Backtest vs Forward Test Comparison

📊 Executive Summary

Overfitting Detected

0.34

Backtest/Forward Correlation

Only 22% of mathematically-validated strategies (5 of 23) proved viable in forward-testing. This low correlation indicates significant curve-fitting in historical backtests.

Strategy Survival Rate

22%

Viable Strategies

5 truly viable (score ≥70), 6 conditionally viable (50-69), 12 eliminated due to negative expectancy or severe degradation.

Portfolio Performance Gap

-20.3%

Return Degradation

Backtest promised 12-18% annual. Forward reality: -8.3% in 3.5 months. Sharpe dropped from 1.2-1.5 to 0.34.

Market Regime

High Vol

Feb 2026 Crypto Crash

BTC -52% from highs, VIX spiked to 35. This stress test exposed overfitting and regime-specific optimization.

📈 Portfolio-Level Performance

Overall Portfolio: Backtest vs Forward Test

Metric	Backtest (Expected)	Forward Test (Actual)	Delta	Status
Total Return (annualized)	12-18%	-8.3% (3.5 mo)	-20.3%	FAILED
Sharpe Ratio	1.2-1.5	0.34	-0.86	FAILED
Max Drawdown	15-20%	31%	+11%	EXCEEDED
Win Rate	54%	46%	-8%	DEGRADED
Profit Factor	1.4	0.89	-0.51	UNPROFITABLE

🎯 Strategy-by-Strategy Comparison

Detailed performance comparison for each strategy that passed mathematical validation. Metrics show backtest projections vs actual forward-test results.

TIER S: Core Strategies (Mathematically Sound)

Funding Rate Arbitrage

Structural Friction

A+ VIABLE

Win Rate

65% → 71%

+6% ✓

Sharpe

1.8 → 1.95

+8% ✓

Expectancy

+0.95R → +1.02R

+7% ✓

Max DD

10% → 8%

-20% ✓

Why It Succeeded

Extreme funding rates during Feb crypto crash created exceptional opportunities
Structural edge from derivatives mechanics, not price prediction
Market-neutral: profits regardless of direction
High backtest/forward correlation: 0.92

Viability Score: 88/100 (Grade A)

Allocation: 15% of portfolio (increased from original plan)

Pairs Trading (Cointegration)

Market Neutral

A- VIABLE

Win Rate

48% → 51%

+3% ✓

Sharpe

0.7 → 0.78

+11% ✓

Expectancy

+0.34R → +0.38R

+12% ✓

Max DD

8% → 7%

-13% ✓

Why It Succeeded

Correlation breakdown during crash created more mean-reversion opportunities
Market-neutral structure protected against directional risk
Works in any regime - volatility increases opportunities
Backtest/forward correlation: 0.85

Viability Score: 79/100 (Grade A-)

Allocation: 12% of portfolio (increased)

Betting Against Beta (BAB)

Risk Premium

A- VIABLE

Win Rate

58% → 61%

+3% ✓

Sharpe

0.9 → 0.94

+4% ✓

Expectancy

+0.45R → +0.51R

+13% ✓

Max DD

12% → 11%

-8% ✓

Why It Succeeded

Low-beta assets outperformed during risk-off rotation
Flight-to-safety benefited defensive positions
Leverage constraint theory holds in volatile regimes
Backtest/forward correlation: 0.78

Viability Score: 77/100 (Grade A-)

Allocation: 13% of portfolio (maintained)

Quality Minus Junk (QMJ)

Quality Factor

B+ VIABLE

Win Rate

56% → 59%

+3% ✓

Sharpe

0.85 → 0.91

+7% ✓

Expectancy

+0.46R → +0.50R

+9% ✓

Max DD

14% → 12%

-14% ✓

Why It Succeeded

Quality stocks (low debt, high profitability) held up during volatility
Defensive characteristics provided stability
Lower turnover reduced transaction cost impact
Backtest/forward correlation: 0.82

Viability Score: 75/100 (Grade B+)

Allocation: 10% of portfolio (maintained)

Flash Crash Reversal

Volatility-Specific

B+ VIABLE

Win Rate

60% → 82%

+22% ✓

Sharpe

1.2 → 2.1

+75% ✓

Expectancy

+0.20R → +1.15R

+475% ✓

Max DD

15% → 10%

-33% ✓

Why It EXCELLED

Literally designed for what happened in February 2026
Captured +45% returns during the 8-day crypto crash
Volatility-specific strategies thrive in tail events
Low correlation to traditional factors (0.45)

Viability Score: 71/100 (Grade B+)

Allocation: 10% of portfolio (new addition)

Note: Performance spike during crash may not be repeatable, but strategy validated for crisis periods.

CONDITIONAL: Maintain with Reduced Allocation

Cross-Exchange Arbitrage

Arbitrage

B+ CONDITIONAL

Win Rate

85% → 82%

-3% ⚠

Sharpe

3.5 → 3.2

-9% ⚠

Expectancy

+0.63R → +0.57R

-10% ⚠

Max DD

2% → 3%

+50% ⚠

Assessment

Maintained profitability but execution risk increased during volatility
Slippage widened, spread volatility increased
Still mathematically sound but capacity limited
Backtest/forward correlation: 0.88

Viability Score: 73/100 (Grade B+)

Allocation: 7% of portfolio (maintained but monitor closely)

Liquidation Cascade Hunter

Crypto Volatility

B CONDITIONAL

Win Rate

60% → 68%

+8% ✓

Sharpe

0.8 → 1.4

+75% ✓

Expectancy

+0.20R → +0.89R

+345% ✓

Max DD

18% → 14%

-22% ✓

Why It Performed Well

Feb crash created massive liquidation cascades
Strategy designed exactly for these events
Captured +38% during the 8-day crash period

Caveats

Low correlation to backtest (0.42) - regime-dependent
Only works during extreme volatility
May have long periods of underperformance

Viability Score: 68/100 (Grade B)

Allocation: 8% of portfolio (opportunistic, monitor regime)

ETF/Institutional Flow

Flow Analysis

B CONDITIONAL

Win Rate

65% → 67%

+2% ✓

Sharpe

1.4 → 1.6

+14% ✓

Expectancy

+0.30R → +0.42R

+40% ✓

Max DD

12% → 10%

-17% ✓

Assessment

Institutional flow signals remained valid during volatility
ETF arbitrage opportunities increased during dislocations
Moderate backtest/forward correlation: 0.65

Viability Score: 65/100 (Grade B)

Allocation: 5% of portfolio (maintained)

❌ ELIMINATED: Catastrophic Failures

VIX Contango Roll

Volatility Risk Premium

F ELIMINATED

Win Rate

70% → 35%

-35% ✗

Sharpe

1.3 → -0.45

-135% ✗

Expectancy

+0.95R → -0.22R

-123% ✗

Max DD

12% → 38%

+217% ✗

Catastrophic Failure

Feb VIX spike from 18 to 35 in one week
Strategy lost -28% in that single week
Classic "picking up pennies in front of a steamroller"
Backtest/forward correlation: 0.15 (essentially random)

Viability Score: 23/100 (Grade F)

Action: ELIMINATE IMMEDIATELY. Never deploy this strategy again.

Residual Momentum

Factor-Based

C- ELIMINATED

Win Rate

55% → 44%

-11% ✗

Sharpe

0.8 → 0.45

-44% ✗

Expectancy

+0.32R → +0.08R

-75% ✗

Max DD

15% → 24%

+60% ✗

Why It Failed

High correlation regime destroyed the residual return signal
Strategy relies on idiosyncratic returns, which vanished during market stress
Essentially became noise during the crash period
Backtest/forward correlation: 0.35

Viability Score: 42/100 (Grade C-)

Action: ELIMINATE. Requires complete re-engineering or regime-specific parameters.

Time-Series Momentum (TSMOM)

Trend Following

C+ ELIMINATED

Win Rate

55% → 48%

-7% ✗

Sharpe

1.2 → 0.71

-41% ✗

Expectancy

+0.65R → +0.29R

-55% ✗

Max DD

15% → 22%

+47% ✗

Why It Degraded

Momentum crashes during Feb crypto collapse caused significant losses
False breakouts and whipsaws increased in high-vol regime
Regime mismatch - optimized for trending markets, got choppy volatility
Backtest/forward correlation: 0.55

Viability Score: 53/100 (Grade C+)

Action: ELIMINATE from core allocation. May be suitable as conditional 2% tactical allocation only.

🔍 Overfitting Analysis

Indicators of Curve-Fitting

The forward-test revealed significant overfitting across the strategy universe. These indicators show why backtests can be misleading:

Backtest/Forward Correlation

0.34

Threshold: >0.7

Correlation of 0.34 indicates strategies ranked similarly in backtest vs forward. Most (61%) had correlation <0.5.

Expectancy Degradation >50%

39%

9 of 23 strategies

Strategies losing more than half their edge in forward-testing. Sign of over-optimization.

Sharpe Degradation >40%

48%

11 of 23 strategies

Risk-adjusted performance collapsed for nearly half the strategies.

Max DD Exceeded >50%

30%

7 of 23 strategies

Drawdowns in forward-test were 50%+ worse than backtest projections.

Common Backtest Illusions Exposed

Over-optimization: Parameters tuned to fit historical data perfectly but failed to generalize
Look-ahead bias: Some strategies inadvertently used future information (e.g., earnings dates, corporate actions)
Transaction cost underestimation: Real slippage during volatility was 3-5x higher than assumed
Regime overfitting: Strategies optimized for low-volatility bull markets failed in high-vol regimes
Correlation assumptions: Historical correlations broke down during stress, invalidating diversification benefits

📋 Forward-Test Methodology

How Forward Testing Was Conducted

Period: November 1, 2025 - February 17, 2026 (108 trading days)

Initial Capital: $1,130,000 (matching planned allocation)

Markets: SPY, QQQ, BTC, ETH, NQ, ES, EURUSD

Data Frequency: 1m, 5m, 15m, 1h, 4h, 1D

Realistic Execution Parameters

Commission: 0.1% (crypto), $0.001/share (equities)
Slippage: 0.05% base + volatility adjustment
Execution: Simulated market orders with 500ms latency

Volatility Adjustments (VIX > 25)

Slippage increased to 0.15%
Spread widening: +50% for crypto, +30% for equities
Execution delay: +200ms during flash events

Black Swan Event Handling (Feb 3-11 Crypto Crash)

BTC dropped $90K → $60K (-33%) in 8 days
Simulated circuit breakers triggered
Daily loss limits enforced (2% max, later reduced to 1.5%)

Risk Management Validation

Circuit breakers triggered 12 times (daily loss limit)
Kill switch response time: 8.2 seconds average
False positives: 2 | Missed triggers: 0
Effectiveness: 85%

📊 Viability Scoring Framework

How Strategies Are Scored

Each strategy receives a viability score (0-100) based on forward-test performance:

Viability Score = (
    0.30 × Backtest/Forward Correlation +
    0.25 × Regime Robustness (1-10) +
    0.25 × Current Market Fit (1-10) +
    0.20 × Expected Future Performance (1-10)
) × 100

Score Categories

TRULY VIABLE (≥70): Deploy with full allocation. Edge persists across regimes.
CONDITIONALLY VIABLE (50-69): Deploy with reduced allocation, monitor closely.
CURVE-FITTED / UNVIABLE (<50): Eliminate or require 6+ months additional forward-testing.

Top 5 Viable Strategies

Rank	Strategy	Score	Grade	B/F Corr	Regime Fit	Allocation
1	Funding Rate Arbitrage	88	A	0.92	8.5/10	15%
2	Pairs Trading	79	A-	0.85	7.5/10	12%
3	Betting Against Beta	77	A-	0.78	8.5/10	13%
4	Quality Minus Junk	75	B+	0.82	7.5/10	10%
5	Flash Crash Reversal	71	B+	0.45	7.0/10	10%

📁 Raw Data & Reports

📊 Backtest Results (JSON)

Complete backtest data for 113 strategies with all metrics

detailed_results.json | v07_results.json

📈 Forward Test Results (JSON)

Forward-test performance data and viability rankings

forward_test_results.json

📋 Quantitative Validation Report

Mathematical analysis of 113 strategies with expectancy calculations

View Report

🔍 Forward Test Validation Report

Detailed forward-test analysis with strategy-by-strategy breakdown

View Report

📄 Executive Summary

Condensed forward-test findings for quick review

View Summary

📊 Strategy Rankings (CSV)

All strategies ranked by overall score with complete metrics

strategy_rankings.csv

💡 Key Findings & Recommendations

What This Comparison Reveals

Only 22% of backtested strategies proved viable - The majority were overfitted to historical data
Backtest/forward correlation of 0.34 - Extremely low, indicating poor generalization
Market-neutral and defensive strategies won - BAB, Pairs Trading, QMJ outperformed
Volatility-specific strategies excelled - Flash Crash Reversal, Liquidation Hunter designed for crashes
Traditional momentum failed - TSMOM, residual momentum, breakout strategies all degraded
VIX selling was catastrophic - -28% in one week, proves tail risk is real

Immediate Actions Required

ELIMINATE: VIX Contango, Residual Momentum, Breakout Scalper, MACD Cross, Technical Pattern Break
REDUCE allocation by 50%: TSMOM, Value-Momentum Combo, all short-term momentum
INCREASE allocation by 50%: Funding Rate Arb, Flash Crash Reversal, Liquidation Hunter, Pairs Trading
Implement revised risk limits: Daily loss 1.5% (from 2%), Max DD 20% (from 25%), Cash reserve 10% (from 5%)
Continue forward-testing: Minimum 6 additional months for conditional strategies

The Bottom Line

"The era of blind backtest worship is over. Forward-test or fail."

Deploy capital ONLY to strategies with viability scores ≥70 that have proven themselves in live markets. The 5 truly viable strategies share common characteristics: market-neutral or defensive, structural edge (not predictive), volatility-agnostic or benefiting, and lower frequency.

⚠️ Disclaimer

This dashboard compares backtested projections against actual forward-test results. Past performance does not guarantee future results. Trading involves significant risk of loss. Only trade with capital you can afford to lose. The strategies shown are for educational and research purposes. Not financial advice.