Recency Bias And Recent Trade Memory

The symptoms, including the flattering one

• Abandoning a tested approach after two or three losses, which is the version everyone recognises.
• Becoming attached to a method after a short good run, which nobody flags as a problem because it feels like confidence rather than bias. It is the same error with the sign reversed.
• Moving stops based on the last outcome. Widening after a stop-out and tightening after a winner, rather than placing them where the structure says.
• Calling an instrument hot or cold from its most recent few moves, and then trading it with more or less size than your rule specifies.

The stop-placement one is the most directly expensive, because it corrupts the arithmetic of every subsequent trade. A stop derived from a measure such as average true range is computed rather than remembered, which is precisely why it resists this bias.

Reading a bad run properly

Dismissing every losing stretch as variance is its own mistake, so the useful skill is telling the two apart. Variance looks like your normal trading with a worse result. Something breaking looks different, and it shows up in the inputs rather than the outcomes.

• Did your behaviour change? Compare execution grades and average size across the bad run against the block before it. Drift here is the most common explanation and the easiest to fix.
• Did the conditions change? A range-bound month, a persistent shift in volatility or wider spreads at your usual hours will all degrade a method without anything being wrong with it.
• Are the losses the size they should be? Losses that keep arriving bigger than the plan allows are not variance. They are an execution problem that a strategy change will not touch.

What to do instead of reacting

Review in blocks of 30 to 50 trades rather than continuously, and keep a rolling figure visible so the recent run is always seen against a longer number. A single line in the journal, updated weekly, showing the last fifty trades in units of risk, does more to defuse recency bias than any amount of telling yourself to stay calm, because it puts the context physically next to the feeling.

This does not mean ignoring recent losses. It means weighting them against the full history of the approach rather than treating them as the newest and therefore truest data. The risk is real, most new traders lose money, and a method that genuinely does not work does need replacing. The distinction is that the replacement decision is made from a block of evidence, at a scheduled review, and not on a Thursday after the third loss. The sizing rules in risk management exist largely to make sure you are still there when the block is complete.

This week, write your last fifty trades as a single net figure at the top of your journal, and update it every Friday. Every bad run then gets read against it rather than on its own.