— Page 1 —

Van Tharp Position Sizing

Risk Management

Hex Omega — June 2026

A comprehensive system for deciding HOW MUCH to bet on each trade, based on how well
your strategy has been performing recently. Most traders spend all their time deciding WHAT to
buy and WHEN to buy it, but they ignore the most important question: how much should I bet?
This document explains, step by step, the math and logic behind a position-sizing system that
automatically adjusts your bet size based on real-time performance metrics.

Table of Contents
1. R-Multiples — Measuring Every Trade on the Same Scale

2. Expectancy — Your Average Profit Per Trade

3. System Quality Number (SQN) — How Consistent Is Your Edge?

4. Kelly Criterion — The Math-Optimal Bet Size

5. Van Tharp Sizer — Combining Four Health Scores Into One Number

6. Asymmetric Expansion Sizer — Grow Slowly, Shrink Quickly

7. Equity Curve Classifier — Labelling How Your Account Is Doing

8. The Complete Pipeline — How It All Fits Together

9. Configuration Reference — All the Knobs You Can Turn

1. R-Multiples — Measuring Every Trade on the Same
Scale
The "R" in R-multiple stands for "Risk." Before you enter any trade, you always know where
your stop- loss is — that's the price at which you'll exit if the trade goes against you. The distance
from your entry price to your stop-loss is called "1R" — one unit of risk. Every trade result is
then expressed as a multiple of that risk unit. Think of R as a universal yardstick for trading. Just
like metres let you compare the height of a person to the height of a building, R-multiples let you
compare any trade to any other trade — regardless of which market it was in, what price the asset
was, or how big your position was. A small forex trade and a large crypto trade can both be
described as "+2R" if the profit was twice the planned risk. This common language is what
makes all the calculations in this document possible. The beauty of R is that it separates the
quality of a trade from the size of the trade. A $50 profit on a $25 risk is exactly the same quality
(+2R) as a $5,000 profit on a $2,500 risk. By converting everything to R- multiples, we can
measure whether our trading edge is real — regardless of how much money is in the account.
How to Calculate R LONG trade: R = (sell price − buy price) / (buy price − stop price) SHORT
trade: R = (sell price − cover price) / (stop price − sell price) Plain-English Examples Example 1
— A winning long trade: Buy at $100, stop at $97, sell at $109. Risk = $3 ($100 − $97). Profit =
$9 ($109 − $100). R = $9 / $3 = +3.0R — you made 3× your risk. Great trade! Example 2 — A
clean stop-loss exit: Buy at $100, stop at $97, stopped out at $97. R = −$3 / $3 = −1.0R — you
lost exactly what you planned to risk. Example 3 — Slippage: Buy at $100, stop at $97, slippage
exits at $95. R = −$5 / $3 = −1.67R — you lost more than planned. Why R Matters R-multiples
are the foundation that everything else in this document builds on. Without a common measuring
stick, you can't calculate averages, quality scores, or optimal bet sizes. Here's why R is so
important: • It lets you compare any trade to any other trade, regardless of the asset's price or
position size. A quick scalp on a penny stock lives on the same scale as a week-long swing trade
on Bitcoin. • It enforces discipline. You must define your stop-loss BEFORE entering, which
means you always know your worst-case scenario upfront. No more "I'll just hold and hope." • It
makes performance measurement objective. Instead of saying "I made $500" (which means
nothing without knowing the risk), you say "I made +2.5R" — which immediately tells you the
quality of the trade relative to the risk taken. • All subsequent calculations — expectancy, SQN,
Kelly, and the position sizers — use R-multiples as their input. Get this right, and everything
downstream works correctly. Trade Labels by R-Multiple Every closed trade is automatically
labelled based on its R-multiple. These labels help you quickly eyeball your trade history and
spot patterns — for example, are you getting too many LARGE_LOSS trades (suggesting your
stops aren't working), or lots of SCRATCH trades (suggesting you're exiting too early)? +3.0R or
higher LARGE WIN — An exceptional trade +1.0R or higher WIN — A standard profitable
trade +0.0R or higher SCRATCH — Roughly break-even −1.0R or higher SMALL LOSS —
Stopped out as planned (good discipline) Worse than −1.0R LARGE LOSS — Lost more than
planned (slippage, gap)

All trades are tracked on a rolling window of the last 100 trades. This means the system always
reflects your recent performance rather than ancient history. If your strategy worked brilliantly
six months ago but has been struggling lately, the rolling window ensures the sizing adjusts to
what's happening NOW, not what happened in the distant past.

2. Expectancy — Your Average Profit Per Trade

Expectancy answers the most fundamental question about any trading system: "On average, how
many R do I make per trade?" It's the single number that tells you whether your strategy makes
money or loses money over time. Think of it like a casino. A casino doesn't win every hand of
blackjack — it loses plenty of individual hands. But it knows that on average, across thousands
of hands, it earns a small edge per hand. That average edge is the casino's expectancy. If your
trading system has a positive expectancy, it means that even though you'll have plenty of losing
trades, the math is on your side over the long run. If expectancy is negative, you're the gambler,
not the casino — and no amount of clever position sizing can save a system that loses money on
average. Expectancy combines two things: how often you win versus lose (your win rate), and
how big your wins are compared to your losses (your payoff ratio). You can have a profitable
system that only wins 30% of the time, as long as the average win is much larger than the
average loss. Conversely, you can win 80% of the time and still lose money if your losers are
huge relative to your winners. Formula Expectancy = average of all recent R-multiples

Alternatively:
Expectancy = (win rate × average winning R) + (loss rate × average losing R) Worked Example
Win 40% of the time, average win = +2.5R Lose 60% of the time, average loss = −1.0R
Expectancy = (0.40 × 2.5) + (0.60 × −1.0) = 1.0 − 0.6 = +0.40R per trade Meaning: on average,
every trade earns you 0.40 times your risk. If you risk $100 per trade, you'd expect to make $40
per trade on average. That doesn't mean every trade makes $40 — some will lose $100, some
will make $250 — but across many trades, the average settles to +0.40R. This is powerful
because it lets you estimate future returns: if you plan to take 200 trades this year at $100 risk
each, you'd expect to earn roughly 200 × $40 = $8,000 (before fees and slippage). Expectancy
Ratings We grade your expectancy into five tiers. This grading determines how the Van Tharp
Sizer (Section 5) adjusts your bet size — higher-rated expectancy earns a larger sizing bonus,
while poor expectancy triggers protective reductions: +1.5R or higher EXCELLENT — An elite
system +0.5R to +1.5R GOOD — A solid, profitable edge +0.1R to +0.5R MARGINAL — A
small edge; needs lots of trades −0.1R to +0.1R BREAKEVEN — No meaningful edge Below
−0.1R POOR — Losing money on average; stop trading

Expectancy is calculated from the last 30 trades. We use a relatively short window because
markets change, strategies evolve, and recent performance is more relevant than what happened
months ago. If fewer than 10 trades have been recorded (for example, when the bot first starts),
we assume GOOD as a starting default — giving the system the benefit of the doubt until there's
enough data to judge.

3. System Quality Number (SQN) — How Consistent Is

Your Edge?
Expectancy tells you the average profit per trade, but it doesn't tell you whether your results are
consistent or wildly all over the place. Two systems can have the exact same expectancy of
+0.5R but feel completely different to trade. SQN (System Quality Number) answers the deeper
question: "Is this edge real and reliable, or just a lucky streak that could vanish at any moment?"
Imagine two restaurants. Restaurant A consistently serves meals rated 8 out of 10. Restaurant B
alternates between spectacular 10s and terrible 2s — the average is still around 8, but you'd never
know what you're going to get. SQN is like a Yelp rating that accounts for both quality AND
consistency. In trading, consistency matters enormously because inconsistent results make it
psychologically brutal to keep following the system, and they make it harder to distinguish a real
edge from random luck. Formula SQN = (average R / standard deviation of R) × √(number of
trades) Let's break this formula down in plain English. The "average R / standard deviation of R"
part is basically asking: "How big is the average win compared to the typical swing between
good and bad trades?" If your average R is high and the swing (standard deviation) is low, you
have a rock-solid system. The "square root of number of trades" part rewards systems with more
data — because 30 trades of consistent results is more convincing than 5 trades of consistent
results. A system that wins $1 per trade like clockwork scores much higher than a system that
alternates between winning $10 and losing $8, even if the average is similar. In the first case, you
can sleep at night knowing each trade will be near +$1. In the second case, you're riding an
emotional rollercoaster, and any given week could wipe out the previous month's gains. SQN
Ratings (Van Tharp's Scale) 5.0 or higher SUPERB — "Holy Grail" territory; extremely rare 3.0
to 5.0 EXCELLENT — A top-tier system 2.0 to 3.0 GOOD — A solid, reliable system 1.0 to 2.0
AVERAGE — Acceptable, worth trading 0.0 to 1.0 POOR — Below average; be cautious Below
0.0 AVOID — The system is losing money

Like expectancy, SQN is calculated from the last 30 trades. If fewer than 10 trades have been
recorded, it defaults to GOOD. Van Tharp himself considered SQN the single most important
measure of a trading system's quality — it's the metric he used to decide whether a system was
worth trading at all.

4. Kelly Criterion — The Math-Optimal Bet Size

The Kelly Criterion is a famous formula originally developed by Bell Labs scientist John L.
Kelly Jr. in 1956 for optimising data transmission over noisy phone lines. It was quickly adopted
by gamblers and investors because it solves a fundamental problem: given that you have an edge
(positive expectancy), what is the mathematically ideal percentage of your bankroll to risk on
each bet to maximise your long- term wealth growth? The intuition is straightforward. If you bet
too little, you leave money on the table — your account grows slowly even though you have a
real edge. If you bet too much, you risk catastrophic losses that wipe out your gains. Kelly finds
the sweet spot that balances these two extremes. It's the bet size that, over many repeated bets,
produces the fastest possible compounding of your account equity. Formula Full Kelly = (win
rate / average loss size) − (loss rate / average win size) Worked Example Win 55% of the time.
Average win = +1.4R. Average loss = −1.0R. Full Kelly = (0.55 / 1.0) − (0.45 / 1.4) = 0.55 −
0.321 = 0.229 (22.9%)

"Bet 22.9% of your account on each trade for maximum growth." The Problem with Full Kelly
Full Kelly has a critical flaw: it assumes your estimates of win rate and payoff ratio are perfectly
accurate. In reality, they never are. Your win rate might be 55% over the last 30 trades, but the
"true" win rate could be anywhere from 45% to 65% — you're working with a small sample. If
your estimates are even slightly too optimistic, Full Kelly can lead to devastating drawdowns.
Academic studies show that Full Kelly can produce drawdowns of 50% or more even with a
genuine edge, simply because the variance is so high. This is why almost nobody — not
professional gamblers, not hedge funds, not quantitative traders — uses Full Kelly at face value.
It's a theoretical maximum, like driving your car at its maximum RPM: technically possible, but
practically disastrous. Our Approach — Quarter Kelly We take the Full Kelly number and
multiply it by 0.25 (one quarter). This is called "Quarter Kelly" and it's a widely respected
approach in professional trading and gambling circles. The math works out beautifully: Quarter
Kelly retains about 75% of the theoretical maximum growth rate while reducing the chance of a
devastating drawdown by roughly 80%. In other words, you give up only a small fraction of
potential profit in exchange for dramatically smoother, more survivable equity growth. Think of
it this way: Full Kelly is like sprinting as fast as you possibly can. Quarter Kelly is like jogging at
a comfortable pace — you still get to the finish line, but you don't collapse halfway there.
Quarter Kelly = Full Kelly × 0.25 Example: 22.9% × 0.25 = 5.7% of your account per trade
Converting Kelly to a Multiplier kelly_multiplier = Kelly risk percentage / base risk percentage

Capped between 0.50 and 2.00.

Example: Kelly says 1.5%, base risk 1% → multiplier = 1.50 Example: Kelly says 0.3%, base
risk 1% → multiplier = 0.30 → capped at 0.50 Kelly needs at least 10 trades to produce a
meaningful calculation. With fewer than 10 trades, the win rate and average payoff estimates are
too unreliable — you might happen to have won three out of three trades, giving a 100% win rate
that is obviously not sustainable. Until 10 trades are recorded, Kelly stays neutral at a multiplier
of 1.0 (no adjustment up or down).

5. Van Tharp Sizer — Combining Four Health Scores

The Van Tharp Sizer is the centrepiece of the position-sizing system. It looks at four different
aspects of your recent trading performance — your account trend, your win/loss streak, your
average profitability, and your consistency — and multiplies them together to produce a single
number that answers the question: "How healthy is our trading right now?" If things are going
well across all four dimensions, the multiplier climbs above 1.0, telling the bot to bet more than
normal. If things are going badly, the multiplier drops below 1.0, automatically shrinking bet
sizes to protect capital. This is the core risk management principle: bet more when you're
winning and everything is aligned, bet less when you're struggling. The four factors are
independent of each other, so even one bad signal will pull the overall multiplier down — it takes
strength across ALL dimensions to earn a big bet size. multiplier = equity_curve × streak ×
expectancy × sqn Factor 1: Equity Curve This factor looks at your account balance relative to its
highest-ever point (the "high-water mark"). If your account is making new all-time highs, the
system is working well and we lean into it with bigger bets. If the account has pulled back
significantly, something may be wrong — the market might have changed, or the strategy may be
in a rough patch — so we pull back on sizing to protect what's left. At all-time highs 1.20 — Bet
20% more Small dip (< 10%) 1.05 — Bet 5% more Moderate dip (10–20%) 0.65 — Bet 35%
less Significant dip (20–35%) 0.80 — Bet 20% less Severe dip (> 35%) 0.50 — Bet 50% less

Factor 2: Streak
Streaks matter because they often signal a regime change in the market. If you've hit 5+ winners
in a row, the strategy is likely well-aligned with current market conditions — so we lean in
slightly. If you've hit 4+ losers in a row, the market may have shifted to a regime where the
strategy doesn't work well — perhaps volatility spiked, or a previously trending market went
sideways. Pulling back during a cold streak gives the system room to recover without digging a
deeper hole. Streaks are detected by looking at the last 10 closed trades and counting how many
consecutive wins or losses appear at the end. 5+ wins in a row 1.20 — Ride the momentum 3–4
wins in a row 1.00 — No change Mixed results 1.00 — No change 3–4 losses in a row 0.80 —
Cool off 5+ losses in a row 0.80 — Cool off

Factor 3: Expectancy
This is the same expectancy metric from Section 2, but here it's converted into a sizing factor. If
your recent trades have been averaging excellent R-multiples, the system is confident in the edge
and rewards it with larger sizing. If expectancy has dropped to break-even or worse, it's a
warning sign that the strategy may be losing its edge, and the sizing shrinks accordingly to limit
damage. EXCELLENT (> +1.5R) 1.20 GOOD (+0.5R to +1.5R) 1.10 MARGINAL (+0.1R to
+0.5R) 1.00 BREAKEVEN (−0.1R to +0.1R) 0.85 POOR (< −0.1R) 0.70

Factor 4: SQN
The same SQN score from Section 3, converted to a sizing factor. A high SQN means your
results are consistent and predictable, so the system trusts them and sizes up. A low or negative
SQN means results are erratic or the system is losing money, so the sizing factor drops to protect
capital. This factor adds an important layer of protection: even if expectancy looks decent, erratic
results (low SQN) will still trigger caution. SUPERB (> 5) 1.20 EXCELLENT (3–5) 1.15 GOOD
(2–3) 1.05 AVERAGE (1–2) 1.00 POOR (0–1) 0.90 AVOID (< 0) 0.75

Worked Examples
Let's walk through two complete examples to show how the four factors combine. Notice how
the multiplication means that one bad factor can significantly drag down the entire result — this
is intentional, as it ensures the system is conservative when even one dimension shows weakness.
Everything going well: Equity at highs (1.20) × hot streak 5+ wins (1.20) × good expectancy
(1.10) × good SQN (1.05) = 1.20 × 1.20 × 1.10 × 1.05 = 1.663 → Bet about 66% more than
normal. Things going badly: Correction, 15% below highs (0.65) × cold streak 4 losses (0.80) ×
marginal expectancy (1.00) × average SQN (1.00) = 0.65 × 0.80 × 1.00 × 1.00 = 0.520 → Bet
about 48% less than normal (roughly half size).
6. Asymmetric Expansion Sizer — Grow Slowly, Shrink
Quickly
This sizer follows one of the most important principles in Van Tharp's work: the speed of risk
adjustment should be ASYMMETRIC. When you're making money, increase your bet size
SLOWLY. When you're losing money, decrease your bet size FAST. Much faster. Why the
asymmetry? Consider the math of drawdowns. If you lose 50% of your account, you need a
100% return just to get back to break-even. Losses hurt more than equivalent gains help, so you
need to clamp down on risk much more aggressively when you're losing than you ramp it up
when you're winning. Slowing down on the way up also prevents a common trap: you've been
winning, so you size up, and then the market turns right as you're at your biggest position size.
By expanding slowly, you're less likely to be over-leveraged at the exact wrong moment. The
sizer tracks your "high-water mark" (HWM) — the highest your account balance has ever been
— and divides your current situation into three zones based on where your account stands
relative to the HWM and your starting capital: Profit Zone Your account is at or above its
all-time high — you're in uncharted territory, making new money. The sizer gradually increases
risk above the base level, because a strategy that's consistently making new highs is clearly
working well in the current market environment. However, there's a hard ceiling of 2.5% risk per
trade so it never gets reckless — even when everything is going perfectly, the system maintains
discipline. Base Zone Your account is above your starting capital but below its all-time high.
You had a peak, then pulled back, but you're still in profit overall. The sizer uses the default base
risk (1%) — no bonus, no penalty. Think of this as "recovery mode": you're not in trouble, but
you're not setting new records either. The system waits patiently for you to either make new
highs (re-entering the Profit Zone) or recover from the dip. Drawdown Zone Your account is
below your starting capital — you're losing real money. This is where the asymmetric
contraction kicks in hardest. The sizer rapidly shrinks your risk, and the deeper the drawdown,
the smaller your bets become. There's a floor of 0.2% so you never bet zero (you want to keep
trading so you can recover), but bets get very small. The logic is simple: when you're in a hole,
the most important thing is to stop digging. Tiny bets limit further damage while still giving you
a chance to catch a winning streak and climb back. Key Numbers Base risk 1% of account per
trade Expansion rate +0.6% risk per 1% profit above HWM Contraction rate −1.6% risk per 1%
drawdown below start Speed difference Risk shrinks 2.67× faster than it grows Maximum risk
2.5% of account (ceiling) Minimum risk 0.2% of account (floor)
 Profit Cushion Mode (Safer Variant)
Profit Cushion Mode adds an extra layer of safety for traders who want to protect their initial
investment at all costs. Instead of calculating risk based on the total account balance, this mode
only risks money from your "profit cushion" — the amount of money above your starting capital.
This means your original seed money is essentially untouchable by the position sizer; only
"house money" (profits you've already earned) gets risked at elevated levels. This is particularly
useful for traders who are deploying capital they can't afford to lose, or who want to
psychologically separate "my money" from "the market's money." It's the safer of the two modes,
at the cost of slightly slower account growth during winning periods. Example: Started with
$10,000, now at $12,000. Profit cushion = $2,000. Cushion-based risk = ($2,000 / $12,000) ×
0.50 = 8.33% → Capped at 2.5% ceiling. Converting to a Multiplier ae_multiplier = calculated
risk / base risk Capped between 0.50 and 2.00.

In profit with 1.5% risk → 1.5 / 1.0 = 1.50× (bet 50% more) In drawdown with 0.5% risk → 0.5 /
1.0 = 0.50× (bet 50% less)

7. Equity Curve Classifier — Labelling Account Health

This is NOT a standalone sizer — it doesn't directly change your bet size. Instead, it's a simple
diagnostic tool that measures how far your current account balance has dropped from its
highest-ever point (the high-water mark) and assigns a human-readable label. That label is then
fed into the Van Tharp Sizer (Section 5) as the Equity Curve factor. Think of it like a
speedometer for your drawdown. At a glance, you can see whether your account is at all-time
highs ("ADVANCING" — green light), in a minor pullback ("PULLBACK" — no alarm), or in
deep trouble ("CRISIS" — red alert). The labels borrow terminology from stock market
commentary: just as the financial news uses "pullback," "correction," and "bear market" to
describe different severity levels of a market decline, this classifier uses similar thresholds to
describe your personal account's drawdown. Formula drawdown = (highest balance ever −
current balance) / highest balance ever Classification Labels Each label corresponds to a
drawdown band. The thresholds are based on common market conventions and Van Tharp's
research on the psychological and mathematical significance of different drawdown levels:
Drawdown = 0% ADVANCING — Making new all-time highs Up to 10% PULLBACK — A
minor, normal dip 10–20% CORRECTION — A noticeable drop; pay attention 20–35%
DRAWDOWN — A significant decline; get cautious Over 35% CRISIS — A deep hole;
consider pausing
8. The Complete Pipeline — How It All Fits Together
Now that you understand each individual component, let's see how they all work together as a
unified pipeline. When the trading bot detects a new trade opportunity (a signal from your
strategy), it doesn't just blindly buy a fixed amount. Instead, it runs through a carefully ordered
sequence of steps that evaluate the health of your trading, the state of the market, and the math of
optimal bet sizing — and uses all of that information to decide exactly how much money to put
into the trade. The key insight is that each sizer approaches the question from a different angle:
Van Tharp looks at four performance dimensions, Kelly uses pure mathematical optimisation,
and the Asymmetric Expansion sizer focuses on profit/loss dynamics. By running all three and
taking the most optimistic result, we get the benefit of diversified analysis — if even one model
sees a reason to trade bigger, we trust it. The regime multiplier then applies a final reality check
based on market conditions. Step 1: Get the starting bet size Freqtrade — the trading bot
framework — proposes a default bet size based on your configuration. For example, if you have
$10,000 and allow 5 open positions at once, it might suggest $2,000 per trade (splitting the wallet
equally). This is the "raw" bet size before any adjustments. Step 2: Run the Van Tharp Sizer (if
enabled) The system checks all four health dimensions — equity curve trend, win/loss streak,
recent expectancy, and SQN consistency — and multiplies the four factors together into a single
multiplier. If all four look healthy, the multiplier exceeds 1.0; if any are struggling, it pulls the
number down. Step 3: Run the Kelly Criterion (if enabled, ≥ 10 trades) Using your actual win
rate and payoff ratio from recent trades, the system calculates the Full Kelly percentage, then
takes 25% of it (Quarter Kelly) for safety. This is converted to a multiplier capped between 0.50
and 2.00. Kelly provides a pure mathematical perspective independent of the other sizers. Step 4:
Run the Asymmetric Expansion Sizer (if enabled) The system checks whether you're in the Profit
Zone, Base Zone, or Drawdown Zone relative to your high-water mark and starting capital. It
calculates an adjusted risk percentage that expands slowly in profits and contracts quickly in
drawdowns, then converts that to a multiplier capped between 0.50 and 2.00. Step 5: Pick the
largest multiplier Of all the active sizers, the system takes the highest (most optimistic)
multiplier. The philosophy: each sizer looks at the situation from a different angle. If ANY one
of them sees a reason to be more aggressive, we trust that signal. This prevents overly
conservative behaviour when one model is bearish but the others see opportunity. Step 6: Apply
market regime and clamp The best sizer multiplier is then multiplied by the market regime factor
— a separate system (not covered in this document) that assesses whether the overall market is
trending strongly, ranging, or in turmoil. The final combined multiplier is hard-clamped between
0.30 (minimum — never less than 30% of the base bet) and 2.00 (maximum — never more than
double). This ensures the system can never go to extremes regardless of what the models say.
Step 7: Update the high-water mark After the bet size is decided, the system checks whether the
current account balance is a new all-time high. If it is, it records this as the new high-water mark,
which becomes the reference point for future drawdown calculations. Example A: Everything Is
Going Great Account: $15,000 (started $10,000) Freqtrade suggested bet: $1,500 Market regime:
strong trend → regime multiplier = 1.25

Equity: at highs (1.20) | Streak: 6 wins (1.20) Expectancy: +0.8R Good (1.10) | SQN: 2.5 Good
(1.05) Van Tharp = 1.20 × 1.20 × 1.10 × 1.05 = 1.663

Kelly → 1.80 | AE → 2.00 (capped) Best sizer = max(1.663, 1.80, 2.00) = 2.00

Combined = 1.25 × 2.00 = 2.50 → capped at 2.00 Final bet = $1,500 × 2.00 = $3,000 Result: The
bot doubles the suggested bet because every measure of health is positive — the account is at
highs, there's a strong winning streak, the strategy is profitable and consistent, and the market is
trending strongly. This is the time to press the advantage. Example B: Things Are Going Badly
Account: $8,500 (started $10,000 — down 15%) Freqtrade suggested bet: $850 Market regime:
choppy → regime multiplier = 0.55

Equity: 15% below highs, Correction (0.65) | Streak: 4 losses (0.80) Expectancy: +0.05R
Breakeven (0.85) | SQN: 0.5 Poor (0.90) Van Tharp = 0.65 × 0.80 × 0.85 × 0.90 = 0.398

Kelly → 0.50 (floor) | AE → 0.60 Best sizer = max(0.398, 0.50, 0.60) = 0.60

Combined = 0.55 × 0.60 = 0.33 Final bet = $850 × 0.33 = $280.50 Result: The bot shrinks the bet
to about one-third of what it would normally be. The strategy is struggling (breakeven
expectancy, poor SQN), the account is in a meaningful drawdown, there's a losing streak, and the
market is choppy with no clear trend. By making tiny bets, the system preserves the remaining
$8,500 so that when conditions improve, there's still meaningful capital to work with. Without
this protection, a series of full-size bets during a bad stretch could destroy the account.

9. Configuration Reference — All the Knobs You Can

Turn
This section lists every configurable parameter in the position-sizing system. The defaults have
been carefully chosen and should work well for most strategies and market conditions. You
generally don't need to change them unless you have a specific reason — for example, if you're
trading a very high- volatility asset class and want tighter risk limits, or if backtesting reveals that
different thresholds produce better results for your specific strategy. Sizer Blending When
multiple sizers are active, the system takes the LARGEST multiplier from whichever sizers are
enabled. For example, if Van Tharp says 1.5×, Kelly says 1.2×, and AE says 1.8×, the blend
result is 1.8×. The most optimistic sizer wins. The rationale is that each sizer analyses the
situation from a different perspective, and if even one of them sees strong conditions for a larger
bet, we allow it. The regime multiplier and the final clamp (0.30 to 2.00) still provide guardrails
against excessive risk. Risk Parameters PDR_BASE_RISK_PCT = 0.01 Default: risk 1% of
account per trade PDR_MAX_RISK_PCT = 0.03 Hard ceiling: never risk more than 3%
PDR_MIN_RISK_PCT = 0.001 Hard floor: never risk less than 0.1%
PDR_MAX_POSITION_PCT = 0.20 Never let one position exceed 20% of account

Asymmetric Expansion Parameters

AE_EXPANSION_RATE = 0.60 +0.6 pp risk per 1% profit above HWM
AE_CONTRACTION_RATE = 1.60 −1.6 pp risk per 1% drawdown (≈ 3× faster)
AE_USE_CUSHION = True Only risk from profit above starting capital

Final Multiplier Limits

Floor = 0.30 At worst, bet 30% of the suggested amount Ceiling = 2.00 At best, bet 200% of the
suggested amount

Van Tharp Position Sizing — Reference Document