# Recency bias

Recency bias is the tendency to give recent events more weight than older but relevant evidence when judging a stock or making a trading decision.

## What it means

A sharp move yesterday can feel more informative than a longer history of quieter trading. Recency bias occurs when that feeling leads someone to overweight the latest price change, headline, or result relative to other relevant evidence.

The problem is not paying attention to new information. A recent event can materially change a stock's outlook or trading conditions. The bias lies in giving it more influence *because it is recent*, without checking how relevant, reliable, or representative it is. Recency bias can affect expectations about returns, estimates of risk, and the urgency of a proposed trade.

## How it is calculated

Recency bias has no standard formula or universal score. One way to illustrate it is to compare the weight assigned to recent information with the weight it would receive under a stated, consistent method:

`Recency overweight = actual weight on recent observations − reference weight on recent observations`

- **Actual weight** is the share of influence recent observations have in the person's assessment, expressed as a percentage or a number from 0 to 1.
- **Reference weight** is the share those observations would receive under a comparison method chosen in advance, using the same units.
- **Recency overweight** is measured in percentage points if both weights are percentages. A positive value shows extra emphasis in this comparison; it does not prove that the judgment is wrong.

This is a diagnostic illustration, not a validated measure of someone's psychology. A reference method must fit the question: equal weighting is simple, but older observations may reasonably matter less after a genuine change in conditions.

## Worked example

Suppose an investor reviews a stock's four most recent monthly returns: +2%, +2%, +2%, and +18%. These are hypothetical round numbers, not a return forecast.

1. **Set a reference.** Under equal weighting, each month receives 25%. The simple average is `(2% + 2% + 2% + 18%) / 4 = 6%`.
2. **Record the investor's weighting.** Suppose the investor gives the latest month 70% of the assessment and splits the remaining 30% equally among the other three months. The weighted average is `(0.10 × 2%) + (0.10 × 2%) + (0.10 × 2%) + (0.70 × 18%) = 13.2%`.
3. **Compare the weights.** The latest month receives `70% − 25% = 45 percentage points` more weight than in the equal-weight reference.

The gap shows how weighting changes the summary, not which weighting is appropriate. The investor would need to examine why the latest month differed before treating it as representative.

## Why it matters before you trade

Overweighting recent gains can make a stock seem less risky or more compelling than the broader evidence supports. Overweighting a recent loss can have the opposite effect. Either response can affect position size, execution conditions, or the rationale for including a stock in a basket.

A pre-trade check can separate three questions: What happened recently? Has anything changed that makes it relevant going forward? How does the proposed trade look under a consistent assessment of liquidity, volatility, costs, and concentration? These checks cannot establish what a stock will do next, but they can make an unexamined reaction easier to spot.

## How PreTrAIde uses it

PreTrAIde does not claim to calculate or diagnose recency bias. Its pre-trade analysis assesses a proposed stock basket before an order is placed, using market data, recent news, and analyst ratings. It produces quantitative liquidity, volatility, and execution scores, then provides a GO, NO-GO, or MODIFY verdict with reasoning for the basket and each line.

Reviewing that reasoning alongside the evidence for a recent move may help a user distinguish a new fact from an assumption based mainly on its timing. The verdict is decision support, not a recommendation; its success probability is a model estimate, not a return forecast or guarantee.

## Common mistakes

- **Treating every response to fresh news as biased.** New information can justify a changed assessment.
- **Assuming a short run will continue.** Recent returns alone do not establish a repeatable pattern.
- **Using a longer history mechanically.** Older data may be less relevant after a material change.
- **Mistaking a diagnostic comparison for proof.** Different weights identify a question to investigate, not the correct trade or outcome.

---
Source: https://pretraide.com/glossary/recency-bias