Recency Bias in Trading: Why Last Month's Results Lie to You

Recency bias is the cognitive shortcut that treats recent data as more important than older data. In everyday life, it’s mostly harmless — yesterday’s weather is a reasonable predictor for today’s weather. In trading, it’s devastating. Markets are not weather. A strategy that made 15% last month might have a 5-year average of 3% per month — meaning last month was an outlier, not the new normal. Traders who can’t distinguish outliers from baselines make the worst decisions at the worst times.

The Recency Trap in Action

The pattern is predictable and universal:

Month 1–3: Strategy produces average returns. You’re neutral.

Month 4: Strategy has an exceptional month (+15%). You feel validated. “I knew this strategy was good.”

Month 5: You increase position size because “the strategy is working.” Maybe you reduce stop distance because “the market is favorable.”

Month 6: Strategy has a losing month (−8%). With increased position size, the actual loss is −12% of your account. Panic sets in.

Month 7: You reduce size or abandon the strategy. Meanwhile, the strategy returns to normal performance — but you’re no longer in it, or you’re in it with reduced size that misses the recovery.

You scaled up at the peak and scaled down at the trough. The strategy’s long-term performance was fine — it was your recency-driven allocation that destroyed the returns.

Why Your Brain Does This

From an evolutionary perspective, recency bias makes sense. A predator spotted yesterday is more relevant than one spotted a year ago. Food sources discovered last week are more reliable than ones known from last season. Our brains are wired to prioritize recent information because in nature, environments change and old data becomes less relevant.

Markets are different. The distribution of returns for a given strategy is relatively stable across years (assuming no structural market changes). Last month’s returns are ONE data point from a distribution that spans hundreds of months. Statistically, one data point tells you almost nothing about the distribution — yet your brain treats it as the most important information available.

The Numbers That Matter

When evaluating a strategy, the relevant metrics span years, not weeks:

Recency Bias Across Market Cycles

Recency bias becomes most dangerous at cycle inflection points. During the final months of a bull market, recency tells you that everything goes up — so you increase exposure, add leverage, and ignore risk management. The backtesting data tells a different story: the highest monthly returns in a cycle tend to cluster in the final 10–15% of the move, right before the reversal. By the time recency bias has you fully committed, the trend is exhausted.

The reverse is equally destructive. At the end of a bear market, recency screams “nothing works, everything loses.” Traders abandon valid strategies during the exact period when those strategies are about to produce their best returns. BTC’s recovery from the 2022 bottom caught most retail traders off guard precisely because recency bias had convinced them that the downtrend was the new normal.

A trader who ran a 5-year backtest through that entire cycle would have seen both the euphoria and the despair in the equity curve. They would have known that −60% drawdowns are part of the distribution, not evidence that the strategy is broken. That context — unavailable to recency-biased intuition — is what keeps experienced traders in the game when others quit.

How Backtesting Counteracts Recency

A 5-year backtest provides the distribution your brain needs but can’t construct from memory. When the backtest shows your strategy averaged 3% per month with a standard deviation of 6%, and last month returned +15%, you can see that +15% is a +2 standard deviation event — rare and unlikely to repeat. Without the backtest data, your brain defaults to treating +15% as the new normal.

Similarly, when the strategy has a −8% month, the backtest shows that −8% months occurred 4 times in 5 years and were always followed by recovery. Your brain screams “abandon ship” while the data says “normal variance, stay the course.”

Practical Defenses

Fixed allocation rules: Pre-commit to position sizes that don’t change based on recent performance. “I risk 1% per trade regardless of last month’s results.” This removes the temptation to scale up after wins or down after losses.

Monthly review against baseline: Every month, compare your results to the full backtest statistics. Plot last month’s return on the historical distribution. Is it within 1 standard deviation? Normal. Between 1–2 sigma? Notable but expected a few times per year. Beyond 2 sigma? Genuinely unusual — investigate, but don’t overreact.

Cooling periods: After any exceptional month (positive or negative), wait 30 days before making allocation changes. This forces recency effects to fade before you act on them.