NursoftAI
schoolLearning Center
Sign InStart Trading Free

Learn

rocket_launchGetting StartedschoolTrading BasicspsychologyStrategieshistoryBacktestingsmart_toyPaper TradingshieldRisk Managementmenu_bookGlossary
Ready to Trade?

Put your knowledge into practice with $100K virtual capital.

Create Free Account
Learnchevron_rightDivergence Tracking
compare

Backtest vs Live Divergence

PRO

Track how your live trading compares to backtest predictions

schedule10 min readAdvanced
trending_up

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."

lightbulb

Real World Example

Your backtest showed the strategy would make 15% annually with a 52% win rate. After running it live for 3 months, you have 8% returns and a 45% win rate. That's divergence - and you need to understand why it's happening.

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.

speed

Execution Delay

Medium

Backtests assume instant execution. Real orders take time to fill.

attach_money

Slippage Differences

High

Real slippage often exceeds backtest estimates, especially in volatile markets.

block

Missed Trades

High

Orders may not fill at expected prices, or markets may be closed.

swap_vert

Market Microstructure

Medium

Order book dynamics, partial fills, and queue priority affect real trades.

trending_down

Regime Change

Very High

Market 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.

85-100
Healthy
Strategy performing as expected
70-84
Warning
Monitor closely, some divergence
<70
Critical
Review strategy immediately
calculate

How It's Calculated

The divergence score combines multiple factors:
  • 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:

MetricBacktestLiveDivergence
Expected Price$150.25$150.42+0.11%
Quantity1001000%
Slippage0.10%0.18%+80%
Signal → Execution Time0ms340msN/A

Alert Levels

The system automatically alerts you when divergence becomes concerning:

warning

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
error

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
tips_and_updates

When to Act

Warning: Investigate but don't panic. Could be temporary market conditions.
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:

tune

Refine Backtests

Use real slippage data to make future backtests more realistic

insights

Identify Patterns

Find when divergence is worst (market open? volatile days?)

speed

Optimize Execution

Adjust timing or order types to reduce slippage

verified

Validate Strategies

Confirm which strategies actually work in live markets

Professional Best Practices

check_circle

Track from Day One

Start measuring divergence as soon as you deploy a strategy live

check_circle

Set Thresholds

Define acceptable divergence levels before going live - not after

check_circle

Regular Reviews

Check divergence weekly, not just when something seems wrong

check_circle

Document Changes

When you update a strategy, note expected divergence impact

check_circle

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 Experiments
arrow_backSlippage ModelingNext: Paper Tradingarrow_forward

Ready to Apply What You've Learned?

Start paper trading with real market data and test your strategies risk-free.

Get Started FreeExplore Strategies