Merger Arbitrage Strategy

The thesis

When two companies announce a definitive merger agreement, the target’s stock typically jumps to a level just below the announced deal price. The remaining gap — the “arbitrage spread” — reflects the market’s aggregate estimate of two things: (a) the probability that the deal closes on the announced terms, and (b) the time-value cost of capital tied up until close. Buy the target after the announcement and hold to deal close, and you collect that spread.

It sounds mechanical, and on average it is. But Mitchell and Pulvino (2001) showed that merger arbitrage’s risk profile is the opposite of what its name suggests: in normal markets the strategy looks riskless and earns ~6% per year over T-bills, but in market crashes it suffers concentrated losses because deals are more likely to break exactly when capital becomes scarce. Their characterization — merger arb is “selling insurance against deal failure” — is the right way to think about it.

Academic basis

Mitchell and Pulvino’s 2001 Journal of Finance paper is the canonical reference. Their sample of 4,750 deals from 1963–1998 produced an annualized excess return of ~4–6% over T-bills with a market beta near zero in normal periods, but a much higher beta in stressed periods — producing a payoff diagram that mirrors a short put position on the market. Baker and Savaşoğlu (2002) decomposed the spread into compensation for capital lockup, regulatory risk, and arbitrageur capital constraints, finding that limited arbitrageur capacity (not just deal-break risk) explains a meaningful portion of the spread.

Subsequent work (Jindra and Walkling 2004; Cornelli and Li 2002) has shown that the spread is wider when arbitrageur capital is depleted, when deal financing involves stock or contingent consideration, and when antitrust scrutiny is elevated. Cash deals in friendly transactions with no regulatory complications tend to close at the highest rates and offer the cleanest arbitrage.

How Alpha Suite implements it

• EDGAR 8-K full-text search — the scanner queries the SEC EDGAR full-text search API for 8-K filings containing M&A trigger language: “agreement and plan of merger”, “definitive merger agreement”, “to be acquired by”, “merger consideration”, “per share in cash”.
• Disqualification keywords — filings mentioning “terminated”, “withdrawn”, or “amended” (without a price increase) are filtered out.
• Deal-price extraction — regex parsing extracts the announced cash consideration per share. Stock-only deals are priced via the acquirer’s current quote; mixed cash-and-stock deals combine both legs. Deals where price cannot be extracted are skipped — we do not fabricate.
• Spread filter — only signals where current price is below deal price qualify (positive spread). Negative spread (price above deal) means the market expects a counter-bid; the arbitrage thesis no longer applies.
• Cash-vs-stock-vs-mixed scoring — cash deals receive +10 points (cleanest payoff), mixed deals +5, stock-for-stock 0. The certainty of the payoff is scored, not just the spread size.
• Regulatory friction haircut — cross-border deals get -5; sectors with elevated antitrust scrutiny (telecom, large healthcare M&A) get -3.
• Annualized-spread tier — spreads above 15% annualized get +5 (compelling carry); spreads above 25% annualized get the same +5 but also -5 (suspicious; market is pricing meaningful break risk).
• Take-profit at deal price; stop-loss at -20% — the take-profit is mechanical (the deal price); the stop-loss matches the typical -20–40% drawdown observed on broken deals (Mitchell-Pulvino).
• Horizon — defaults to 90 days, with cash deals scheduled at 60 days and large strategic / regulatory-sensitive deals at 120–180 days when the filing discloses an expected close timeframe.
• Long-only — the suite does not short the acquirer leg of stock deals; users with shorting access can construct that leg themselves.

When it fires

Signals appear within hours of an 8-K M&A announcement and persist until the deal closes or breaks. The cleanest fires are friendly, all-cash deals with announced expected-close dates, no significant regulatory overhang, and a target market cap above $2B (large enough to clear antitrust thresholds, big enough to attract arbitrageur capital that compresses the spread to a fair level).

Confluence with the activist 13D engine is informative: when a 13D filer accumulates a stake in a target that subsequently announces a deal, the original 13D often was the catalyst. The model surfaces these as confirmation rather than independent signals to avoid double-counting the same event.