Recency Bias: Why Last Year's Winners Usually Lose Next Year

"Gold crushed it in 2024. Time to load up on gold."

"Tech stocks won 2023. Here's the ETF to buy."

"Bitcoin pumped in 2021. Mortgage the house."

Every year, financial media crowns last year's winners as next year's picks. Every year, most of those picks disappoint.

This is recency bias. The trap of extrapolating short trends into forever.

What It Is

Recency bias: giving too much weight to recent events when forecasting the future.

Your brain is optimized for this. The last meal you ate informs what you expect next meal to taste like. Last week's weather informs your outfit today.

In trading, this bias is expensive. The market doesn't remember last year. It remembers nothing. Yet we keep betting as if the recent past projects forward.

The Data

Academic study after academic study shows: the best-performing sector LAST year is more likely than average to UNDERPERFORM the next year.

This is called mean reversion. It's one of the few reliable patterns in markets.

Why does it happen?

Prices overshoot. Enthusiasm pushes winners above fair value.
Capital rotates. Money chases winners until they're crowded.
Competition arrives. High returns attract imitators, who erode the edge.
Reversion pressure. Statistical mean-reversion is a baseline force in any volatile asset.

Over 50+ years of sector data in the US, the previous year's top sector beats SPY only about 40 percent of the time. Worse than a coin flip.

Where People Get Confused

Recency bias feels like pattern recognition. "Gold worked last year because inflation. Inflation is still an issue. Gold will work again."

The problem: most of the drivers that powered last year's winner are already priced in. The next move requires a NEW catalyst. Which you don't know.

Meanwhile, the drivers of last year's loser may have bottomed. The next move up could be huge.

Mean reversion says: bet slightly against recent winners, slightly for recent losers. In practice, this is called "value investing" and also "contrarian investing." Both have long academic literature.

The Momentum Confusion

Wait. Didn't I publish a momentum strategy? Buy last year's winners?

Yes. And this seems to contradict recency bias / mean reversion.

Here's the nuance.

Momentum works at the 12-month horizon. Last 12 months = next 3-6 months tend to continue. This is the Jegadeesh-Titman effect from 1993.

Mean reversion works at the 3-5 year horizon. Last 3-5 years = next 3-5 years tend to reverse. This is the DeBondt-Thaler effect from 1985.

They coexist. Short-term continuation, long-term reversion. The academic term is "momentum followed by reversal."

The momentum strategy exploits the 12-month continuation. Mean reversion strategies exploit the 3-5 year reversal. Both work at different horizons.

Recency bias goes wrong when you apply short-term patterns to long-term decisions, or vice versa. Buying last year's winner for the next decade is recency bias. Buying last year's winner for the next 3 months is momentum.

How It Burned Me

In early 2022, I nearly doubled my crypto allocation based on the 2020-2021 bull run. "Bitcoin has been up every year recently."

Then 2022 happened. Bitcoin -77 percent.

My recency-biased bet lost about 40 percent of its value. Would have been worse without a cycle filter I had in place.

The bet wasn't crazy. It was just too big, because my recent memory made the risk feel smaller than it was.

The Sector Example

Look at US sector winners and losers:

2019: Tech (+50%). 2020: Tech still strong (+43%). Two years of continuation.
2020: Tech crushed (+43%). 2021: Real Estate (+46%), Energy (+54%). Rotation to other sectors.
2022: Energy (+65%). 2023: Communication Services (+56%), Tech (+58%). Energy flat.
2023: Tech & Comm Svcs. 2024: Tech continued strong (+30+%). Some persistence.

It's not clean. Sometimes persistence. Sometimes rotation. Any "obvious" pattern doesn't hold.

The way to navigate: systematic signals instead of narrative. Momentum at a defined horizon. Not "feeling" about "what's been hot."

The Retail Trader's Trap

Retail investors tend to pour money into funds that did well last year. Vanguard and Fidelity publish net flow data each quarter.

The pattern: big inflows to the fund that was top-ranked last year, followed by disappointing returns, followed by withdrawals after another year.

You buy high, sell low, on the same fund, on a 2-year cycle.

The academic estimate of the "behavior gap": investors underperform the funds they invest in by about 1-2 percent per year, due to this timing pattern. Over 30 years, that's a third of your potential wealth.

The Fix

1. Beware of "this year's winner" narratives. Most commentary in December-January is recency bias in action.

2. Allocate systematically, not emotionally. If you believe in momentum, run the strategy with discipline. Don't pick today's "hot" sector by gut.

3. Use 5-10 year rolling returns for decisions. Not last year. Not last quarter.

4. Rebalance against your winners. If one position grew to 40 percent of your portfolio because it mooned, rebalance back to target. Boring, counterintuitive, correct.

5. Track your prediction accuracy. Humbly.

What This Means for Sandra

Sandra's universe is 15 stocks that did well in the last decade. That's recency bias at the selection level.

A portfolio built on 2015 selections would include very different names. Some would have been excellent picks. Some would have been disasters.

Her current portfolio is protected from the worst recency traps because her selection process (3-pillar fundamental + moat + management) filters out sentiment-only winners.

But she still can't know: which of her 15 will be 2035's winners, which will be 2035's has-beens? The answer is unknowable. The mix is unknowable. Her best protection is position sizing, diversification, and not betting the house.

-> Previous: Regime Bias -> Next: Data-Snooping -> Back to pillar