Trader Recency Bias — When Only the Last Trade Counts

A trader whose emails I followed for a few months wrote to me after three winning trades in a row on the pound: "I think I've finally got this." A week later the message arrived in a completely different tone — after two losses in a row he wanted to scrap a strategy he had tested for half a year, one with a genuine positive edge. Nothing in his system had changed. The only thing that had changed was which three or four results sat fresh in his memory. That is recency bias in its purest form, and it costs retail traders more than most technical mistakes combined — one of the behavioural failures covered in the ForexMechanics trading psychology section.

What recency bias is and where it comes from

Recency bias is the mind's tendency to put too much weight on what happened most recently, at the expense of the entire history before it. Amos Tversky and Daniel Kahneman described the mechanism back in 1973 and called it the availability heuristic: we judge how likely something is by how easily it comes to mind, not by how often it actually occurs. The last trade is the most available memory — it is fresh, emotional, tied to real money — so the mind treats it as the best sample of reality. In truth it is a sample of size one.

In a trader the same mechanism runs in two directions, and both are expensive. After a run of wins the brain flashes "I have this figured out," confidence rises, and position size rises with it. After a run of losses the message becomes "this has stopped working," paralysis sets in, and proven setups get skipped. In both cases the decision rests on the last three or four results, while the real edge of a strategy only shows up across dozens of trades. That is why two losing positions in a row can derail a plan that would be comfortably profitable over a hundred trades.

How recency bias sabotages sticking to a system

The most dangerous variant is abandoning a sound strategy after a short losing streak. Imagine a trader — this is a hypothetical example, meant to illustrate the mechanism — running a system with roughly a 45% win rate and a one-to-two risk-to-reward ratio. That system is profitable, but with a win rate below half, runs of four or five losses in a row are a mathematical norm in it, not a malfunction. A trader who quits after the fifth loss and jumps to "something that works now" trades statistical certainty for chasing the last result. After a few such jumps there is no system left — only a bundle of reactions to the last win and the last loss.

The second variant is the mirror image of the first: over-trusting a setup that has just worked a few times. Three good entries on the same pattern and the mind promotes it to a sure thing, even though three results are still noise, not signal. The third variant is extrapolating the last few candles into the future — "this trend will never end" — which makes a trader add to a position exactly when the move is most stretched. All three share one root: mistaking the freshest sample for the truth about the market.

The most dangerous effect — position size drift

Recency bias hits hardest not in the choice of entry but in risk management. After a good streak the hand quietly adds a zero to the lot size; after a bad one it shrinks the position below the level at which the strategy makes sense. The effect is insidious because it runs counter to the maths: a trader raises risk just before a statistically inevitable losing streak and cuts it just before the return to the mean. The largest positions land on the worst trades and the smallest on the best. Even a system with a genuine edge can end up flat this way — not because it stopped working, but because position size detached from the rule and started following mood.

Picture a concrete number for illustration: a trader holds a standard of 1% risk per trade, but after three wins lifts the fourth position to 3%. If that fourth one happens to be a loss — and the probability is exactly the same as for any other — a single slip wipes out the gain from the three preceding winners. A handful of such episodes in a year are enough to leave the equity curve flat despite a positive expected value across the whole system.

"We are far too willing to reject the belief that much of what we see in life is random." — Daniel Kahneman, Thinking, Fast and Slow, 2011.

How to spot recency bias in yourself

• Position size drifts with recent results — a bigger lot after wins, a smaller one after losses, even though the risk rule has not changed.
• The urge to change strategy after a short losing streak — "this has stopped working" shows up after four or five trades, not after a hundred.
• Sudden trust in a setup that has just hit a few times — you lower the entry bar "because lately it always worked out."
• Adding to a position in a stretched trend — convinced the last few candles promise an endless continuation.
• Your decision changes when you mentally reshuffle recent trades — the cleanest test that a sample, not the statistics, is steering you.

I catch myself at this most often after an unusually good week — that is when the quiet thought appears, "I could take a little more." I do not trust that thought, because I know where it comes from: three fresh wins, not an analysis of a hundred trades in the journal. Awareness of the source does not switch the bias off, but it lets me not act on it.

Concrete defences — separating the base rate from the latest sample

The first defence is judging the system over a large sample, not over the latest streak. Until you have several dozen trades under the same rule, the last five results are noise and should change nothing. Only an expected value calculated over a meaningful sample tells you whether the strategy has an edge — and that number moves slowly and resists single slips. The second defence is a position-sizing rule set in advance and cut off from emotion. If risk per trade is 1%, then it is 1% after three wins and 1% after three losses alike. Position size should come from a formula, not from how you feel about the last result.

The third defence is a long-horizon journal. Logging every trade — setup, size, emotion, outcome — strips the latest streak of its monopoly on attention, because the whole broad history sits in front of you rather than only what you remember. After a few weeks, valuing the process over a single outcome stops being a slogan and becomes the habit of reading your own data. The fourth defence is rules written in advance that survive a drawdown. The decision "I do not change the system below fifty trades," made while calm, is worth more than ten good intentions in the middle of a losing streak, when your sense of your own skill is unsteady anyway. Recency bias also rarely acts alone — when the whole market reacts to the latest event the same way, herd pressure amplifies the temptation to chase the fresh move.

What to do tonight

Before you sit down for the next session, open your journal and calculate the expected value over at least your last thirty trades — one number that is your true base rate and that the most recent result must not be allowed to obscure. Write one sentence on a card: "I risk X% per trade regardless of how the last positions went," and stick it where you place orders. Add a second rule, a threshold — the minimum number of trades below which you may not change or abandon the strategy. These three things will not remove recency bias from your head, because it is a built-in way the mind works. But they will move the decision from the memory of the last few candles to the statistics of the whole system — the only place where a trader's edge actually lives.