Performance Metrics

Understanding Rolling Win Rate and Expectancy

Learn how rolling metrics detect strategy degradation and edge consistency over time.

6 min readBeginner friendly

What you'll learn

Learn how rolling metrics detect strategy degradation and edge consistency over time.

Rolling metrics show how your strategy's performance changes over time. Instead of a single overall win rate or expectancy, you see these metrics calculated over sliding windows of trades.

In Simple Terms: Rolling metrics answer: "Is my strategy's edge consistent, or does performance drift over time?" They help detect strategy degradation before it's too late.

Why Rolling Metrics Matter

Overall statistics hide critical information:

MetricOverallReality
Win Rate55%Was 65% in first 100 trades, now 45% in last 100
Expectancy$500/tradeStarted at $800, declining to $200

The overall numbers look acceptable, but the strategy is clearly degrading. Rolling metrics catch this.

Rolling Win Rate

Calculated over a 20-trade window (or 25% of total trades, whichever is smaller). Shows your win rate for the most recent N trades at each point.

What to Look For:

PatternWhat It MeansAction
Stable around 50-60%Consistent edgeStrategy is reliable
Declining trendEdge deterioratingMarket may have changed, investigate
Wild swingsSmall sample or inconsistent strategyNeed more data or strategy improvement
Below 50% recentlyCurrently in losing phaseReduce position size until recovery

Example Interpretation:

Start of backtest: Rolling win rate = 62%
Middle of backtest: Rolling win rate = 58%
End of backtest: Rolling win rate = 54%

This strategy is slowly losing its edge. In live trading, it might
decline further to 50% or below.
Red Flag: If rolling win rate shows a clear declining trend across the entire backtest, the strategy is not stationary. It worked better in earlier periods than later ones. This will likely continue in live trading.

Rolling Expectancy

Shows average profit per trade over the rolling window. More important than win rate because it accounts for both win rate and win/loss size.

Key Insights:

  • Positive and stable: Strategy has a consistent edge
  • Positive but declining: Edge is weakening, prepare to stop trading
  • Crosses zero line: Strategy has periods of negative expectancy (losing edge temporarily)
  • Volatile but positive: Edge exists but is inconsistent, size positions conservatively

Example Patterns:

Pattern 1: Consistent Strategy

Rolling expectancy stays between $400-$600 per trade
Fluctuates but no trend
Good strategy, normal variance

Pattern 2: Degrading Strategy

Started at $800 per trade
Gradually declined to $200
Now approaching zero
Stop trading soon, edge is disappearing

Pattern 3: Regime-Dependent Strategy

High expectancy in trending markets
Negative expectancy in ranging markets
Need to add regime filter or only trade certain conditions

Using Both Together

Win rate and expectancy tell different stories:

Win RateExpectancyWhat's Happening
DecliningDecliningStrategy failing completely
StableDecliningStill winning, but winners getting smaller
DecliningStableLosing more often but winners are bigger
StableStableConsistent performance

Window Size Selection

VivaTrades uses a 20-trade window (or 25% of total trades if less than 80 trades):

  • Smaller window (10 trades): More responsive, but noisier
  • Medium window (20 trades): Balanced, catches trends while filtering noise
  • Larger window (50 trades): Smoother, but slower to detect changes
Pro Tip: If you see large swings in rolling metrics with constant crossings above/below average, your strategy might be too sensitive to market noise. Consider increasing timeframes or adding filters.

Setting Live Trading Rules

Use rolling metrics to create systematic rules:

Rule Example 1: Exit Condition

If rolling expectancy (20-trade window) stays negative 
for 30 consecutive trades, stop trading the strategy.

Rule Example 2: Position Sizing

Base position size: $10,000
If rolling win rate > 55%: Trade full size
If rolling win rate 45-55%: Trade 50% size
If rolling win rate < 45%: Trade 25% size or pause

Rule Example 3: Strategy Health Check

Every 50 trades, compare:
- Current rolling expectancy vs. backtest average
- If current < 50% of backtest: Strategy degraded, investigate
- If current < 25% of backtest: Stop trading

Common Patterns to Watch

The Fade Pattern

Rolling metrics start strong and gradually decline throughout the backtest. This strategy is not robust. Either:

  • Overfitted to earlier data
  • Market structure changed
  • Other traders started using similar strategies

The Cycle Pattern

Rolling metrics oscillate predictably. Strategy works in certain market conditions but not others. Consider:

  • Adding a regime filter
  • Only trading when conditions are favorable
  • Pausing during unfavorable regimes

The Noise Pattern

Rolling metrics swing wildly with no pattern. Strategy might be:

  • Trading too frequently (small sample in each window)
  • Too sensitive to noise
  • Lacking a real edge

Real-World Example

Trader reviews a strategy with 200 trades:

  • Overall win rate: 58%
  • Overall expectancy: $450/trade
  • Rolling win rate: Starts at 68%, ends at 48%
  • Rolling expectancy: Starts at $700, ends at $200

Decision: Don't trade this strategy. The overall numbers look good, but both metrics are declining sharply. The strategy had an edge in earlier periods but is losing it. In live trading, performance will likely continue declining.

Limitations

  • Requires sufficient trades: Need 50+ trades minimum for meaningful rolling metrics
  • Window size affects sensitivity: No perfect window size for all strategies
  • Past performance: Even stable rolling metrics don't guarantee future performance
  • Can't predict regime changes: Markets can shift suddenly, invalidating historical patterns

In VivaTrades

Find rolling win rate and rolling expectancy charts in the Risk Analysis tab. VivaTrades automatically calculates the optimal window size based on your trade count and displays both metrics with interpretation guides.

Ready to test this?

Apply what you've learned with real Indian stock data.