Investment Strategies

Event-driven investment strategies use the same machinery as an academic event study, then try to capture the part of the market's reaction that prices have not yet fully absorbed. The central object is the abnormal return: the slice of a security's return that the chosen expected-return model cannot explain. A tradable strategy exists wherever abnormal returns around an identifiable event are predictable, whether as a positive drift to ride, an overreaction to fade, or a deal spread to harvest. The same event study toolkit that documents these patterns is also the honest referee, because it tells you whether an apparent edge survives risk adjustment, transaction costs, and out-of-sample decay.

This page sets out the abnormal-return patterns that strategies are built on (post-earnings drift, momentum and reversal, and corporate-event signals), states their empirical magnitudes, explains the methodology that distinguishes a real edge from a statistical mirage, and shows how to run the analysis with our calculators. The governing caution comes from Fama (1998): every abnormal return is measured against an assumed model, so an apparent strategy is always a joint test of the event effect and that model.

What the research shows

Three families of return predictability have survived decades of scrutiny well enough to anchor real strategies. None is a free lunch, and the most important recent finding is that publication itself erodes them.

Post-earnings-announcement drift: the most robust tradable signal

After an earnings surprise, prices keep drifting in the direction of the surprise for weeks. Bernard and Thomas (1989), building on Foster, Olsen and Shevlin (1984), sorted firms into deciles on standardized unexpected earnings (SUE) and found that a portfolio long the highest-surprise decile and short the lowest earned a cumulative abnormal return of roughly 18% annualized over the following 60 trading days. Bernard and Thomas (1990) tied the drift to investors underreacting to the autocorrelation in earnings. This is the cleanest event-driven strategy: a datable catalyst, a measurable surprise, and a directional drift to trade.

Momentum and reversal: horizon is the whole game

Return predictability flips sign with horizon. Jegadeesh and Titman (1993) document intermediate-horizon momentum: stocks ranked on their past three-to-twelve-month returns continue in the same direction, with a long-short portfolio earning about 1% per month over the next year. At multi-year horizons the sign reverses, as De Bondt and Thaler (1985) showed for long-term reversal and Lakonishok, Shleifer and Vishny (1994) for contrarian value. The catch is tail risk: Daniel and Moskowitz (2016) show momentum suffers infrequent but severe crashes in market rebounds, which any abnormal-return backtest must stress-test for.

Corporate-event signals: deal spreads and post-event drift

Corporate actions generate their own strategies. Merger arbitrage buys the target and (in stock deals) shorts the acquirer to capture the spread between the offer and the market price: Mitchell and Pulvino (2001) show its payoff resembles a short put on the market, earning a steady premium in calm markets and losing in crashes, with risk-adjusted returns of a few percent per year after realistic costs. Post-event drift also appears after buybacks, where Ikenberry, Lakonishok and Vermaelen (1995) document about 12.1% of abnormal return over the four years following open-market repurchase announcements, and in the opposite direction after equity issuance, where Loughran and Ritter (1995) document long-run underperformance after IPOs and SEOs.

The catch: efficiency, costs, and decay

Three forces stand between a documented abnormal return and a profitable strategy. First, costs: Jensen (1978) framed market efficiency as the absence of profit after transaction costs, and many paper anomalies do not clear realistic spreads, commissions, and price impact. Second, limits to arbitrage (Shleifer and Vishny, 1997): the strategies are strongest exactly where they are hardest to trade, in small, illiquid, hard-to-short names. Third, and most important, decay: McLean and Pontiff (2016) find that returns to a typical anomaly fall by about 58% after the academic paper documenting it is published, as arbitrageurs trade it away. A live strategy must therefore assume the published magnitude is an upper bound.

How to run this kind of event study

The general workflow (estimation window, expected-return model, event window, abnormal returns, significance testing) is covered in our Introduction to Event Study Methodology and the step-by-step Application Blueprint. Building a strategy on top adds four design choices.

• Evaluate in calendar time, not just event time. An event-time cumulative abnormal return shows that a signal exists, but a strategy is a portfolio held over the calendar. Form a calendar-time portfolio of all firms currently inside their post-event window, then regress its monthly excess return on the market, size, value, and momentum factors. A significant intercept (Jensen's alpha) is the tradable edge, and this approach handles the cross-sectional correlation that inflates naive event-time t-statistics.

• Choose the expected-return model deliberately. Because every abnormal return is conditional on a model, test robustness across the expected-return models (market model, CAPM, Fama-French three and five factor, Carhart four factor). An edge that vanishes when you add the momentum or size factor was a risk premium, not alpha.

• Test significance honestly. Event-driven strategies cluster in calendar time (many firms report earnings in the same weeks), which biases standard errors downward, so use the robust statistics on our significance tests page (Boehmer-Musumeci-Poulsen and the Kolari-Pynnonen adjusted tests) rather than the plain cross-sectional t-test.

• Subtract realistic costs and assume decay. Net every backtested return of spreads, commissions, short-borrow fees, and price impact, and discount the historical magnitude for post-publication decay before sizing any position.

Run it with our tools

Our calculators implement this workflow over the short windows where the statistics are well-behaved.

• Abnormal Return Calculator (ARC) is the core tool. Build an event file keyed to the catalyst date (the earnings release, the announcement, the issuance), pick an estimation window and expected-return model, and compute the abnormal returns and cumulative abnormal returns with parametric and non-parametric tests. To prototype a signal, rank the cross-section on the surprise variable (SUE, past return, or deal type) and compare the CAAR trajectories of the top and bottom groups.

• News Analytics (CATA) turns the news and disclosure flow into a tone or surprise score, the raw material for a sentiment-driven or surprise-driven signal.

• Event Date Identifier (EDI) pins down the precise catalyst date, on which every short-window result depends.

• Abnormal Volume Calculator (AVC) adds a volume confirmation: a drift backed by abnormal volume is more durable than one on thin trading.

Related use cases

Investment strategies sit alongside the other investing applications of this methodology and share their design questions. See the closely related Tactical Asset Allocation Signals page, which lifts the same drift and reversal signals to the portfolio and asset-class level; the Earnings Announcements page, whose SUE and drift machinery underpins the strongest strategy here; the Mergers and Acquisitions page behind merger arbitrage; and the broader Comparative Event-Type Analyses. For the full catalogue, return to the Practical Applications overview.