Backtest vs Live Divergence
PROTrack how your live trading compares to backtest predictions
The #1 Quant Metric: Professional trading firms track divergence religiously. If your live results differ too much from backtests, something is wrong - maybe market conditions changed, or your backtest was flawed. This is how real quants validate their strategies.
What is Divergence?
Divergence measures how much your live trading results differ from what your backtest predicted. It's the gap between "what should have happened" and "what actually happened."
Real World Example
Why Backtests and Live Trading Differ
There are many reasons why live results don't match backtests. Understanding these helps you build more realistic expectations.
Execution Delay
MediumBacktests assume instant execution. Real orders take time to fill.
Slippage Differences
HighReal slippage often exceeds backtest estimates, especially in volatile markets.
Missed Trades
HighOrders may not fill at expected prices, or markets may be closed.
Market Microstructure
MediumOrder book dynamics, partial fills, and queue priority affect real trades.
Regime Change
Very HighMarket conditions today may differ from historical data used in backtests.
Divergence Score
We calculate a Divergence Score from 0-100 that tells you how closely your live trading matches your backtest. Higher is better.
How It's Calculated
- Return Divergence: Difference in overall returns
- Win Rate Divergence: Difference in trade success rate
- Slippage Divergence: Expected vs actual execution costs
- Trade Matching: Percentage of backtest trades that executed live
What We Track
For each trade, we compare the backtest prediction against the live execution:
| Metric | Backtest | Live | Divergence |
|---|---|---|---|
| Expected Price | $150.25 | $150.42 | +0.11% |
| Quantity | 100 | 100 | 0% |
| Slippage | 0.10% | 0.18% | +80% |
| Signal → Execution Time | 0ms | 340ms | N/A |
Alert Levels
The system automatically alerts you when divergence becomes concerning:
Warning Alert
Triggered when:
- Return divergence exceeds 5%
- Win rate diverges by more than 10 percentage points
- Slippage is 50%+ higher than expected
- More than 10% of trades are missed
Critical Alert
Triggered when:
- Return divergence exceeds 15%
- Win rate diverges by more than 20 percentage points
- Slippage is 100%+ higher than expected
- More than 25% of trades are missed
When to Act
Critical: Consider pausing the strategy. Review your backtest assumptions and market conditions before continuing.
Using Divergence Data
Divergence tracking isn't just about monitoring - it helps you improve your strategies:
Refine Backtests
Use real slippage data to make future backtests more realistic
Identify Patterns
Find when divergence is worst (market open? volatile days?)
Optimize Execution
Adjust timing or order types to reduce slippage
Validate Strategies
Confirm which strategies actually work in live markets
Professional Best Practices
Track from Day One
Start measuring divergence as soon as you deploy a strategy live
Set Thresholds
Define acceptable divergence levels before going live - not after
Regular Reviews
Check divergence weekly, not just when something seems wrong
Document Changes
When you update a strategy, note expected divergence impact
Compare Across Strategies
Some strategy types naturally have higher divergence than others
bookmarkKey Takeaways
- check_circleDivergence tracking is how professional quants validate strategies
- check_circleA divergence score above 85 means your strategy is performing as expected
- check_circleCritical alerts mean you should pause and review before continuing
- check_circleUse divergence data to make your backtests more realistic over time
View Your Divergence Dashboard
Track how your live experiments compare to their backtests in real-time.
monitoringOpen ExperimentsReady to Apply What You've Learned?
Start paper trading with real market data and test your strategies risk-free.