Ray Ball & Philip Brown; Victor Bernard & Jacob Thomas
Post-earnings drift: the market gets the direction right and the speed wrong
Textbook markets absorb news instantly. In 1968, Ball and Brown showed that prices keep moving in the direction of an earnings surprise for weeks after the report is public — and two decades later Bernard and Thomas showed the pattern was underreaction, not risk. It is one of the oldest documented anomalies in finance, which cuts both ways: the evidence is unusually deep, and so is the crowd that has read it.
- Style
- Factor & anomaly
- Approach
- Mechanical
- Difficulty
- Intermediate
- Horizon
- Swing (days to weeks)
- Holding period
- About one quarter
- Time needed
- Concentrated around earnings season
- Markets
- Small and mid-cap stocks · Post-earnings movers
- Source
- Ball & Brown (1968); Bernard & Thomas (1989) — Ray Ball & Philip Brown; Victor Bernard & Jacob Thomas
The rule set
- Trade only on the bar where a new report actually lands
- Require the reported growth to be strongly positive, not merely positive
- Enter immediately after the announcement rather than waiting for a pullback
- Hold through the drift window — roughly one quarter
- Keep a stop: a strong report can still be sold into a falling market
What makes it distinctive
- One of the oldest documented anomalies, and it has survived being published since 1968
- The entry is an event with a known date, not a pattern someone has to see
- The holding period is the strategy — there is no exit signal to wait for, only the window running out
When it works
Mid and small caps with light analyst coverage, where information takes longer to work into the price — the fewer professionals are paid to react in the first hour, the more repricing is left for the following weeks.
When it fails
Large, heavily covered names price the surprise within minutes and there is nothing left to drift. Entering the day after an announcement can also mean buying the gap high, and the drift is smaller than the gap you paid for.
How a decision moves through it
Input
Reported fundamentals and daily prices
Two feeds with different clocks: prices tick daily, reported figures change only when a filing lands. The mismatch between those clocks is not a nuisance here — it is the raw material the whole system is built from.
Measure
Year-over-year EPS growth, laid out as a step series
Reported figures are placed on the price timeline by publication date, so the growth column holds its value for a whole quarter and steps only on the day a new report arrives.
Measure
Event detection: did the number change on this bar?
Today's growth figure minus yesterday's. Because the series is a step, a non-zero difference means exactly one thing — a filing landed today. This is what separates the system from an ordinary growth screen, which would hold the stock all quarter.
Decide
A strong report just landed
Both halves must be true on the same bar: the figure changed today, and the reported year-over-year EPS growth is above 25%. The threshold is the seed's proxy for 'strong surprise' — deliberately well above merely positive.
Act
Buy the surprise, immediately, and hold for 60 bars
No pullback wait — the drift is measured from the post-announcement price, and the pullback you would wait for is, on average, the drift failing. Sixty trading days is roughly one quarter: the window the literature documents, ending around the time the next report resets everything.
Size & protect
A 10% stop under each position
The drift is an average across many events, not a promise about this one. The stop exists because a strong report can still be sold into a falling market, and no earnings figure protects a position from that.
The finding: prices finish reacting weeks after the news is public
Ball and Brown were not trying to find a trading strategy. Their 1968 paper asked whether accounting earnings carried information at all, and found that they did — but with a detail that should not have been there: the stocks with good earnings news kept outperforming for weeks after the announcement, and the bad-news stocks kept underperforming. In an efficient market the reaction should have been complete on day one.
Two decades later Bernard and Thomas pinned down what the drift was. Not compensation for risk, and not a statistical artefact: underreaction. The market behaved as if it did not fully appreciate that earnings surprises repeat — a company that just beat handily tends to beat again next quarter, and prices moved on the next announcement in a way that was partly predictable from the last one.
Read carefully, the anomaly is about speed, not direction. The market's first move is right; it is just incomplete. That framing decides every rule below — you are not betting against the crowd's judgement, you are betting on its slowness.
What counts as a surprise, as this system ships it
The academic studies define surprise against an expectation — earnings versus the analyst consensus, or versus the same quarter last year scaled by its volatility. The shipped system uses a simpler proxy: on the day a new filing lands, year-over-year EPS growth must be above 25%. Strongly positive reported growth stands in for a positive surprise, on the reasoning that growth that strong is rarely fully anticipated in the under-covered names this hunts.
The proxy is honest but it is a proxy, and the difference matters at the edges: a company universally expected to grow 40% that reports 30% has passed this screen while genuinely disappointing the market. The literature's drift follows the surprise, not the growth — which is one reason the shipped rules keep a stop where the papers keep a portfolio.
There is no exit signal, and that is the design
Most systems in this library exit on information: a band is touched, a level breaks, a story changes. This one exits on a clock — sixty trading days, roughly the gap to the next report. The documented anomaly is a drift over approximately one quarter, so the holding period is not a parameter someone tuned; it is the finding itself, transcribed.
That makes patience the system's only ongoing task. Nothing that happens in week three is actionable. The position was right or wrong at the moment of entry, and the window either pays or it does not.
Five ways into this system
- The rules: detecting the filing day, qualifying the surprise, and why you buy immediatelyThree rules do all the work: recognise the day a report lands, demand strongly positive growth on that day, and buy before the market has finished thinking. The rest is a calendar.7 min read
- Sizing PEAD: many small positions, because the edge is an averageThe drift is a statistical statement about hundreds of announcements, not a forecast about the next one. Everything about sizing follows: small, uniform, and spread across as many qualifying events as the account can carry.6 min read
- Where the drift lives: under-covered names, and a calendar instead of a scheduleThe anomaly is not evenly spread across the market. It concentrates where attention is thin — and the same thinness that preserves the drift charges for access through spreads and slippage.6 min read
- When the drift fails: a fading anomaly, the crowding argument, and the gap you paid forPEAD's biggest risk is not any single trade. It is that the anomaly itself has been shrinking in the decades since it was published — and the reasons why are the most instructive content in the whole dossier.8 min read
- Post-earnings drift for beginners: why news can be public and still not priced inThe strategy in one sentence: when a smaller company reports surprisingly strong profits, buy and hold for about three months, because prices tend to finish reacting slowly. This page explains why that ever works.7 min read
The ideas behind it
This system assumes you already know these. Each one is explained from scratch in Investing 101.
Compare with
- 12-1 Momentum FactorBuy the stocks that have risen most from twelve months ago to one month ago, hold while the measure stays positive, and rebuild the list on a fixed schedule.
- CAN SLIM Growth SystemBuy the market's leading growth stocks — accelerating earnings, institutional buying, a fresh high — but only from a proper chart base, and only while the market itself is advancing.
These are documented methods described for study. Nothing here is investment advice, a recommendation, or a claim about future returns — every system on this page has losing periods, and the pages say where.
Reading about a system is not having one.
Plutux is where you write your own rules down, test them against real data, and keep the record your memory would otherwise rewrite. Join the waitlist for early access.