QUANT STRATEGY

PLAYBOOK

Mean Reversion, Momentum, Factor Investing

A Practical Guide to Quantitative Trading
 What is Quant Strategy?

Definition:
Quant strategy uses mathematical models and data analysis to identify trading opportuni-
ties. Rules are explicit and testable. Decisions come from statistical evidence rather than
discretion. Systematic execution removes emotion from the process. The approach applies
across all asset classes and timeframes.
P
Expected Return Decomposition: E[R] = Rf + βmarket · MRP + i βi · Factori + α
Core Philosophy:
Markets exhibit patterns that persist over time. Statistical relationships can be measured
and exploited. Diversification across signals reduces strategy-specific risk. Transaction costs
must be smaller than expected returns. Backtesting provides evidence but forward testing
validates.

What Quant Strategy Requires:

Rigorous data analysis and statistical testing. Understanding of market microstructure and
execution. Risk management frameworks that adapt to changing conditions. Continuous
monitoring and strategy refinement. Acceptance of drawdown periods when markets shift.
 Foundation Concepts

Signal Construction:
Signal transforms market data into trading direction. Strength indicates conviction level.
Signals combine features using statistical relationships. Normalization ensures comparability
across assets. Signal decay determines holding period.
Alpha vs Beta:
Beta represents systematic market exposure. Alpha measures return above benchmark
adjusted for risk. Pure alpha strategies hedge market exposure. Factor exposure sits between
alpha and beta. Portfolio construction separates desired from unintended exposures.
E[Rp −Rb ] Alpha
Information Ratio: IR = σRp −Rb = Tracking Error

Strategy Capacity:
Capacity measures maximum capital before returns degrade. Market impact increases non-
linearly with position size. Lower frequency strategies have higher capacity. Crowding
reduces available alpha as more capital chases signal. Capacity constraints force diversifi-
cation across strategies.
 Mean Reversion: Fundamentals

Core Principle:
Prices that deviate from fundamental value tend to revert. Short-term overreactions create
profit opportunities. Supply and demand imbalances correct over time. Liquidity provision
and market making exploit mean reversion. Statistical tests identify mean-reverting pro-
cesses.
Ornstein-Uhlenbeck Process: dXt = θ(µ − Xt )dt + σdWt where θ is reversion speed, µ
is long-term mean.
Entry and Exit Logic:
Define equilibrium level using moving average or fair value. Measure deviation using standard
deviations or z-scores. Enter when price exceeds threshold distance from equilibrium. Exit
when price returns to equilibrium or stop loss triggers. Position size scales with deviation
magnitude.
Pt −µt
Z-Score Calculation: zt = σt Trade when |zt | > k for threshold k ∈ [1.5, 3.0].

Regime Change Risk:

Mean can shift during structural changes. What appears mean-reverting may be trending.
Stop losses protect against continued divergence. Shorter lookback windows adapt faster
to regime changes. Combine with trend filters to avoid catching falling knives.
 Pairs Trading Strategy

Strategy Mechanics:
Pairs trading exploits relative mispricing between related securities. Long undervalued asset
and short overvalued asset. Market-neutral construction hedges systematic risk. Profit
comes from spread convergence regardless of market direction. Cointegration ensures long-
run relationship stability.
Cov(PA ,PB )
Hedge Ratio: β = Var(PB ) Spread: St = PtA − βPtB
Pair Selection Process:
Screen universe for fundamental similarity. Same sector, related business models, or input-
output relationships. Test for cointegration using Engle-Granger or Johansen tests. Calcu-
late half-life of mean reversion. Reject pairs with half-life too long or unstable. Monitor
rolling correlation to detect relationship breakdown.
ln(2)
Half-Life Estimation: Half-Life = λ where λ is mean reversion rate from AR(1) model.

Execution Challenges:
Leg risk arises from non-simultaneous execution. One side fills while other side does not.
Borrowing costs for short positions reduce profitability. Liquidity differences between legs
create basis risk. Transaction costs consume profits on tight spreads. Slippage increases
with position size.
 Statistical Arbitrage

Multi-Asset Mean Reversion:

Statistical arbitrage extends pairs trading to portfolios. Identify baskets of securities with
mean-reverting relationships. Principal component analysis extracts common factors. Trade
deviations from factor model predictions. Portfolio approach diversifies idiosyncratic risk.
Factor Model Construction:
Returns decompose into systematic and idiosyncratic components. Systematic risk comes
from exposure to common factors. Residual returns exhibit mean reversion properties.
PCA identifies factors explaining maximum variance. Trade residuals while hedging factor
exposure.
PK
Return Decomposition:
P Ri = αi + k=1 βik Fk + ϵi Trade signal based on ϵi while neu-
tralizing βik .

Portfolio Construction:
Weight assets by signal strength and conviction. Constrain gross and net exposure. Optimize
for target volatility and maximum drawdown. Rebalance when positions deviate from target.
Transaction cost optimization balances turnover against signal decay.
 Momentum: Fundamentals

Core Principle:
Assets with strong recent performance continue outperforming. Momentum persists across
asset classes and timeframes. Behavioral biases create under-reaction to information. Trend-
following captures directional moves. Statistical evidence spans decades of market data.
Return Predictability: E[Rt+1 | Rt ] = α + βRt where β > 0 indicates momentum.
Time-Series Momentum:
Each asset trades based on its own past returns. Positive past returns predict positive future
returns. Implementation uses lookback windows from 1 to 12 months. Skip most recent
month to avoid microstructure reversals. Volatility scaling normalizes position sizes across
assets.
P 
13 σtarget
Signal Construction: Signalt = sign i=2 rt−i · σ̂t where rt−i is monthly return i
months ago.

Cross-Sectional Momentum:
Rank assets by past performance. Buy top performers and sell bottom performers. Market-
neutral by construction through long-short portfolio. Holding period determines strategy
frequency. Monthly rebalancing balances signal persistence against costs.
 Momentum Implementation

Lookback Window Selection:

Short lookbacks (1-3 months) capture recent trends. Medium lookbacks (6-12 months) rep-
resent classic momentum. Long lookbacks (12-24 months) detect sustained trends. Multiple
timeframes provide diversification. Combine signals using equal or volatility weighting.
Momentum Crashes:
Sharp reversals occur during market stress. Winners become losers rapidly in panic selling.
Losses concentrate in short positions during rebounds. Volatility scaling reduces exposure
in high-volatility regimes. Stop losses limit downside but create whipsaw risk. Dynamic
leverage adjustment improves risk-adjusted returns.
σtarget
Volatility Scaling: wt = w ∗ · σt where w ∗ is base weight, σt is realized volatility.

Risk Management:
Target constant volatility through position sizing. Reduce exposure when volatility spikes.
Combine with mean reversion for diversification. Monitor correlation between momentum
and market returns. Increase cash allocation during high correlation periods.
 Factor Investing: Overview

What Are Factors:

Factors are characteristics that explain return differences. Systematic factors represent com-
mon risk sources. Stock-specific factors capture individual attributes. Academic research
identifies persistent factor premiums. Factor investing tilts portfolios toward rewarded char-
acteristics.
Multi-Factor Model: Ri = αi + βi,1 F1 + βi,2 F2 + · · · + βi,K FK + ϵi
Factor Premiums:
Value premium rewards buying cheap assets. Size premium compensates for small-cap risk.
Momentum premium captures trend persistence. Quality premium pays for stable earnings.
Low volatility premium arises from leverage constraints. Each factor delivers positive long-
run returns.

Factor Exposure:
Portfolio returns decompose into factor contributions. Factor loadings measure sensitivity to
each factor. Pure factor portfolios isolate individual factor returns. Factor timing attempts
to predict factor performance. Most implementations use static factor tilts.
 Value Factor

Value Principle:
Value stocks trade below fundamental value. Price-to-book, price-to-earnings, and dividend
yield identify value. Academic evidence shows value premium across markets. Behavioral
explanation involves overreaction to bad news. Risk-based explanation links value to distress
risk.
   
1 1
Value Score: Value = rank P/B + rank P/E + rank(Div Yield)
Implementation Details:
Multiple valuation metrics reduce noise. Book value adjusts for accounting treatment.
Earnings use normalized or forward estimates. Enterprise value ratios account for debt.
Composite scores combine multiple measures. Universe screening removes distressed stocks.

Value Traps:
Cheap stocks can get cheaper during deterioration. Permanent impairment destroys value
thesis. Financial distress leads to bankruptcy. Combine value with quality to avoid traps.
Momentum overlay exits deteriorating positions. Diversification limits single-stock impact.
 Quality Factor

Quality Characteristics:
Quality measures business fundamentals strength. High profitability indicates competitive
advantage. Stable earnings reduce uncertainty. Low leverage decreases financial risk. Ef-
ficient capital allocation drives shareholder value. Quality stocks outperform over long
horizons.
Quality Metrics:
Return on equity measures profitability. Earnings stability calculated from time series. Ac-
cruals ratio detects earnings manipulation. Debt-to-equity captures leverage. Profit margin
expansion signals pricing power. Composite quality score combines metrics.
1 1 1
Quality Score: Quality = w1 · ROE + w2 · σEarnings + w3 · Accruals + w4 · D/E

Quality-Value Combination:
Quality and value historically exhibited negative correlation. Combining factors improves
risk-adjusted returns. Quality filters remove value traps. Value discipline prevents overpaying
for quality. Equal weighting or optimization determines allocation.
 Low Volatility Factor

Low Volatility Anomaly:

Low-risk stocks deliver higher risk-adjusted returns. Contradicts capital asset pricing model
predictions. Leverage constraints force investors into high-beta stocks. Behavioral biases
favor lottery-like payoffs. Institutional constraints limit arbitrage of anomaly.
Risk-Adjusted Return: Sharpelow vol > Sharpehigh vol
Implementation Approaches:
Minimum variance optimization minimizes portfolio volatility. Low volatility portfolios
weight by inverse volatility. Low beta portfolios target market beta below one. Rank-
ing by volatility constructs long-short factor. Each approach captures low-risk premium
differently.
Minimum Variance: minw w T Σw subject to
P
wi = 1 where Σ is covariance matrix.

Crowding Concerns:
Low volatility investing gained popularity recently. Increased capital may reduce future
returns. Valuations of low-vol stocks elevated relative to history. Performance cyclical with
underperformance in bull markets. Diversification across factors mitigates concentration
risk.
 Multi-Factor Strategies

Factor Diversification:
Combining factors reduces strategy-specific risk. Factor returns exhibit low correlation.
Diversification benefit from factor rotation. Single-factor strategies experience extended
drawdowns. Multi-factor portfolios smooth return profiles.

Portfolio Construction Methods:

Sequential approach applies factors in order. Combination approach blends factor scores.
Optimization maximizes factor exposure subject to constraints. Equal weighting provides
naive diversification. Risk parity weights by factor volatility contribution.
PK
Combined Score: Scorei = k=1 wk · Factork,i where wk are factor weights.
Factor Timing:
Factor performance varies across market regimes. Value performs during recoveries. Mo-
mentum thrives in trending markets. Quality outperforms during stress. Timing factors
requires regime prediction. Static allocation avoids timing risk.
 Backtesting Framework

Data Requirements:
Historical price data with adjustments for splits and dividends. Fundamental data aligned to
reporting dates. Point-in-time data prevents lookahead bias. Survivorship-bias-free universe
includes delisted stocks. Transaction cost estimates for execution simulation.

Common Pitfalls:
Lookahead bias uses future information in past decisions. Survivorship bias inflates returns
by excluding failures. Overfitting optimizes to historical noise. Data snooping bias from
repeated testing. Selection bias from cherry-picking favorable periods. Transaction cost
neglect overstates profitability.
Validation Process:
Split data into training and testing periods. Walk-forward analysis uses expanding or rolling
windows. Out-of-sample results confirm strategy robustness. Parameter stability across pe-
riods indicates reliability. Monte Carlo simulation quantifies uncertainty. Compare multiple
strategies to benchmark.
 Risk Management

Position Sizing:
Kelly criterion provides theoretical optimal leverage. Fractional Kelly reduces risk of ruin.
Fixed fractional sizing limits per-trade risk. Volatility targeting maintains constant risk
exposure. Position limits cap concentration risk.
µ
Kelly Formula: f ∗ = σ2 where µ is expected return, σ 2 is variance.

Drawdown Management:
Maximum drawdown measures peak-to-trough decline. Drawdown duration affects psy-
chological tolerance. Reduce exposure during drawdowns to limit losses. Risk-off periods
preserve capital for recovery. Dynamic risk allocation adjusts to market conditions.
Factor Risk Control:
Factor exposures create unintended bets. Measure factor loadings using regression. Hedge
unwanted exposures using derivatives or shorts. Sector neutrality removes industry concen-
tration. Market neutrality eliminates directional exposure.
 Performance Measurement

Risk-Adjusted Metrics:
Sharpe ratio divides excess return by standard deviation. Sortino ratio penalizes only down-
side deviation. Calmar ratio uses maximum drawdown in denominator. Information ratio
measures excess return per tracking error. Each metric captures different risk dimensions.
√
Sharpe Ratio: Sharpe = R̄−R σR
f
Annualize by multiplying by 252 for daily returns.
Factor Attribution:
Decompose returns into factor contributions. Factor exposures explain performance. Alpha
represents excess return after factor adjustment. Attribution identifies return sources. Iso-
lates skill from systematic factor bets.
PK
Attribution: Rp = α + k=1 βk Fk + ϵ where α is residual performance.

Transaction Cost Analysis:

Measure implementation shortfall from decision to execution. Compare execution price to
arrival price. Analyze slippage by order characteristics. Assess market impact of trades.
Optimize execution to minimize costs.
 Learning Path: Foundations

Mathematics Prerequisites:
Linear algebra for portfolio optimization. Calculus for continuous-time models. Probability
theory and distributions. Statistical inference and hypothesis testing. Time series analysis
for return modeling. Stochastic processes for price dynamics.

Statistics Fundamentals:
Regression analysis for factor models. Cointegration and unit root tests. Correlation versus
causation understanding. Hypothesis testing for signal validation. Bootstrap and Monte
Carlo methods. Bayesian inference for parameter estimation.
Programming Skills:
Python for research and backtesting. Pandas for data manipulation. NumPy for numerical
computation. Statistical libraries like statsmodels and scipy. Visualization with matplotlib
and seaborn. Version control using git.
 Learning Path: Intermediate

Finance Theory:
Capital asset pricing model fundamentals. Arbitrage pricing theory framework. Efficient
market hypothesis and anomalies. Market microstructure mechanics. Portfolio theory and
mean-variance optimization. Factor models from Fama-French to multi-factor.

Strategy Development:
Study academic papers on factor premiums. Replicate published strategies from research.
Build backtesting framework with proper data handling. Implement transaction cost models.
Test strategies across different market regimes. Document assumptions and limitations
clearly.
Data Management:
Clean and validate historical data. Handle corporate actions properly. Align fundamental
data to availability dates. Build point-in-time databases. Store data efficiently for fast
access. Implement data quality checks.
 Learning Path: Advanced

Machine Learning Applications:

Feature engineering for financial data. Cross-validation techniques preventing overfitting.
Ensemble methods for signal combination. Dimensionality reduction using PCA. Neural
networks for non-linear relationships. Online learning for adaptive strategies.
Portfolio Optimization:
Mean-variance optimization with constraints. Black-Litterman model incorporating views.
Risk parity allocation methods. Robust optimization under uncertainty. Transaction cost
aware rebalancing. Multi-period optimization frameworks.
Mean-Variance: maxw w T µ − λ2 w T Σw subject to constraints on weights and expo-


sures.

Alternative Data:
Sentiment analysis from news and social media. Satellite imagery for economic activity.
Credit card transactions for consumer spending. Web traffic and app downloads. Evaluate
signal quality and decay rates. Combine with traditional factors.
 Practical Implementation

Strategy Selection:
Start with simple well-documented strategies. Mean reversion easier to test than momentum
initially. Single-factor before multi-factor complexity. Paper trade before live capital. Scale
gradually as confidence builds.

Backtesting Best Practices:

Use realistic transaction cost assumptions. Include bid-ask spreads in simulations. Model
slippage based on order size. Account for market impact of trades. Test across multiple
time periods. Validate on out-of-sample data exclusively.
Common Mistakes:
Overfitting parameters to historical data. Ignoring transaction costs in backtests. Using
too short backtesting periods. Failing to test across market regimes. Not accounting for
survivorship bias. Confusing in-sample with out-of-sample.
 Resources and Next Steps

Key Academic Papers:

Fama and French three-factor and five-factor models. Jegadeesh and Titman momentum
studies. Carhart four-factor model. Asness, Moskowitz, Pedersen value and momentum
everywhere. Frazzini and Pedersen betting against beta.

Books to Study:
Quantitative Trading by Ernie Chan. Active Portfolio Management by Grinold and Kahn.
Advances in Financial Machine Learning by Marcos Lopez de Prado. Expected Returns by
Antti Ilmanen. Your Complete Guide to Factor-Based Investing by Berkin and Swedroe.
Practice Steps:
Download historical data from free sources. Implement basic mean reversion strategy. Back-
test with transaction costs included. Build momentum strategy on same universe. Combine
strategies in multi-factor portfolio. Paper trade and track performance. Document learnings
and iterate.
 Follow Johannes Meyer

For More Quant Finance Content!

Join the Quant Enthusiasts Discord Today!