Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 1 of 32

WILLIAM A. ISAACSON (Pro hac vice) CHRISTOPHER S. YATES (Pro hac vice)
1 wisaacson@[Link] [Link]@[Link]
2 JESSICA PHILLIPS (Pro hac vice) LATHAM & WATKINS LLP
jphillips@[Link] 505 Montgomery Street, Suite 2000
3 PAUL, WEISS, RIFKIND, WHARTON & San Francisco, CA 94111
GARRISON LLP
4 2001 K Street, NW SEAN M. BERKOWITZ (Pro hac vice)
Washington, DC 20006 [Link]@[Link]
5 LATHAM & WATKINS LLP
6 DONALD J. CAMPBELL (No. 1216) 330 North Wabash Ave, Suite 2800
djc@[Link] Chicago, IL 60611
7 J. COLBY WILLIAMS (No. 5549)
jcw@[Link] LAURA R. WASHINGTON (Pro hac vice)
8 CAMPBELL & WILLIAMS [Link]@[Link]
700 South 7th Street LATHAM & WATKINS LLP
9
Las Vegas, NV 89101 10250 Constellation Blvd, Suite 1100
10 Los Angeles, CA 90067

11 Attorneys for Defendant Zuffa, LLC

12

13
UNITED STATES DISTRICT COURT
14 DISTRICT OF NEVADA

15

16 Cung Le, Nathan Quarry, Jon Fitch, Brandon Vera, Case No.: 2:15-cv-01045-RFB-BNW
Luis Javier Vazquez, and Kyle Kingsbury on behalf
17 of themselves and all others similarly situated, DEFENDANT ZUFFA, LLC’S
MOTION TO EXCLUDE CERTAIN
18
Plaintiffs, OPINIONS OF DR. HAL J. SINGER
19
v. HEARING REQUESTED
20
Zuffa, LLC, d/b/a Ultimate Fighting Championship
21 and UFC,
22 Defendant.
23

24

25

26

27

28

Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW

 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 2 of 32

1 TABLE OF CONTENTS
Page
2
PRELIMINARY STATEMENT .....................................................................................................1
3

4 BACKGROUND .............................................................................................................................2

5 ARGUMENT .................................................................................................................................11

6 I. Opinions Relying On Multivariate Regressions Still Must Satisfy Rule 702 ....................11

7 II. Dr. Singer’s Model Does Not Measure Compensation In A Reliable Way ......................12
8 A. The Model Does Not Measure Actual Compensation ...........................................12
9 B. Following Generally Accepted Principles Results In No Impact ..........................12
10 C. No Court Has Ever Accepted This Method Of Using Revenue Share ..................13
11
D. Dr. Singer Relied on Fraudulent Evidence In Lieu Of Accepted Methods ...........15
12
E. Revenue Share Cannot Distinguish Legal From Illegal Conduct ..........................16
13
F. The Model Does Not Account For Major Revenue And Compensation
14 Drivers....................................................................................................................16
15 III. The Regression’s Key “Foreclosure Share” Variable Does Not Fit .................................19
16 IV. The Regression Does Not Model Impact On Any Individual Fighter ...............................22
17
CONCLUSION ..............................................................................................................................24
18

19

20

21

22

23

24

25

26

27

28

i
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 3 of 32

TABLE OF AUTHORITIES
1
Page(s)
2

3 CASES

4 Apotex, Inc. v. Cephalon, Inc.,

321 F.R.D. 220 (E.D. Pa. 2017) ...........................................................................................9, 19
5
Been v. O.K. Indus., Inc.,
6 398 F. App’x 382 (10th Cir. 2010) ..........................................................................................15
7 Big Bear Lodging Ass’n v. Snow Summit, Inc.,
182 F.3d 1096 (9th Cir. 1999) .................................................................................................19
8

9 In re Capacitors Antitrust Litig.,

2020 WL 870927 (N.D. Cal. Feb. 21, 2020) .............................................................................9
10
Clark v. Takata Corp.,
11 192 F.3d 750 (7th Cir. 1999) ...................................................................................................19
12 Conrad v. Jimmy John’s Franchise,
2021 WL 718320 (S.D. Ill. Feb. 24, 2021) ..............................................................9, 11, 12, 13
13

14 Contreras v. City of L.A.,

656 F.2d 1267 (9th Cir. 1981) .................................................................................................18
15
In re Cox Enters., Inc. Set-Top Cable Tele. Box Antitrust Litig.,
16 2011 WL 6826813 (W.D. Okla. Dec. 28, 2011) ........................................................................9

17 Craftsmen Limousine, Inc. v. Ford Motor Co.,

2005 WL 3263288 (W.D. Mo. Dec. 1, 2005), aff’d, 491 F.3d 380 (8th Cir. 2007) ................20
18
Daubert v. Merrell Dow Pharmaceuticals,
19
509 U.S. 579 (1993) ...............................................................................................................1, 7
20
F.T.C. v. OMICS Grp.,
21 374 F. Supp. 3d 994 (D. Nev. 2019) ........................................................................................16

22 F.T.C. v. OMICS Grp.,

302 F. Supp. 3d 1184 (D. Nev. 2017) ................................................................................15, 16
23
Fleer Corp. v. Topps Chewing Gum, Inc.,
24 658 F.2d 139 (3d Cir. 1981).....................................................................................................19
25
Fraser v. Major League Soccer,
26 284 F.3d 47 (1st Cir. 2002) ......................................................................................................19

27 In re Google Play Store Antitrust Litig.,

2023 WL 5532128 (N.D. Cal. Aug. 28, 2023) ................................................................ passim
28

ii
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 4 of 32

In re High-Tech Emp. Antitrust Litig.,

1 985 F. Supp. 2d 1167 (N.D. Cal. 2013) ...................................................................................12
2
Jinro Am. Inc. v. Secure Invs.,
3 266 F.3d 993 (9th Cir.), amended on denial of rehr’g, 272 F.3d 1289 (9th Cir. 2001).............7

4 John Doe 1 v. Abbott Labs.,

571 F.3d 930 (9th Cir. 2009) ...................................................................................................21
5
Kamakahi v. Am. Soc’y for Reprod. Med.,
6 305 F.R.D. 164 (N.D. Cal. 2015) ...............................................................................................9
7 King Drug Co. of Florence v. Cephalon,
8 2015 WL 12645766 (E.D. Pa. Dec. 22, 2015) .........................................................................20

9 Kumho Tire Co. v. Carmichael,

526 U.S. 137 (1999) ...................................................................................................................7
10
In re LIBOR-Based Fin. Instrument Antitrust Litig.,
11 299 F. Supp. 3d 430 (S.D.N.Y. 2018)......................................................................................11
12 In re Live Concert Antitrust Litig.,
13 863 F. Supp. 2d 966 (C.D. Cal. 2012) ...............................................................................17, 18

14 In re MacBook Keyboard Lit.,

2021 WL 1250378 (N.D. Cal. Apr. 5, 2021) .................................................................9, 11, 12
15
Mazda v. Carfax,
16 2016 WL 7231941 (S.D.N.Y. Dec. 9, 2016), aff’d, 726 F. App’x 66 (2d Cir. 2018)..........9, 19
17 In re Mirena Ius Levonorgestrel-Related Pros. Liab. Litig. (No. II),
341 F. Supp. 3d 213 (S.D.N.Y. 2018)......................................................................................14
18

19 In re NCAA Student-Athlete Name & Likeness Licensing Litig.,

2013 WL 5979327 (N.D. Cal. Nov. 8, 2013) ..........................................................................22
20
Pacific Bell Tel. Co. v. Linkline Commc'ns,
21 555 U.S. 438 (2009) .................................................................................................................13
22 Penk v. Or. State Bd. of Higher Educ.,
816 F.2d 458 (9th Cir. 1987) ...................................................................................................11
23
In re Pharmacy Benefit Managers Antitrust Litig.,
24
2017 WL 275398 (E.D. Pa. Jan. 18, 2017) ..............................................................................22
25
In re Photochromic Lens Antitrust Litig.,
26 2014 WL 1338605 (M.D. Fla. Apr. 3, 2014) .............................................................................9

27 Reed Constr. Data v. McGraw-Hill,

49 F. Supp. 3d 385 (S.D.N.Y. 2014), aff’d, 638 F. App’x 43 (2d Cir. 2016) ..........................18
28

iii
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 5 of 32

Sali v. Corona Regional Medical Center,

1 889 F.3d 623 (9th Cir.), superseded by, 909 F.3d 996 (9th Cir. 2018)......................................5
2
Sardis v. Overhead Door Corp.,
3 10 F.4th 268 (4th Cir. 2021) ................................................................................................7, 14

4 Spinelli v. Nat’l Football League,

96 F. Supp. 3d 81 (S.D.N.Y. 2015)..........................................................................................19
5
United States v. Syufy Enterprises,
6 903 F.2d 659 (9th Cir. 1990) ...................................................................................................20
7 United States v. Tranowski,
8 659 F.2d 750 (7th Cir. 1981) ...................................................................................................14

9 Williamson Oil v. Philip Morris,

346 F.3d 1287 (11th Cir. 2003) ...............................................................................................16
10
In re Young Broad. Inc.,
11 430 B.R. 99 (S.D.N.Y. 2010)...................................................................................................14
12 RULES
13
Fed. R. Evid. 104(a) .........................................................................................................................7
14
Fed. R. Evid. 702 ................................................................................................................... passim
15

16

17

18

19

20

21

22

23

24

25

26

27

28

iv
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 6 of 32

1 GLOSSARY OF ABBREVIATIONS

2 For the Court’s convenience in the interest of efficiency given the large number of

3 exhibits and documents cited herein, this motion uses the abbreviations shown below. These

4 materials are attached to the Declaration of attorney William A. Isaacson (“Decl.”), filed

5 concurrently herewith, and assigned the exhibit numbers shown below, or are publicly

6 available at the docket number indicated below.

7 Document Citation Decl. Ex.

8 Order, In re Google Play Store Antitrust Litig., No. 20-cv-5761 (N.D. N/A Ex. 19
Cal.), ECF No. 457
9
Letter, U.S. Supreme Court (April 24, 223) N/A Ex. 20
10 Expert Report of Prof. Robert H. Topel (Oct. 27, 2017) TR1 Ex. 21
11 Prof. Robert H. Topel’s Reply to the Supplemental Expert Report of TR3 Ex. 22
Hal J. Singer (May 7, 2018)
12 Expert Report of Paul Oyer (Oct. 27, 2017) OR1 Ex. 23
13 Expert Report of Hal J. Singer (Aug. 31, 2017) SR1 Ex. 24
Supplemental Expert Report of Hal J. Singer (Apr. 3, 2018) SR3 Ex. 25
14
Excerpt of Deposition of Hal J. Singer (Sept. 27, 2017) Singer Dep. Ex. 26
15 Declaration of Gregory K. Leonard (Dec. 1, 2023) Leonard Ex. 27
16 SCImago Journal & Country Rank (accessed Nov. 30, 2023) N/A Ex. 28
WorldCat, Journal of Business and Economics (accessed Dec. 1, 2023) N/A Ex. 29
17
Academic Star Publishing Webpage, FAQ (accessed Nov. 30, 2023) N/A Ex. 30
18 Transcript of Class Certification Hearing Day 1 (Aug. 26, 2019) CC1 Hrg. Tr. ECF 724
19 Transcript of Class Certification Hearing Day 2 (Aug. 27, 2019) CC2 Hrg. Tr. ECF 726
Transcript of Class Certification Hearing Day 3 (Aug. 28, 2019) CC3 Hrg. Tr. ECF 730
20 Transcript of Class Certification Hearing Day 4 (Aug. 30, 2019) CC4 Hrg. Tr. ECF 734
21 Transcript of Class Certification Hearing Day 5 (Sept. 12, 2019) CC5 Hrg. Tr. ECF 741

22

23

24

25

26

27

28

v
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 7 of 32

1 Defendant Zuffa, LLC (“Zuffa”) submits this motion to exclude certain opinions of

2 Plaintiffs’ designated expert Dr. Hal J. Singer pursuant to Federal Rule of Evidence 702 and

3 Daubert v. Merrell Dow Pharmaceuticals, 509 U.S. 579 (1993).

4 PRELIMINARY STATEMENT

5 Zuffa has done more to improve career opportunities for professional mixed martial arts

6 (“MMA”) athletes than any other U.S. company. Before Zuffa acquired the Ultimate Fighting

7 Championship (“UFC”) in 2001, the vast majority of states outlawed the sport of MMA, and

8 broadcasters refused to air MMA events. Decl. Ex. 21, TR1 ¶ 49. Between 2001 and 2004, UFC’s

9 owners lost nearly $40 million from their expenditures to improve the prospects for professional

10 fighters and grow the sport of MMA. Id. ¶ 50. Despite those losses, UFC’s owners pressed on.

11 They paid the entire $10 million cost of production to air The Ultimate Fighter in 2005 when no

12 cable network would buy it, and made countless more investments with no certainty of

13 recoupment. Id. ¶¶ 50-52. After years of set-backs, Zuffa’s efforts finally began to pay off:

14 consumer demand for MMA and UFC events started to increase. UFC’s payments to fighters rose

15 too. From 2005 to 2016, UFC’s compensation to fighters per-bout increased by 18% per year, on

16 average. Id. ¶ 58. And from 2011 through 2016, UFC athletes received as much as 37%, 52%, or

17 even 61% more in per-bout compensation, depending on fight performance. Id. ¶ 171 & Ex. 17.

18 Plaintiffs argue Zuffa should have done more. They hired Dr. Hal Singer to opine on

19 whether Zuffa’s use of exclusive contracts with fighters lasting up to thirty months caused each

20 and every UFC bout class member to receive less compensation than he or she would have

21 otherwise, and whether common evidence could establish that causal impact. Dr. Singer intends

22 to tell the jury that he performed economic tests—regressions—that are proof of anticompetitive

23 impact and confirm both of those conditions. In reality, Dr. Singer’s models are incapable of

24 answering either of those questions. His regressions do not evaluate fighters’ actual compensation,

25 and instead evaluate fighters’ compensation as a percentage of Zuffa’s event revenues. They do

26 not test whether the challenged contractual clauses harmed competition or directly assess the

27 effects of those contracts; relying instead on a distorted figure Dr. Singer calls “foreclosure share”

28 that is hard-wired to produce the negative correlation that plaintiffs are seeking to show. And his

Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW

 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 8 of 32

1 regressions do not establish or measure impact on any individual fighter, even when using his

2 “wage share” proxy for compensation and “foreclosure share” proxy for anticompetitive conduct.

3 Standard robustness checks for econometric models demonstrate that Dr. Singer’s models

4 cannot support the opinions he has proffered. His models do not fit the facts of this case. And his

5 opinions do not reflect a reliable application of accepted principles or methods. Numerous courts

6 have found Dr. Singer’s work deficient for similar reasons—and another district court in the Ninth

7 Circuit recently excluded his opinions and decertified a class even though Dr. Singer used a

8 generally accepted economic model because his assumptions about the application of the model to

9 the facts of that particular case were “not supported by the evidence” and did not “give the jury a

10 sound basis on which to make a reasoned and reasonable judgment about antitrust impact and

11 damages.” In re Google Play Store Antitrust Litig., 2023 WL 5532128, at *9 (N.D. Cal. Aug. 28,

12 2023); Decl. Ex. 19, Order, In re Google Play Store Antitrust Litig., No. 20-cv-5761 (N.D. Cal.

13 Sept. 13, 2023), ECF No. 457 (decertifying class after excluding “Dr. Singer’s pass-through

14 formula”). His impact opinions should be excluded here as well.

15 BACKGROUND

16 Plaintiffs claim that athletes committing by contract to fight exclusively for the UFC for

17 up to thirty months (if a one-time extension for a title-holding champion is triggered) denied rival

18 promoters access to a critical mass of “elite” MMA fighters, limited those fighters’ options, and

19 anticompetitively caused their compensation to be less than it would have otherwise been.1 The

20 facts on the ground do not support that claim. Multiple MMA promoters, including Bellator and

21 Strikeforce, used exclusivity and champion’s clauses in their fighter contracts, because those

22 clauses are recognized to incentivize promoters to invest for fighters’ benefit. Decl. Ex. 21, TR1

23 ¶¶ 90-97, 107-09. Moreover, Zuffa’s payments to fighters increased during the relevant period

24 and Zuffa paid fighters more than other MMA promoters. Id. ¶¶ 44, 153, 169-71, 288 & Exs. 14,

25 17-18. For example, when Zuffa acquired Strikeforce in 2011, the compensation of Strikeforce

26 athletes that moved to UFC increased sharply. Id. ¶¶ 44, 180-83 & Ex. 21. But plaintiffs proffer

27

28 1
A “champion’s clause” grants Zuffa the limited opportunity to retain a current weight-class
champion for one extension of one year or a defined number of bouts. Decl. Ex. 21, TR1 ¶ 107.
2
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 9 of 32

1 Dr. Singer to opine that these common contract clauses nonetheless suppressed fighters’

2 compensation, according to his “impact” regression.2 This motion focuses on those opinions.

3 Dr. Singer’s Impact Opinions: Dr. Singer’s impact regression is supposed to test the

4 relationship between two novel variables: what Dr. Singer calls “fighter share” of event revenues

5 (or “wage share” or “revenue share”) and “foreclosure share.” Revenue share is intended to

6 measure the percentage of revenues from individual MMA events paid to a Zuffa fighter or to a

7 Strikeforce fighter (before Zuffa acquired Strikeforce). Id. ¶ 181. Importantly, this is not a

8 measure of fighters’ actual compensation. The key independent variable on the other side of

9 Dr. Singer’s regression is foreclosure share, which is a ratio of the fighters with an exclusive Zuffa

10 contract of thirty months (if counting the champion’s clause extension), compared to all fighters

11 that Dr. Singer finds in his alleged relevant input market. Id. ¶ 171. Dr. Singer took the measure

12 a step further by weighting each fighter with his or her promoter’s average pay-per-view and gate

13 revenues for all events. Id. ¶¶ 170, 182.3 When Dr. Singer performed this impact regression,

14 which included additional variables meant to control for factors that might influence a fighter’s

15 compensation, he found a negative relationship between revenue share and foreclosure share—

16 meaning that as the revenue-weighted foreclosure share went up, the fighters’ share of revenue

17 went down. Id. ¶¶ 183, 187. From this result, Dr. Singer opines that Zuffa’s exclusive contracts

18 of up to thirty months (counting the champion’s clause) anticompetitively harmed UFC fighters’

19 compensation during the class period. Id. ¶ 183.

20 Professor Robert H. Topel: Dr. Robert Topel, Distinguished Service Professor of

21 Economics at The University of Chicago Booth School of Business, analyzed Dr. Singer’s model

22 and found it fundamentally flawed in many ways. Most relevant to this motion, Prof. Topel

23 explained that using fighters’ share of event revenues, instead of their actual compensation, to

24
2
Although plaintiffs challenge conduct other than Zuffa’s contracts with fighters, Dr. Singer’s
25 “measure of economic harm” includes “only the exclusionary effects of Zuffa’s Fighter contracts.”
Decl. Ex. 24, SR1 ¶ 187 n.456; see also Decl. Ex. 21, TR1 ¶ 201 (“Dr. Singer makes no attempt
26
to establish that there was anticompetitive harm from other elements of Plaintiffs’ Challenged
27 Conduct apart from the exclusive aspect of PAR contracts . . . .”).
3
For a “submarket” of only the top fifteen fighters per weight class (the “headliner” submarket),
28 Dr. Singer weighted fighters using the inverse of their rank—meaning the top fighter per weight
class was deemed five times more valuable than the fifth-ranked fighter. Decl. Ex. 24, SR1 ¶ 170.
3
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 10 of 32

1 estimate how the challenged conduct affected fighters financially is inconsistent with generally

2 accepted economic principles. Decl. Ex. 21, TR1 ¶¶ 124-49, 284-88. That is particularly the case

3 here, Prof. Topel explained, because Zuffa’s athletes were not paid on a percentage-of-revenue

4 basis, the athletes’ actual compensation consistently increased over the class period, and

5 Dr. Singer’s own regression estimates no harm if compensation is used instead of revenue share.

6 Id. ¶¶ 146-47 & Ex. 13.4

7 Prof. Topel found numerous other examples proving that Dr. Singer’s regression did not

8 reliably model what was driving fighters’ compensation. For example, the regression does not

9 show any increase in revenue share in response to the fighter winning a bout, despite Dr. Singer

10 admitting that, in the real world, the winning fighter’s base pay is nearly always doubled. Infra

11 p.18. Moreover, Dr. Singer’s model did not control, at all, for UFC’s expenditures to host,

12 promote, and produce events, which are obvious drivers of event revenue and could exceed

13 $10 million for a single event. Decl. Ex. 22, TR3 ¶ 14 & Ex. 5. In other words, Dr. Singer’s model

14 failed to account for the possibility of those costs contributing to event revenue, even though the

15 academic literature speaks to that likelihood. See Decl. Ex. 21, TR1 ¶¶ 133-40 & App. A; Decl.

16 Ex. 23, OR1 ¶ 36. This omission was significant, because if event revenues increase, that

17 mechanically causes both the fighters’ revenue share to decrease and the foreclosure share to

18 increase—creating the appearance of a negative correlation. See infra pp.20-21.

19 When Dr. Singer tried to correct this failing with new regression analyses during class-

20 certification briefing by adding some annualized promotion expenses to his regression, his impact

21 assessment was cut nearly in half. Decl. Ex. 22, TR3 ¶ 19. But even that late effort to control for

22 known drivers of revenue was flawed because Dr. Singer used annualized expenses instead of per-

23 event expenses, which masks how per-event expenses influence per-event revenues, and he still

24 failed to include any measure of Zuffa’s production costs. See infra § II(F).

25
4
To convert Dr. Singer’s regression to a measurement of actual compensation, Prof. Topel
26
replaced Dr. Singer’s incorrect dependent variable with the natural logarithm of an athlete’s
27 compensation from an event, measured in dollars. Decl. Ex. 21, TR1 ¶ 146. Using the log of
compensation is standard practice when estimating compensation regressions, which allows the
28 regression to be informative about changes in athlete’s compensation in percentage terms. Id.
¶ 146 & n.212.
4
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 11 of 32

1 Prof. Topel also found foundational flaws in Dr. Singer’s approach to “weighting” fighters

2 to estimate a foreclosure share, including that use of promoter revenue in both foreclosure share

3 and revenue share will “mechanically generate ‘evidence’ of anticompetitive impact, even if such

4 impact is impossible and even if compensation of MMA fighters is increasing.” Decl. Ex. 21, TR1

5 ¶¶ 27, 219-30; infra pp.20-21. As proof of this distortion in Dr. Singer’s foreclosure measure,

6 Prof. Topel found that removing the use of revenue weighting caused the regression to indicate no

7 harm to fighters (even when relying on Dr. Singer’s revenue-share proxy). Id. ¶¶ 165-67 & Ex. 16.

8 Daubert Briefing at Class Certification: When plaintiffs moved to certify a class, Zuffa

9 filed a motion to exclude certain of Dr. Singer’s opinions (and other expert opinions) because they

10 fail Daubert and Rule 702. Briefing was completed in 2018. ECF Nos. 524, 534, 551, 658-1, 569.

11 In September 2018, the Court denied the motion as “premature,” and instructed that it

12 would “consider the arguments” as part of class certification, but that “the Court is not required to

13 determine the admissibility of evidence in ruling on the Motion to Certify Class nor does such

14 evidence have to be admissible for consideration by the Court.” Minute Or., ECF No. 600. The

15 Court cited to Sali v. Corona Regional Medical Center, in which the Ninth Circuit ruled that the

16 “manner and degree of evidence required at the preliminary class certification stage is not the same

17 as at the successive stages of the litigation,” and that class-certification orders may be “amended

18 before final judgment.” 889 F.3d 623, 631 (9th Cir.), superseded by, 909 F.3d 996 (9th Cir. 2018).

19 After holding a multi-day evidentiary hearing in 2019, the Court discussed Zuffa’s motion

20 to exclude in an August 2023 order denying in-part and granting in-part class certification. Class

21 Cert. Or., ECF No. 839. The Court characterized Zuffa’s Daubert challenges as attacking

22 Dr. Singer’s choice to apply regression analysis to “a set of facts in a new market or industry,”

23 which the Court ruled “does not alter or undermine the generally accepted nature of their

24 methodological techniques.” Id. at 12-15. The Court also ruled that “at this stage” (i.e., class

25 certification) concerns “about the categorization or definitions of data Dr. Singer used” and his

26 variable choices go to “matters of weight and probative value for a jury to evaluate,” not to

27 admissibility. Id. at 51 (citation omitted). With respect to future stages, the “Court reiterate[d]

28

5
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 12 of 32

1 that its findings in this order are not meant to be dispositive as to any factual dispute beyond class

2 certification.” Id. at 21 n.21.5

3 As to Dr. Singer’s use of revenue share for the dependent variable in his regressions, the

4 Court determined this was “an appropriate proxy” for estimating the theoretical compensation that

5 labor would receive in a competitive market, which in economic verbiage is known as marginal

6 revenue product or “MRP.” Id. at 49 n.42, 52-54. The Court found this novel approach was

7 justified because this industry is “unique in several respects,” including that the “fighters,

8 themselves, are the product,” there was a large amount of “event-level data” available, and

9 compensation for boxers is “guaranteed as a percentage of event revenue.” Id. at 54-56. The Court

10 also cited business documents that listed fighters’ wages as a percentage of revenue, and concluded

11 that Prof. Topel and another UFC expert, Prof. Oyer,6 have referred to wage share in some writings

12 about sports industries. Id. at 54-55.

13 As to Zuffa’s criticisms of the weighting Dr. Singer applied to his foreclosure share, the

14 Court found those arguments did not “fundamentally or substantively undermine the reliability

15 and explanatory effect of Plaintiffs’ modeling,” and the Court was “not persuaded that they defeat

16 class certification.” Id. at 59.

17 Recent Developments: Several recent developments relating to the law on evaluating

18 expert testimony and the reliability of Dr. Singer’s analyses put a finer point on Zuffa’s 2018

19 arguments, and confirm that Dr. Singer’s impact opinions cannot be presented to a jury.

20
5
In a November 29, 2023 meet and confer regarding a pretrial schedule, counsel for plaintiffs
21 took the position that the Court’s Class Certification Opinion denied Zuffa’s prior motion to
exclude Dr. Singer’s opinions. The Order concluding the Class Certification Opinion, however,
22
does not include a denial of that motion. See Class Cert Or. at 79-80. Furthermore, the Court’s
23 statements on the docket and in the Class Certification Opinion indicate that the Court’s discussion
of the motion to exclude was for class-certification purposes, not the merits stage. See supra pp.5-
24 6; see also Class Cert. Or. at 21 n.21 (“When the Court indicates here that Plaintiffs or Defendant
have ‘established’ a particular fact or position, this simply means that they have satisfied the
25 necessary legal threshold in the context of determining class certification as set forth in Olean.”).
6
Professor Oyer is the Fred H. Merrill Professor of Economics at Standard University’s
26
Graduate School of Business. He is a Research Assistant at the National Bureau of Economic
27 Research, and was the Editor-in-Chief of The Journal of Labor Economics. Professor Oyer
provided opinions in this case on the economically-accepted methods of evaluating compensation,
28 including in allegedly monopsonized markets, and assessed Dr. Singer’s use of fighter pay as a
percentage of revenue, among other analyses and opinions. See Decl. Ex. 23, OR1 ¶¶ 1, 5.
6
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 13 of 32

1 First, the U.S. Supreme Court ordered an amendment to Federal Rule of Evidence 702

2 effective December 1, 2023.7 The amendment “is not a sea change but rather an amplification of

3 existing FRE 702 standards.” Google Play Store, 2023 WL 5532128, at *5. The amendment

4 “clarifies that an expert witness’s opinion testimony is admissible” only if “the proponent

5 demonstrates to the court that it is more likely than not that the proposed testimony satisfies” all

6 elements of Rule 702, including that the testimony “is based on sufficient facts or data” and “is the

7 product of reliable principles and methods.” Id. (citation omitted). The amendment also refines

8 the last Rule 702 element (Rule 702(d)), to require that an expert opinion “reflects a reliable

9 application of the principles and methods to the facts of the case.” Id. at *5 (citation omitted).

10 This amendment arose because the Advisory Committee on Evidence Rules detected a

11 “pervasive problem” of courts delegating to the jury the judicial responsibility of critically

12 screening expert testimony. Sardis v. Overhead Door Corp., 10 F.4th 268, 283-84 (4th Cir. 2021)

13 (quoting Advisory Comm. on Evidence Rules, Agenda for Committee Meeting 17 (Apr. 30,

14 2021)). Because “expert evidence can be both powerful and quite misleading,” Daubert v. Merrell

15 Dow Pharmaceuticals, 509 U.S. 579, 595 (1993) (quotation omitted), Rule 702 imposes a “special

16 obligation” on trial courts to act as a “gatekeeper” of expert testimony, Kumho Tire Co. v.

17 Carmichael, 526 U.S. 137, 147 (1999). The “proper exercise of that gatekeeping function is

18 critically important.” Jinro Am. Inc. v. Secure Invs., 266 F.3d 993, 1005 (9th Cir.), amended on

19 denial of rehr’g, 272 F.3d 1289 (9th Cir. 2001). But as the Advisory Committee explained,

20 “unfortunately many courts have held that the critical question of the sufficiency of an expert’s

21 basis [for his testimony], and the application of the expert’s methodology, are generally questions

22 of weight and not admissibility.” Sardis, 10 F.4th at 284. The amendment “reject[s]” that

23 “incorrect application of Rules 702 and 104(a),” id., and “emphasize[s] that each expert opinion

24 must stay within the bounds of what can be concluded from a reliable application of the expert’s

25 basis,” Proposed Am. to Fed. R. of Evid. 106, 615, & 702, 344 F.R.D. 850, 857-58 (Oct. 19, 2022).

26 Second, several courts have recently found Dr. Singer’s regression analyses and economic

27 modelling inadmissible because he committed the same types of errors that Zuffa and its experts

28
7
See Decl. Ex. 20, Letter, U.S. Supreme Court (Apr. 24, 2023).
7
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 14 of 32

1 identify here. The most recent example was a ruling in In re Google Play Store Antitrust

2 Litigation, just weeks after this Court’s class-certification order, where the court excluded

3 Dr. Singer’s economic opinions at the merits stage, despite previously relying on them for class-

4 certification purposes. 2023 WL 5532128 (N.D. Cal. Aug. 28, 2023). In Google Play Store, the

5 consumer-plaintiffs proffered Dr. Singer to support their claims that Google illegally monopolized

6 the alleged Android-device app distribution market, causing them to be overcharged for apps. Id.

7 at *1. The “critical element” in Dr. Singer’s overcharge analysis at both class certification and the

8 merits stage was a “pass-through” formula that he opined could “quantify the portion of the

9 supracompetitive cost imposed on developers by Google that was ‘passed through’ to, or more

10 aptly, paid by, consumers.” Id. at *2. Dr. Singer’s pass-through formula relied on a “logit model”

11 to approximate the demand for apps sold in the Google Play Store. Id. at *3.

12 After the court relied on Dr. Singer’s impact opinion at class certification, a new expert,

13 Dr. Gregory Leonard, “took a fresh look” at Dr. Singer’s opinions at the merits stage. Id. at *4.

14 Dr. Leonard did not dispute that logit had been validated “in the peer-reviewed economics

15 literature,” and was “widely used by economists” to model demand. Id. at *5-6. But as

16 Dr. Leonard explained, reliance on logit requires that “all goods in the market where demand is

17 being studied are substitutes of one another in proportion to their share of that market.” Id.

18 Dr. Singer failed to confirm this condition existed in the market he was studying and also failed to

19 “compare the ‘fit’ of the logit model with ‘that of an alternative demand model.’” Id. at *9.

20 Dr. Leonard, by contrast, showed that many apps being sold alongside each other in the Google

21 Play Store clearly were not substitutes, such as language-learning apps and plant-identification

22 apps. Id. at *7. The court found that Dr. Leonard’s opinions “put a much finer point” on flaws in

23 Dr. Singer’s model and demonstrated Dr. Singer was using “the logit model in an overly simple

24 way,” based on an “unproven assumption,” which did “not give the jury a sound basis on which to

25 make a reasoned and reasonable judgment about antitrust impact and damages.” Id. at *7-9.

26 The Google Play Store decision is the most recent example of a court excluding

27 Dr. Singer’s economic opinions, but it follows a growing crowd of courts rejecting his opinions as

28 unsuited for presentation to a jury. For example, in 2021, after this Court’s class-certification

8
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 15 of 32

1 hearing, a court rejected Dr. Singer’s opinions in a labor monopsony case because he committed

2 “a methodological flaw” rendering his models “unreliable” and his estimates “inflated” in favor of

3 plaintiffs. Conrad v. Jimmy John’s Franchise, 2021 WL 718320, at *16-19 (S.D. Ill. Feb. 24,

4 2021). Less than two months later, another court found Dr. Singer had made “untenable

5 assumptions,” and excluded his model as a result. In re MacBook Keyboard Lit., 2021 WL

6 1250378, at *5 (N.D. Cal. Apr. 5, 2021). And a year earlier, a court excluded Dr. Singer’s

7 “qualitative” analysis of whether the challenged conduct was anticompetitive because he merely

8 relied on two sources from antitrust agencies to assess competitive impact, which he lacked

9 “scientific, technical, or other specialized knowledge” to do. In re Capacitors Antitrust Litig.,

10 2020 WL 870927, at *2 (N.D. Cal. Feb. 21, 2020). There are even more. In 2017, Dr. Singer

11 offered two opinions that were “contrary to the record” and “contrary to the law” (in the case of

12 one opinion) and based on assumptions with “no rational or legal basis” (in the case of the

13 second)—the court excluded both. Apotex, Inc. v. Cephalon, Inc., 321 F.R.D. 220, 236, 237 (E.D.

14 Pa. 2017). In 2015, a court excluded Dr. Singer’s opinions because “his analysis d[id] not reliably

15 support his conclusion.” Kamakahi v. Am. Soc’y for Reprod. Med., 305 F.R.D. 164, 179–82 (N.D.

16 Cal. 2015). And in 2014, a court found Dr. Singer’s opinion “legally deficient” because he relied

17 on “irrelevant” data. In re Photochromic Lens Antitrust Litig., 2014 WL 1338605, at *24-25 (M.D.

18 Fla. Apr. 3, 2014). Even more cases have criticized Dr. Singer’s opinions and found them

19 unreliable or unsuitable for a jury, even if not expressly excluding them.8

20 A third development is that Dr. Leonard has now evaluated aspects of Dr. Singer’s models

21 and found flaws similar to the errors that led the court to exclude Dr. Singer’s opinions in Google

22 Play Store.9 Dr. Leonard did not seek to re-do the work completed by Zuffa’s other experts. Decl.

23
8
See, e.g., Mazda v. Carfax, 2016 WL 7231941, at *12-14 (S.D.N.Y. Dec. 9, 2016), aff’d, 726
24 F. App’x 66 (2d Cir. 2018) (holding that “no reasonable jury could rely” on “Singer’s foreclosure
finding”); In re Cox Enters., Inc. Set-Top Cable Television Box Antitrust Litig., 2011 WL 6826813,
25 at *12-16 (W.D. Okla. Dec. 28, 2011) (expressing doubts on Dr. Singer’s analysis).
9
Dr. Leonard earned a Ph.D. in economics from the Massachusetts Institute of Technology. He
26
has served as an assistant professor at Columbia University, as the Vice Chair for Economics of
27 the Board of Editors of the Antitrust Law Journal, and has given invited lectures on antitrust issues
at the Federal Trade Commission, the U.S. Department of Justice, the Directorate General for
28 Competition of the European Commission, the Fair Trade Commission of Japan, and China’s
Supreme People’s Court and Ministry of Commerce. Decl. Ex. 27, Leonard ¶¶ 2-6.
9
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 16 of 32

1 Ex. 27, Leonard ¶ 7 n.1. Instead, Dr. Leonard analyzed whether Dr. Singer’s model was a reliable

2 method of answering the key question of whether the challenged contracts injured any individual

3 fighter. Id. ¶ 7. As described below and in his accompanying declaration, Dr. Leonard found,

4 unambiguously, that the impact regression cannot answer that question. See, e.g., id. ¶ 9.

5 As Dr. Leonard explained, it should be undisputed that Dr. Singer’s impact regression

6 attempts to measure only an average impact across all fighters. Id. ¶¶ 12-17. The unspoken

7 assumption by Dr. Singer, then, is that the challenged contracts must have impacted every fighter

8 identically—because the model produces only one impact measure (or “coefficient”). Id. ¶ 15.

9 But results from both a generally accepted statistical test, and Dr. Singer’s own regression (if

10 permitted to estimate different foreclosure effects for individual fighters), reject the assumption

11 that every fighter’s compensation-as-a-share-of-revenue was similarly affected. Id. ¶¶ 18-21. In

12 fact, allowing Dr. Singer’s own regression to show effects for individual fighters (instead of

13 lumping the effect together into an average), shows that for over 80% of fighters, their

14 compensation-as-a-share-of-revenue did not decrease a statistically significant amount as

15 foreclosure share increased. Id. ¶ 19, Fig. 1.10 Thus, even using Dr. Singer’s preferred revenue-

16 share measure, foreclosure-share measure, and regression, no anticompetitive impact is observed

17 for the vast majority of individual fighters. Id. ¶¶ 18-21, Figs. 1, 2.

18 Dr. Leonard further explained that it was unreasonable for Dr. Singer to have failed to

19 perform this check. Economic literature on theories of harm from exclusive contracts teaches that

20 any impact is likely to be inconsistently felt by the various dealers (e.g., the fighters in this case).

21 Id. ¶¶ 10-11. Accepted checks for identical impact are readily available. Id. ¶¶ 14, 19 & n.20.

22 And the “compensation structure” and “common factors” regressions Dr. Singer performed to try

23 to show similar impact indirectly (1) did not evaluate impact from foreclosure share; and (2) did

24 not show a shared effect across individuals anyway. Id. ¶¶ 24-31.

25 As the court did in Google Play Store, this Court should consider Dr. Leonard’s

26 assessment, and other recent developments, when evaluating if Dr. Singer’s impact regression is

27

28 10
This finding is robust across the tracked and ranked input markets that Dr. Singer utilizes, as
well as the headliner submarket. See Decl. Ex. 27, Leonard Fig. 1.
10
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 17 of 32

1 qualified to be the first model ever presented to a federal jury intended to prove that contractors’

2 compensation growth not keeping pace with a business’s revenue growth is anticompetitive harm.

3 ARGUMENT

4 Dr. Singer’s opinions about antitrust impact and damages, flowing from his “impact”

5 regression, do not satisfy the requirements of Federal Rule of Evidence 702, especially given the

6 guidance of the Supreme Court and the drafters of the amendment to Rule 702. The conclusions

7 that Dr. Singer draws from his regression are not the product of reliable principles and methods,

8 and do not reflect the reliable application of such principles and methods to the facts of this case.

9 I. Opinions Relying On Multivariate Regressions Still Must Satisfy Rule 702

10 Zuffa takes no issue with the statistical method of multivariate regression analysis. When

11 properly specified, a regression analysis can be informative of impact and damages, and the

12 regressions that Zuffa’s experts have run show that neither occurred in this case. But as the Ninth

13 Circuit has made clear, there is no “blanket approval to the introduction of all evidence derived

14 from multiple regression analyses.” Penk v. Or. State Bd. of Higher Educ., 816 F.2d 458, 464-65

15 (9th Cir. 1987). Regressions may be excluded when they are used to draw inferences that the

16 regressions are not “validate[d]” to support, Google Play Store, 2023 WL 5532128, at *8, or when

17 they are based on an assumption that “is not factually accurate” or on a “misconception,” In re

18 MacBook Keyboard Litigation, 2021 WL 1250378, at *4. And regressions may be excluded when

19 they rely on a “methodological flaw.” Jimmy John’s Franchise, 2021 WL 718320, at *18-19.

20 These three examples come from courts that excluded Dr. Singer’s regression analyses in

21 just the time since the parties completed the class-certification briefing. There are many other

22 scenarios when an “expert ‘may employ this reliable methodology in an unreliable way.’” Id. at

23 *16 (quoting Phillip E. Areeda & Herbert Hovenkamp, Antitrust Law: An Analysis of Antitrust

24 Principles and Their Application ¶ 399 (5th ed. 2020)); see, e.g., In re LIBOR-Based Fin.

25 Instrument Antitrust Litig., 299 F. Supp. 3d 430, 559 & n.136 (S.D.N.Y. 2018) (“It also bears

26 repeating that the acceptance of regressions as a statistical method generally does not imply that

27 all regressions will be admissible under Daubert.”).

28

11
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 18 of 32

II. Dr. Singer’s Model Does Not Measure Compensation In A Reliable Way
1
The first methodological flaw in Dr. Singer’s impact analyses is that his model does not
2
reliably measure compensation paid to fighters for several reasons, including (1) it does not
3
measure compensation at all (nor does it attempt to); (2) the standard compensation-assessment
4
measure, and how UFC fighters were actually compensated, shows no impact; (3) the proxy for
5
compensation that Dr. Singer uses has never been accepted by a court as a reliable indicator of
6
anticompetitive impact; (4) the article Dr. Singer relied upon comes from a fraudulent journal;
7
(5) the proxy cannot distinguish between competitive and anticompetitive conduct; and (6) the
8
model does not account for obvious non-fighter drivers of event revenue.
9
A. The Model Does Not Measure Actual Compensation
10
Plaintiffs’ theory of harm is that Zuffa’s fighters would have received more actual
11
compensation in the absence of Zuffa’s contracts, not that they would have received a higher “wage
12
share.” CC1 Hrg. Tr. at 122:2-7, ECF No. 724 (“Q. And if UFC had monopsony power, what
13
would that mean for UFC fighters? What would we see in the data? A. Well, what we would see
14
is a wage level that wasn’t competitive. . . .”). This is intuitive because, as Prof. Topel explained,
15
“workers care about the dollar value of their compensation, not compensation as a share of their
16
employer’s revenue.” Decl. Ex. 21, TR1 ¶ 140. By failing to even attempt to measure any
17
correlation between the challenged conduct and actual compensation, Dr. Singer’s model does not
18
fit plaintiffs’ theory of damages or the facts of this case, and should be excluded. See In re
19
MacBook Keyboard Litig., 2021 WL 1250378, at *4 (excluding Dr. Singer’s opinion because his
20
“misconception” caused his opinions to “not accurately relate to Plaintiffs’ theory of damages in
21
this case”).
22
B. Following Generally Accepted Principles Results In No Impact
23
The generally accepted method of attempting to calculate harm to labor compensation from
24
monopsonization is to measure the challenged conduct’s effect on actual compensation. E.g., In
25
re High-Tech Emp. Antitrust Litig., 985 F. Supp. 2d 1167, 1207-08 (N.D. Cal. 2013). Just two
26
years ago, Dr. Singer employed that approach to test monopsony impact on deli workers allegedly
27
harmed from “no-poach” clauses. Jimmy John’s Franchise, 2021 WL 718320, at *2-3. Dr. Singer
28

12
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 19 of 32

1 relied upon a regression to compare “the compensation that each Class Member actually received

2 to the compensation they would have received in the absence of the No-Poach Provision.” Id.11

3 Following the generally accepted method in this case (using Dr. Singer’s own model, but

4 with actual compensation) results in a finding of no impact. Decl. Ex. 21, TR1 ¶¶ 146-48 & Ex. 13.

5 No expert has produced a model in this case that shows any negative effect on fighters’ actual

6 compensation as a result of the challenged conduct.12 If Dr. Singer’s model reliably fit the facts

7 of this case, at a minimum it should produce consistent results when using compensation, but it

8 does not. See id. ¶ 149; CC2 Hrg. Tr. at 197:8-12, ECF No. 726 (“[I]f variation in foreclosure

9 share is a measure of variation in monopsony power that affects wages, then a higher foreclosure

10 share ought to be associated with a lower wage within the context of their model. It doesn’t

11 happen.”). Dr. Singer’s failure to “compare the ‘fit’” of his proffered model with “that of an

12 alternative,” generally accepted model (a model he has endorsed elsewhere), is grounds for

13 exclusion, just as it contributed to the recent exclusion of his opinions in Google Play Store. 2023

14 WL 5532128, at *9.

15 C. No Court Has Ever Accepted This Method Of Using Revenue Share

16 The problem is not limited to Dr. Singer employing a model that is different from the

17 generally accepted approach; the problem is also that the method he selected is unprecedented in

18 monopsony antitrust litigation. No other court has ever approved this type of use of revenue share.

19 And there are important reasons why no court should, including the fact that economics does not

20 predict that companies should or will pay labor a defined share of revenues, Decl. Ex. 21, TR1

21 ¶ 127; Decl. Ex. 23, OR1 ¶ 39, and “[c]ourts are ill suited ‘to act as central planners, identifying

22 the proper price, quantity, and other terms of dealing,” Pacific Bell Telephone Co. v. Linkline

23 Communications, 555 U.S. 438, 452 (2009) (citation omitted). Thus, Amended Rule 702 confirms

24 that trials should not be the testing ground for advancing novel disciplines and methods of

25
11
In that case, Dr. Singer’s regression was still excluded because of his incorrect assumptions
26
about the data. See Jimmy John’s Franchise, 2021 WL 718320, at *18-19.
12
27 Dr. Singer has not opined that it is impossible to apply a model that follows the generally
accepted approach of measuring actual compensation in this case. A model following the wage-
28 measurement approach, if properly specified, could still account for the fact that wages were
increasing over time, for example. CC2 Hrg. Tr. at 132:18-136:9; 197:8-12, ECF No. 726.
13
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 20 of 32

1 regression analysis. See Sardis, 10 F.4th at 282-84 (finding motivation for Rule 702 amendment

2 in the “pervasive problem” of courts holding that questions of the “sufficiency of basis and reliable

3 application of principles and methods” were issues “of weight for the jury”); United States v.

4 Tranowski, 659 F.2d 750, 757 (7th Cir. 1981) (“The trial court should not be used as a testing

5 ground for theories supported neither by prior control experiments nor by calculations with indicia

6 or reliability.”). That approach would flip the court’s gatekeeping responsibility on its head by

7 having courts serve as front-runners of fringe theories and deeming them reliable if a jury accepts

8 them. Rule 702 requires the opposite, as lay jurors are not expected to weed out unreliable

9 methodologies or to be capable of doing so—that is the Court’s role, as the amendments to Rule

10 702 make clear. Sardis, 10 F.4th at 283-84. An economic method should be debated, tested, and

11 generally accepted in academic discourse before it is presented as a reliable methodology in a court

12 as a basis for liability because “[l]aw lags behind science; it does not lead it.” In re Mirena Ius

13 Levonorgestrel-Related Pros. Liab. Litig. (No. II), 341 F. Supp. 3d 213, 270-71 (S.D.N.Y. 2018)

14 (excluding expert’s “theory of causation [that] would invite the jury to guess as to the validity of

15 a novel and untested theory based essentially on his say-so”); In re Young Broad. Inc., 430 B.R.

16 99, 127 (S.D.N.Y. 2010) (excluding testimony based on method that had not “been tested or relied

17 upon by other experts,” “subjected to peer review or discussed in any publication,” or “employed,

18 discussed, [or] generally accepted in any academic or professional community”). But plaintiffs

19 have never identified a single scholarly article promoting the use of revenue share to show an

20 anticompetitive effect, where using actual compensation in the same model shows no

21 anticompetitive effect.13

22 The argument that in MMA “the fighter is the product,” does not justify this new and

23 unsubstantiated approach. Class Cert. Or. at 49. The MMA event is literally the product, as fans

24
13
Professors Topel and Oyer were asked at the class-certification hearing if there was any
25 economics literature using revenue share to “measure the effects of a monopsony” and they said
there was no such published literature. CC2 Hrg. Tr. at 247:7-15, ECF No. 726 (zero regressions
26
with revenue share in the literature to measure the effects of monopsony power); CC5 Hrg. Tr. at
27 96:16-19, ECF No. 741 (articles “never run a regression the way Singer did, never”). Plaintiffs
failed to rebut this testimony, only asking Dr. Singer about published articles on wage share and
28 not asking him about published articles promoting the use of wage share “to measure the effect of
monopsony.” CC3 Hrg. Tr. at 149:3-6, ECF No. 730.
14
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 21 of 32

1 do not buy tickets to see a single fighter stand alone in the Octagon. Even Dr. Singer acknowledged

2 that “television distributors are interested in Zuffa producing high-quality events,” and the “precise

3 identities” of individual fighters is “less critical than Zuffa’s offering generic, top-ranked Fighters

4 who are competitive[ly] matched and effectively promoted.” SR1 ¶ 284. But even if the fighter

5 was the “product,” that does not explain diverging from generally accepted practices into

6 unsubstantiated theories, as even monopsony cases where the buyer allegedly underpaid for the

7 “product” use impact regressions testing the actual amount paid (not price as a percent of revenue).

8 See, e.g., Been v. O.K. Indus., Inc., 398 F. App’x 382, 386, 397 (10th Cir. 2010) (affirming analysis

9 of monopsony impact based on prices paid to poultry farmers, not prices as a percentage of the

10 buyer’s revenues).

11 D. Dr. Singer Relied On Fraudulent Evidence In Lieu of Accepted Methods

12 The fact that Dr. Singer relied on a falsified journal as a basis for his divergence from

13 generally accepted principles further confirms that his methodology lacks sufficient reliability.

14 Dr. Singer admitted that he arrived at using revenue share as a proxy for measuring fighters’

15 marginal revenue product, not from testing the relative revenue contributions by fighters versus

16 Zuffa’s investments, but instead by relying on what he thought was an academic article. CC2 Hrg.

17 Tr. at 95:6-15, ECF No. 726 (discussing article McGowan & Mahon, Demand for the Ultimate

18 Fighting Championship: An Econometric Analysis of PPV Buy Rates, J. of Bus. & Econ. (June

19 2015)); Decl. Ex. 26, Singer Dep. at 118:22-119:24.14 That article is from a “journal” deceptively

20 named the Journal of Business and Economics to mislead authors and readers to believe it is the

21 legitimate publication, The Journal of Economics and Business, a known deceptive and misleading

22 practice. See F.T.C. v. OMICS Grp., Inc., 302 F. Supp. 3d 1184, 1191 (D. Nev. 2017) (part of

23 deceptive scheme involved the use of “names that are nearly identical to other respected journals”).

24 The former is not a legitimate academic journal, but rather the product of an academic vanity press,

25 the Academic Star Publishing Company. See ECF No. 743-11 (McGowan & Mahon article).

26 Academic conferences have posted warnings about Academic Star’s “phishing scam” where it

27

28 14
This article is publicly available at ECF No. 743-11, and was submitted as JCCX72 during the
class certification hearing.
15
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 22 of 32

1 spams academic conference attendees seeking submissions and then demands a $50 per page

2 publishing fee.15 This pay-to-play journal boasts no standard metric of journal quality,16 and is not

3 cataloged in any libraries worldwide (except for one in Malaysia).17 Its website promises a 3-4

4 week “peer review” process and has an FAQ section that begins each of five identical pages: “Core

5 values: Guangzhou love Bosch help.”18 Reliance on an article that was a fraudulent sham requires

6 exclusion of Dr. Singer’s use of fighter share as a proxy for MRP.

7 E. Revenue Share Cannot Distinguish Legal From Illegal Conduct

8 At the class certification hearing, Professors Topel and Oyer explained that in standard

9 economics, a regression measuring an effect on the fighter share of event revenues could not

10 distinguish a competitive market or a market with a legal monopsony from one with

11 anticompetitive restraints because a rise in revenue in either would decrease the fighter share of

12 revenue. CC2 Hrg. Tr. at 105:6-19, 238:23-239:3, ECF No. 726; CC5 Hrg. Tr. at 99:8-21, ECF

13 No. 741. Dr. Singer did not respond to this testimony. Thus, it is undisputed that under accepted

14 economic principles, Dr. Singer’s regressions cannot prove that a declining fighter share is linked

15 to the challenged contractual clauses. This also requires exclusion of Dr. Singer’s opinions. See

16 Williamson Oil v. Philip Morris, 346 F.3d 1287, 1323 (11th Cir. 2003) (affirming exclusion of

17 expert opinion that “could not have aided a finder of fact to determine whether appellees’ behavior

18 was or was not legal” because “he did not differentiate between legal and illegal pricing behavior”).

19 F. The Model Does Not Account For Major Revenue And Compensation Drivers

20 Dr. Singer’s model is also unreliable and should be excluded because it fails to account for

21 a major driver of event revenues: Zuffa’s investments to promote, produce, and present those very

22
15
23 See, e.g., ABAI Phishing Scam Alert re: Academic Star Publishing Company (accessed on
September 12, 2019), available at ECF No. 743-12.
24 16
See Decl. Ex. 28, SCImago Journal & Country Rank (accessed Nov. 30, 2023). The SCImago
Journal Rank is a recognized means to measure journal rankings. See OMICS Grp., 374 F. Supp.
25 3d at 1017.
17
See Decl. Ex. 29, WorldCat, Journal of Business and Economics (accessed Dec. 1, 2023),
26
[Link]
18
27 See Decl. Ex. 30, Academic Star Publishing Webpage, FAQ (accessed Nov. 30, 2023),
[Link] By contrast, a legitimate “peer review
28 process typically takes several months.” F.T.C. v. OMICS Grp., 374 F. Supp. 3d 994, 1003 (D.
Nev. 2019).
16
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 23 of 32

1 same events. See In re Live Concert Antitrust Litig., 863 F. Supp. 2d 966, 974 (C.D. Cal. 2012).

2 It is well documented (and intuitive) that Zuffa’s expenses contribute to event revenues. Decl. Ex.

3 21, TR1 ¶¶ 133-40 & App. A; Decl. Ex. 23, OR1 ¶ 36. Even Dr. Singer acknowledges that, “as a

4 matter of theory,” Zuffa’s non-compensation expenses, such as marketing, promotion, production,

5 and other factors could influence event revenues, Decl. Ex. 26, Singer Dep. at 119:8-121:21, and

6 that “[t]elevision distributors are interested in Zuffa producing high-quality events that will attract

7 viewers,” Decl. Ex. 24, SR1 ¶ 284. But Dr. Singer failed to include any measure of Zuffa’s event

8 expenses in his initial model. Id. ¶ 185, Tabs. 4-5 (listing regression variables). The exclusion of

9 these revenue drivers is inexplicable, as it eliminates any possibility for the model to estimate that

10 Zuffa’s investments and efforts contributed to increased revenue, and instead attributes any gains

11 from those investments solely to the athletes.

12 Courts exclude opinions when there is “some indication that the excluded variables would

13 have impacted results.” Live Concert Antitrust Litig., 863 F. Supp. 2d at 974. Here, Dr. Singer’s

14 own analysis provides proof of that impact on results (and Zuffa’s expert confirmed it as well).

15 Attempting to correct for the exclusion of event expenses, Dr. Singer presented a new model during

16 the class certification Daubert process that attempted to control for some event “promotion” cost,

17 which is a subset of non-compensation event costs, and the changes to his regression were

18 substantial. Decl. Ex. 22, TR3 ¶ 19. This partial accounting for Zuffa’s costs caused the coefficient

19 on foreclosure share for Dr. Singer’s ranked market (the measure he uses for damages) to decrease

20 by nearly half. Id. ¶ 19; compare Decl. Ex. 24, SR1 Tab. 6 (original coefficient of -0.0427), with

21 Decl. Ex. 25, SR3 Tab. A2 (revised coefficient of -0.0231). That fall-off comes from accounting

22 for just a small portion of Zuffa’s investments to drive event demand. Promotional costs

23 represented, at most, 28% of event costs (excluding athlete compensation). Production costs—the

24 costs of broadcasting events—were far more substantial, often triple the promotion costs;

25 production costs could make up more than half of all event costs (excluding athlete compensation).

26 Decl. Ex. 22, TR3 ¶¶ 17, 18 & Ex. 1.

27 Moreover, Dr. Singer annualized the promotion costs, rather than applying them to his

28 regression on a per-event basis. This was an indefensible choice because the relationship being

17
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 24 of 32

1 studied should have been per-event expenses’ effect on per-event revenues. By instead adding

2 together the entire year’s expenses and applying only the average to each event, Dr. Singer

3 obscured the huge variances in expenses per-event, and how those expenses drive event revenue.

4 For example, in 2016, Zuffa spent over $10 million (excluding athlete compensation) on a single

5 marquee event, which was over 20% of event revenues. Id. at Ex. 5. For another 2016 event,

6 Zuffa spent less than $2 million, and resulting revenues for that event were much lower. Id.

7 Dr. Singer merely assigned $1.1 million in promotional costs for both events (the 2016 annual

8 average), which eliminates the model’s ability to attribute different revenue results from different

9 expenses at each UFC event. Id. at Exs. 1, 5. Event-level cost information was available for

10 promotional expenses, production expenses, and other expenses. Dr. Singer even organized it in

11 his backup materials. But he did not use it in his analysis. Had he done so, his model would have

12 estimated no foreclosure effect, as Prof. Topel found when he tested it. Id. ¶ 22. This is proof of

13 the huge—dispositive—way that Dr. Singer’s exclusion of Zuffa’s expenses per-event “impact[ed]

14 results” from his regression. Live Concert Antitrust Litig., 863 F. Supp. 2d at 974; see also

15 Contreras v. City of L.A., 656 F.2d 1267, 1273 n.4 (9th Cir. 1981) (“Statistics are not trustworthy

16 when minor numerical variations produce significant fluctuations.”); Reed Constr. Data v.

17 McGraw-Hill, 49 F. Supp. 3d 385, 400-04 (S.D.N.Y. 2014), aff’d, 638 F. App’x 43 (2d Cir. 2016).

18 Similarly, with respect to fighters’ revenue share, Dr. Singer’s regression nonsensically

19 shows no relationship between compensation and his variable for a fighter winning a bout (the

20 “WinFlag” variable), even though Dr. Singer admits that 94% of the time fighters double their

21 base pay by winning their fights. Decl. Ex. 24, SR1 ¶ 30; Decl. Ex. 26, Singer Dep. at 111:23-

22 112:17; Decl. Ex. 23, OR1 ¶ 52 & Tab. 1; Decl. Ex. 21, TR1 Ex. 13. Singer’s regression also

23 shows no relationship between variables for Rank and sharing of PPV revenues and fighter share.

24 Decl. Ex. 21, TR1 ¶149; CC2 Hrg. Tr. at 220:7-15, ECF No. 726. But when the model is adjusted

25 to instead use actual compensation as the dependent variable, it produces the expected results of

26 compensation increasing if their rankings improve or if they are entitled to a portion of Zuffa’s

27 PPV revenues. See Decl. Ex. 21, TR1 ¶ 149.

28

18
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 25 of 32

III. The Regression’s Key “Foreclosure Share” Variable Does Not Fit
1
Dr. Singer’s regression is further flawed because the “foreclosure share” variable it relies
2
on to measure antitrust impact does not isolate anticompetitive conduct, and is mismeasured in any
3
event. To show impact, plaintiffs’ “injury must be attributable to an anticompetitive aspect of the
4
practice under scrutiny.” Big Bear Lodging Ass’n v. Snow Summit, Inc., 182 F.3d 1096, 1102 (9th
5
Cir. 1999) (citation omitted). The foreclosure share Dr. Singer utilizes, however, is only a
6
weighted counting of the challenged practice (e.g., 30-month exclusive contracts), without any
7
verification that the practice had any adverse effect on competition. As Dr. Singer explained in
8
deposition, “the regression takes the foreclosure share as an input,” which merely “flows from
9
which market definition and weighting model I use.” Singer Dep. at 39:23-40:3, 40:23-41:8. He
10
assumes that any contract capable of being exclusive for thirty months had an adverse effect on
11
competition, even though that is something plaintiffs must instead prove, and numerous courts
12
have found exclusive contracts even longer than thirty months had no such negative effects.19
13
“Simply put, an expert does not assist the trier of fact” when he “starts his analysis based upon
14
[an] assumption” that is “the very question that he was called upon to resolve.” Clark v. Takata
15
Corp., 192 F.3d 750, 757 (7th Cir. 1999). As other courts have found when Dr. Singer’s testimony
16
relied on assumption rather than verification, this is grounds to exclude his opinions based on
17
“foreclosure share.” See, e.g., Carfax, 2016 WL 7231941, at *13 (rejecting Dr. Singer’s
18
foreclosure conclusions, in part, because “Dr. Singer assumed that all of the ‘exclusive’ CPO
19
agreements foreclosed competition”); Apotex, Inc. v. Cephalon, Inc., 321 F.R.D. 220, 235-37 (E.D.
20

21
19
See, e.g., Fraser v. Major League Soccer, 284 F.3d 47, 68-69 (1st Cir. 2002) (holding that “a
22
garden variety exclusive dealing arrangement limited to three years” is “not inherently unlawful”);
23 Fleer Corp. v. Topps Chewing Gum, Inc., 658 F.2d 139, 142, 144, 153-54 (3d Cir. 1981) (holding
that exclusive athlete agreements of five and eight years did not exclude competition under Section
24 2); Spinelli v. Nat’l Football League, 96 F. Supp. 3d 81, 117 (S.D.N.Y. 2015) (“exclusivity periods
of no more than three years . . . do not foreclose competition and are not anticompetitive as a matter
25 of law” where the contracts subject to re-bidding). Dr. Singer supports his assumption that the 30-
month contracts in this case are exclusionary with only a citation to an antitrust treatise, where the
26
cited section says that “[w]e suggest presumptively that periods of less than one year be approved.”
27 Decl. Ex. 24, SR1 ¶ 172 (citing Areeda & Hovenkamp ¶ 1821d3). Far from saying that all 30-
month exclusive contracts are irrefutably, or even presumptively, anticompetitive, the treatise
28 explains that, when there are a large number of “dealers”—which are the fighters in this context—
“even contracts with long terms need not be anticompetitive.” Areeda & Hovenkamp ¶ 1821d3.
19
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 26 of 32

1 Pa. 2017) (excluding Dr. Singer’s opinions based on assumption lacking factual support); King

2 Drug Co. of Florence v. Cephalon, 2015 WL 12645766, at *6 (E.D. Pa. Dec. 22, 2015) (excluding

3 testimony based on Dr. Singer’s “legal musings” and “legal opinions”); see also Craftsmen

4 Limousine, Inc. v. Ford Motor Co., 2005 WL 3263288, at *7 (W.D. Mo. Dec. 1, 2005) (excluding

5 regression relying on “antitrust indicator” variable where the expert “never analyzes the alleged

6 restraints . . . to determine if they were anticompetitive”), aff’d, 491 F.3d 380 (8th Cir. 2007).

7 Dr. Singer compounds his flawed reliance on foreclosure share by distorting the ratio with

8 a “weight” based on portions of MMA promoters’ average event revenues. See Decl. Ex. 24,

9 SR1 ¶ 128. This approach artificially inflates the estimated foreclosure share—and confuses size

10 for anticompetitive conduct. Decl. Ex. 21, TR1 ¶¶ 162, 224. Moreover, relying on event

11 revenues—which are part of the output market (e.g., MMA events)—to calculate supposed

12 foreclosure in the input market (e.g., fighter services), makes no sense because the exact same

13 fighter, with the exact same rank, would be assigned a higher foreclosure share when fighting for

14 Zuffa than when fighting for any other promoter, merely because Zuffa’s average event revenues

15 were higher than other promoters. See id. ¶¶ 221-23 (proving examples of this flaw in the data).

16 In United States v. Syufy Enterprises, the Ninth Circuit considered the same gimmick as part of

17 market-share calculations and rejected it because there was not a clear relationship between

18 downstream revenues and upstream market power. 903 F.2d 659, 663-64 (9th Cir. 1990).

19 The same outcome is even more deserving here because Dr. Singer includes a weight from

20 the output market not only on the foreclosure share variable, but also on the dependent variable

21 revenue-share variable. With event revenue being counted in both the foreclosure share and the

22 revenue share, the automatic effect of Zuffa’s revenues increasing (all else held constant) is for

23 both the foreclosure share to increase and the fighter’s compensation as a percentage of event

24 revenues to decrease. Thus, the regression is hard-wired to produce a negative correlation between

25 foreclosure share and revenue share. Decl. Ex. 21, TR1 ¶¶ 143, 150, 161-67. 20 As Prof. Topel

26 20
For a “submarket” of his fighter services input market limited only to the top 15 ranked fighters
27 by weight class, Dr. Singer weights fighters based on an inverse measure of their rank, rather than
by the promoters’ revenue. See Decl. Ex. 24, SR1 ¶ 128. Under this weighting, the top-ranked
28 fighter in a weight class is assigned 1,000% more foreclosure share than the tenth-ranked fighter.
Dr. Singer applies this inverse-rank measure only to a submarket (not the relevant market) and it
20
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 27 of 32

1 explained, “[t]his mechanical negative correlation, rather than being evidence of Zuffa’s

2 monopsony power, occurs because Dr. Singer included Zuffa’s event revenue on both the left and

3 right hand side of the regression (as the denominator of the dependent variable and as the weight

4 that increases the key explanatory variable).” Id. ¶ 143. But when the use of revenue-weighting

5 is removed from foreclosure share (and pre-acquisition Strikeforce bouts are removed), the model

6 produces no negative correlation between revenue share and the challenged contracts. Id. ¶ 225.

7 The combined flaws of weighting foreclosure share by revenue and failing to verify that

8 foreclosure share isolates anticompetitive effects creates an unacceptable risk that the jury will

9 make a legally erroneous finding of liability based on permissible conduct. Because Dr. Singer

10 weights his measure of foreclosure share by the promoter’s revenue, his measure of foreclosure

11 share is virtually indistinguishable from his calculation of Zuffa’s market share during the class

12 period.21 The regression “finds a relationship between that foreclosure share and the fighters’

13 wage share controlling for all other things and then the regression is done.” Decl. Ex. 26, Singer

14 Dep. at 41:9-14. But by failing to verify the anticompetitive effect of foreclosure share, and

15 allowing for increased revenues to mechanically increase foreclosure share and decrease wage

16 share, the model is incapable of differentiating between lower compensation arising merely from

17 Zuffa’s increase in size. That, however, would be an impermissible inference, as it is an

18 indisputable “basic rule that mere possession of monopoly power and the practice of charging

19 monopoly prices does not run afoul of § 2.” John Doe 1 v. Abbott Labs., 571 F.3d 930, 934 (9th

20 Cir. 2009); see also Decl. Ex. 21, TR1 ¶ 139 (“[A]ll that Dr. Singer has shown is that as Zuffa’s

21

22
is not the basis of his damages estimate, so it is less relevant. Id. ¶¶ 1, 128, 162, 251. Nonetheless,
23 it has no basis in economic literature or the facts of this case, because there is no basis to assume
such dramatic variability in fighter quality among the top 15 fighters in a weight class. See Decl.
24 Ex. 21, TR1 ¶¶ 228-30.
21
Dr. Singer’s calculated foreclosure share during the class period “fluctuated between 91
25 percent and 98 percent (using the Tracked measure) and between 68 and 90 percent (using the
Ranked measure),” while his calculated market share “fluctuated between 94 and 99 percent (using
26
the Tracked measure) and between 71 and 91 percent (using the Ranked measure).” Decl. Ex. 24,
27 SR1 ¶¶ 129, 173; compare id. at Fig. 1 (market share), with id. at Fig. 3 (foreclosure share). The
closeness between Dr. Singer’s calculated foreclosure share and market share derives from his
28 conclusion that all UFC fighters’ contracts are exclusive, and “[v]irtually all (about 94 percent) of
Zuffa’s [promotion and ancillary rights agreements] contain a champion’s clause.” Id. ¶¶ 66, 69.
21
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 28 of 32

1 business grew relative to its competition, its revenues grew at a faster rate than the earnings of

2 MMA athletes. This reveals nothing about the Challenged Conduct or its effects.”).

3 IV. The Regression Does Not Model Impact On Any Individual Fighter

4 Independent of the flaws with Dr. Singer’s reliance on revenue share and foreclosure share,

5 his impact regression cannot be relied upon to identify the challenged conduct’s impact on any

6 individual fighter, which is the issue at the core of his opinion. See In re Google Play Story

7 Antitrust Litig., 2023 WL 5532128, at *9 (N.D. Cal. Aug. 28, 2023) (excluding Dr. Singer model

8 that “does not give the jury a sound basis on which to make a reasoned and reasonable judgment

9 about antitrust impact and damages”); In re Pharmacy Benefit Managers Antitrust Litig., 2017 WL

10 275398, at *20 (E.D. Pa. Jan. 18, 2017) (finding “insurmountable Daubert fit problem arises from

11 the use of national averages in the expert model since averages cannot demonstrate antitrust impact

12 for individual class members”); In re NCAA Student-Athlete Name & Likeness Licensing Litig.,

13 2013 WL 5979327, at *8 (N.D. Cal. Nov. 8, 2013) (denying class certification where plaintiffs

14 failed to identify “a feasible way to determine which members of the Damages Subclass were

15 actually harmed by the NCAA’s allegedly anticompetitive conduct”).

16 Although Dr. Singer’s descriptions often omit this fact, Dr. Singer does occasionally

17 acknowledge that his impact regression is only intended to measure an average effect, on the

18 average of fighters, not the actual effect on any individual fighter’s share of revenue.22 His impact

19 regression produces only a single impact coefficient associated with foreclosure share for each of

20 his relevant markets, not an individual impact coefficient for each fighter. Decl. Ex. 24, SR1 ¶ 187,

21 Tab. 6; Decl. Ex. 27, Leonard ¶¶ 13, 15-16. For his single measure to reflect the impact to

22 individual fighters, Dr. Singer must assume that the conduct identically affected every fighter.

23 Decl. Ex. 27, Leonard ¶¶ 15-17. That is because, if the fighters experienced different effects, then

24 a single average would not reveal how any individual fighter was affected. A simple example

25
22
E.g., Decl. Ex. 24, SR1 ¶ 187 (“[I]f the Foreclosure Share were to decrease from 90 percent to
26
80 percent, the Fighter Share for the average Fighter would increase by (0.9-0.8)*(0.0319) =
27 0.00319, or about 0.319 percentage points. . . . Given that the average Fighter Share is only about
1.2 percent (as seen in Table 5 above), the average Fighter’s compensation would increase by
28 about 27 percent (equal to 0.00319/0.012) if the Foreclosure Share were to decrease from 90
percent to 80 percent.”); see also Decl. Ex. 27, Leonard ¶¶ 12-17.
22
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 29 of 32

1 makes this clear. If the challenged conduct harmed one fighter (returning a coefficient of -0.04),

2 but did not harm two fighters (returning coefficients of 0.00 and 0.01), the average effect on the

3 group would be a negative coefficient of -0.01, even though two fighters were unharmed. Id. ¶ 16.

4 There are well-recognized and generally accepted checks to determine if the model

5 estimates the same impact for each fighter—but Dr. Singer did not perform them. Id. ¶¶ 18-19.

6 Instead, Dr. Singer commits the error that other courts have repeatedly found in his work—he

7 assumes an identical effect; but does so incorrectly. Id. ¶¶ 18-19; see supra pp.8-9.

8 Because the data was readily available, Dr. Leonard performed a check on the model using

9 generally accepted statistical tests of Dr. Singer’s implicit assumption. The results unambiguously

10 show that foreclosure share does not have the same effect on each fighter’s share of revenue. Decl.

11 Ex. 27, Leonard ¶¶ 19-21, Figs. 1, 2. First, Dr. Leonard performed an F-test for joint

12 significance—a check that Dr. Singer endorses, but did not apply—to test the hypothesis that the

13 foreclosure share effects were identical for all fighters.23 Id. ¶ 19 & n.21. The test rejects that

14 assumption by showing that not allowing foreclosure share to have an effect that varies across

15 fighters is a flawed approach. Id. ¶ 19 n.21. Second, Dr. Leonard affirmatively confirmed that the

16 effect of foreclosure share varied across fighters by applying Dr. Singer’s own regression, but

17 allowing for the coefficient on foreclosure to vary for each fighter. Id. ¶¶ 19-21, Figs. 1, 2. The

18 results of this check show that, for more than 80% of fighters, Dr. Singer’s model does not estimate

19 that the fighter’s share of revenue decreased as foreclosure share increased by a statistically

20 significant amount. Id. ¶ 19, Tab. 1. Thus, because Dr. Singer’s model estimates only one

21 foreclosure coefficient, and there is no evidence that the majority of fighters were harmed by

22 increasing foreclosure share, Dr. Singer’s impact regression cannot determine impact to any

23 individual fighter. Id. ¶ 21.

24 The “compensation structure” and “common factors” regressions that Dr. Singer

25 performed cannot compensate for the flawed impact regression. Dr. Singer’s compensation

26 structure regression does not even attempt to answer the question of whether a fighter’s

27
23
Dr. Singer employs a similar test to see whether the foreclosure share effect was the same for
28 Zuffa fighters and Strikeforce fighters before Zuffa acquired Strikeforce. See Decl. Ex. 24, SR1
¶ 183 n.454; Decl. Ex. 27, Leonard ¶¶ 18-19 n.20.
23
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 30 of 32

1 compensation was correlated with foreclosure share. Id. ¶ 25. Rather, as Dr. Singer acknowledges,

2 it attempts to measure whether changes in compensation “broadly” due to any factor are “shared”

3 across fighters. Decl. Ex. 24, SR1 ¶¶ 228-29. Of course some factor might cause a change in all

4 fighter compensation—general inflation, for example. Decl. Ex. 27, Leonard ¶ 25. But that does

5 nothing to connect the shared movement to the factor of interest—foreclosure share. As described

6 above, Dr. Singer could have checked whether individual fighters’ experiences with foreclosure

7 share were the same, but he does not do so. Id. ¶ 26. Moreover, like his impact regression, the

8 compensation structure regression employed by Dr. Singer is informative, at best, only of the

9 average rather than the individual. Examining the full set of correlations between pairs of

10 individual fighters, Dr. Leonard found that approximately half of the 245,000 pairs of fighter

11 combinations have correlations that are either negative, or are very small (e.g., less than 10%). Id.

12 ¶ 27, Fig. 3. And the average or median compensation correlation for many fighters is less than

13 zero. Id. ¶ 27, Figs. 4, 5. The substantial number of fighters with little or no positive compensation

14 correlation is inconsistent with Dr. Singer’s claim of a common compensation structure. Id. ¶ 27.

15 Dr. Singer’s “common factors regression” fares no better at testing if foreclosure share

16 impacted any individual fighter. This regression purports to identify “what proportion of the

17 variation in Bout Class compensation (above and beyond the average) is attributable to common

18 factors.” Decl. Ex. 24, SR1 ¶ 227. But, incredibly, Dr. Singer excludes foreclosure share from the

19 set of common factors he tested. Decl. Ex. 27, Leonard ¶ 31. By excluding the key variable,

20 Dr. Singer rendered his common factor regression incapable of testing that variable’s effect on

21 individual fighters. Id. ¶ 31. He also failed to check whether any of the “common factors” tested

22 actually had the same effect across fighters. Id. ¶ 32. When Dr. Leonard applied this approach,

23 he found that even the three variables with the highest statistical significance within Dr. Singer’s

24 model did not have the same effect for all fighters. Id. ¶ 32.

25 CONCLUSION

26 For the foregoing reasons, Zuffa respectfully requests that the Court exclude from trial, and

27 consideration at summary judgment, Dr. Singer’s opinions regarding antitrust impact that rely

28 upon his attempts to show a negative correlation between wage share and foreclosure share.

24
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 31 of 32

2 Dated: December 8, 2023 Respectfully Submitted,

3 /s/ Christopher S. Yates

WILLIAM A. ISAACSON (Pro hac vice) CHRISTOPHER S. YATES (Pro hac vice)
4
wisaacson@[Link] [Link]@[Link]
5 JESSICA PHILLIPS (Pro hac vice) LATHAM & WATKINS LLP
jphillips@[Link] 505 Montgomery Street, Suite 2000
6 PAUL, WEISS, RIFKIND, WHARTON & San Francisco, CA 94111
GARRISON LLP Tel: (415) 395-8095
7 2001 K Street, NW
Washington, DC 20006 SEAN M. BERKOWITZ (Pro hac vice)
8
[Link]@[Link]
9 BRETTE M. TANNENBAUM (Pro hac vice) LATHAM & WATKINS LLP
btannenbaum@[Link] 330 North Wabash Ave, Suite 2800
10 YOTAM BARKAI (Pro hac vice) Chicago, IL 60611
ybarkai@[Link]
11 PAUL, WEISS, RIFKIND, WHARTON & LAURA WASHINGTON (Pro hac vice)
12 GARRISON LLP [Link]@[Link]
1285 Avenue of the Americas LATHAM & WATKINS LLP
13 New York, NY 10019 10250 Constellation Blvd, Suite 1100
Los Angeles, CA 90067
14 DONALD J. CAMPBELL (No. 1216)
djc@[Link] DAVID L. JOHNSON (Pro hac vice)
15 J. COLBY WILLIAMS (No. 5549) [Link]@[Link]
16 jcw@[Link] LATHAM & WATKINS LLP
CAMPBELL & WILLIAMS 555 Eleventh Street NW, Suite 1000
17 700 South 7th Street Washington, D.C. 20004
Las Vegas, Nevada 89101
18 Tel: (702) 382-5222 Attorneys for Defendants Zuffa, LLC
19

20

21

22

23

24

25

26

27

28

25
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW
 Case 2:15-cv-01045-RFB-BNW Document 929 Filed 12/08/23 Page 32 of 32

CERTIFICATE OF SERVICE
1

2 The undersigned hereby certifies that the foregoing Motion To Exclude Certain Opinions
3 of Dr. Hal J. Singer was served on December 8, 2023 via the Court’s CM/ECF electronic filing
4 system addressed to all parties on the e-service list.
5

6
/s/ Christopher S. Yates
7 Christopher S. Yates of
LATHAM & WATKINS LLP
8

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

26
Def.’s Mot. to Exclude Certain Singer Opinons No. 2:15-cv-1045-RFB-BNW