Trading strategies·event

Event arbitrage

◆still alpha
Reviewed 4 June 2026. As of 2026: a real edge still exists for those who can run it well.

Take the other side of predictable repricing around a defined event. Cleanest on prediction markets, where the payoff is bounded and the resolution is unambiguous.

The idea

Event arbitrage annotated diagramfigure
Take the other side of predictable repricing around a defined event. Cleanest on prediction markets, where the payoff is bounded and the resolution is unambiguous.

Reference figure. This concept is explained in prose and diagram; the interactive widgets live on the flagship pages it links to under Where this fits.

What is event arbitrage?

Event arbitrage profits from a predictable repricing, one driven by a known forced flow or a quantifiable outcome probability, not by surprise. You take the side that a model says is mispriced relative to the event's likely resolution: buy the target a merger will probably complete on; front-run the buying an index rebalance forces. The edge is in the predictability, and it is not riskless. It is the structural member of the directional event trading family.

Intuition first. Some repricings are foreseeable. When a stock is added to the S&P 500, every index fund tracking it must buy it by the effective date: predictable forced demand. When a company agrees to be acquired at $50, its shares trade just below $50 until the deal closes, and the gap is the market's odds the deal completes. In both, the event's mechanics are known; you position for the known resolution.

The contrast with news trading is the cleanest way to place it: news trading reacts to surprise, where you do not know the content until it breaks. Event arbitrage trades the known: the rebalance is announced; the merger is agreed; the resolution is a probability you can model. It is closer in spirit to statistical arbitrage (a modelled spread), but the spread closes on an event, not on mean reversion. "Arbitrage" is again loose: these are risk arbitrages. The merger can break (the spread gaps down); the rebalance flow can be smaller or already priced (no pop). You earn a risk premium for bearing event-completion risk, not a riskless profit.

The tradable gap is the expected resolution value minus the current price. When your modelled completion odds or forced-flow estimate differ from the market's implied figure, that difference (net of costs and break risk) is the edge.
edge  =  E[ Vresolve ]⏟your model  −  Pnow⏟market  −  costs\text{edge} \;=\; \underbrace{\mathbb{E}[\,V_{\text{resolve}}\,]}_{\text{your model}} \;-\; \underbrace{P_{\text{now}}}_{\text{market}} \;-\; \text{costs}

Index and ETF rebalances: the forced-flow trade

When an index adds or drops a name, every fund tracking it must trade it to stay in line: a large, predictable, price-insensitive flow concentrated at the effective date. Event arbitrageurs position ahead of that forced buying/selling and unwind into it. The edge has shrunk as the trade became famous and index providers added measures to disguise the flow.

The mechanic: an index addition (e.g. S&P 500 reconstitution) forces tracker funds to buy the new constituent by the effective date, regardless of price. That is the textbook price-insensitive demand (Shleifer, 1986, on downward-sloping demand curves for stocks). The predictable pop into the effective date, and the partial reversion after the forced flow clears, is the arb. The trade: accumulate the addition (and short the deletion) on the announcement, then unwind into the forced flow at or around the effective date. ETF rebalances and leveraged-ETF daily rebalancing flows are related cases.

Forced demand is price-insensitive: trackers must buy roughly a fixed fraction of float by the effective date whatever the price. The pop is that demand meeting finite available supply; the partial reversion is what is left once the inelastic buyer is done.
ΔPpop  ≈  λ⋅QforcedADV,Qforced=(index AUM)×wnew\Delta P_{\text{pop}} \;\approx\; \lambda \cdot \frac{Q_{\text{forced}}}{\text{ADV}}, \qquad Q_{\text{forced}} = (\text{index AUM})\times w_{\text{new}}

Why it has decayed: the trade is decades-public and crowded, so the pop is increasingly priced on announcement rather than the effective date, and index providers and funds use volume-weighted execution and longer transition windows to blunt front-running (Petajisto, 2011, on the index premium and its hidden cost). The surviving edge is in less-tracked indices, the exact flow estimation, and the execution to capture it without paying impact back. See market impact and execution. The HFT angle: the fast version is reacting in milliseconds to the index-change announcement (a scheduled-ish event), pure speed, while the slow version is the capital trade into the effective date.

Merger (risk) arbitrage, and its fast end

Merger arbitrage buys the target of an announced acquisition below the offer price and (in a stock deal) shorts the acquirer, capturing the spread if the deal completes. The spread is the market's deal-completion odds. It is mostly a capital-and-analysis game on a months-long horizon, but its fast end, reacting in milliseconds to deal-status news, is HFT.

The setup: Company Y is offered $50/share and trades at, say, $47, a $3 spread. If the deal completes you make $3 (≈6.4%\approx 6.4\%); if it breaks, Y falls back toward its undisturbed price (a much larger loss). The spread compensates for deal-break risk, and the implied completion probability is recoverable from the spread and the downside. It is a risk arbitrage: negatively skewed, like relative value, with many small completions and the occasional large break (a regulatory block, financing fails, a shareholder revolt). Sizing must survive the break, not assume completion.

The price is a probability-weighted blend of the two outcomes. Invert it and the spread tells you the market's implied completion odds; trade only when your analysis puts the true probability meaningfully above what the spread implies.
Pnow=p Poffer+(1−p) Pundisturbed  ⟹  p=Pnow−PundisturbedPoffer−PundisturbedP_{\text{now}} = p\,P_{\text{offer}} + (1-p)\,P_{\text{undisturbed}} \;\Longrightarrow\; p = \frac{P_{\text{now}} - P_{\text{undisturbed}}}{P_{\text{offer}} - P_{\text{undisturbed}}}

The HFT fast end. The slow merger-arb position is held for months and is a fundamental/legal-analysis game (Mitchell & Pulvino, 2001, on the risk and return of risk arbitrage). But reacting to deal news (the announcement itself, regulatory rulings, competing bids, deal-break headlines) is a millisecond reaction trade and squarely HFT. This is where merger arb meets news trading and scheduled-vs-unscheduled events: the deal-status headline reprices the spread instantly, and whoever reads it first and correctly captures the move. In 2026 the slow leg is a structural, capital-intensive strategy run by dedicated arb funds (alive, a premium for risk); the fast leg is a latency-and-interpretation HFT game on deal news. Neither is "free".

Cross-asset event arbitrage and prediction markets

Cross-asset event arb exploits a predictable relationship between how different instruments reprice on the same event: equity versus its options, a future versus cash, a token versus its perp, or a prediction-market contract versus the asset it references. Prediction markets are the purest case: the contract is the event, so every event view is directly tradable and arbitrageable against correlated assets.

Cross-asset linkage: the same event hits linked instruments with predictable relative timing and magnitude. The equity moves, then the options' implied vol; the spot moves, then the perp funding; the news hits the stock, then the index. Event arb trades the lagging leg against the leading one, overlapping with latency arbitrage when the lag is purely speed.

Prediction markets are the cleanest substrate. A Polymarket contract pays $1 if an event happens and $0 otherwise, so its price is a probability. Three consequences follow: every event view is a direct trade; you can arbitrage a prediction-market price against the correlated asset (if a "Fed cuts" contract and rates futures imply different odds, there is a cross-asset trade); and mutually-exclusive contracts must sum to ≤$1\le \$1 (plus fees), a clean Dutch-book arbitrage when they do not.

A Dutch book: if the prices of a complete set of mutually-exclusive outcomes sum to more than $1, the set is overpriced versus the $1 it is guaranteed to pay, so selling the overpriced pair down toward $1 (net of fees) is the riskless-in-principle edge.
∑iπi  >  $1+fees  ⟹  sell the set; arb=(∑iπi)−$1\sum_i \pi_i \;\gt\; \$1 + \text{fees} \;\Longrightarrow\; \text{sell the set; arb} = \Bigl(\textstyle\sum_i \pi_i\Bigr) - \$1

Honest caveats: prediction-market books are thin (impact is large), resolution can be ambiguous or disputed, and the cross-asset hedge is imperfect. But for an independent quant they are the most accessible event-arb arena: open data, no prime broker. The young venue is also where mispricings persist longest. See prediction markets and the wider crypto venues.

Is event arbitrage still profitable in 2026?

In parts. Merger arbitrage's slow leg is a structural risk-premium strategy (alive, capital-intensive); its fast deal-news leg is competitive HFT. Index/ETF rebalance arb is crowded and decayed in major indices but survives in less-tracked ones and in execution skill. Prediction-market and cross-asset event arb are the youngest, most accessible edges: thin books, but real and persistent mispricings.

Alive, structural: the merger-arb risk premium (slow leg), for capital with the analysis to model deal odds and survive breaks. Competitive HFT: reacting to deal/rebalance announcements in milliseconds, which overlaps news trading. Crowded/decayed: front-running major-index rebalances, where the pop is increasingly priced on announcement. Youngest edge: prediction markets and cross-asset event arb (Dutch-book violations, mispriced contracts versus correlated assets), accessible to an independent quant but constrained by thin liquidity. For the wider verdict see is HFT still profitable in 2026.

Worked example

Two synthetic event-arb setups, as of 2026 (illustrative numbers). Index rebalance. Stock Z is announced for index inclusion, effective in 5 trading days; index funds tracking it must buy about 3%3\% of Z's float by the effective date. On announcement Z pops +4%+4\%; the arb accumulates, expecting the forced effective-date buying to push it further and partially revert after. By the effective date Z is +6%+6\%; the arb unwinds into the forced flow and Z reverts to +3.5%+3.5\% over the following week. Captured ≈\approx the +2%+2\% from announcement-level to effective-date peak, minus the impact of building and unwinding the position, which is the whole skill.

Merger arb. Target Y is offered $50 cash; Y trades $47 (spread $3, ≈6.4%\approx 6.4\%) with about 4 months to expected close. With an undisturbed price ≈$38\approx \$38 (a $9 downside), solving p⋅50+(1−p)⋅38=47p\cdot 50 + (1-p)\cdot 38 = 47 gives p≈0.69p \approx 0.69. If your analysis says completion is more like 85%85\%, the spread is too wide and you buy Y. Fast-end overlay: a regulatory-approval headline at 10:02:03 reprices the spread instantly; the HFT reaction (buy on "approved", sell on "blocked") is a millisecond news trade layered on top of the slow position.

Prediction-market Dutch book: two mutually-exclusive Polymarket contracts ("Candidate A wins" at $0.55, "Candidate A loses" at $0.47) sum to $1.02. The tradable arb is to sell the overpriced pair down toward $1.00 (net of fees), a clean event-arb edge that exists only because the venue is young and thin.
$0.55+$0.47=$1.02  >  $1.00  ⟹  sell the pair, arb≈$0.02−fees\$0.55 + \$0.47 = \$1.02 \;\gt\; \$1.00 \;\Longrightarrow\; \text{sell the pair, arb} \approx \$0.02 - \text{fees}

The live event/duration toy for this family (IX-DURATION) lives on irregular time; this page is diagram-only. Numbers are synthetic and illustrative. Real spreads, flows, completion odds and fees must be measured per deal/index/venue and dated. Educational only, not investment advice.

Where this fits

↑ Up · building block of Directional event trading ↑ Up · building block of Trading strategies ↔ Across · composes with News trading ↔ Across · composes with Latency arbitrage → Apply · makes money in Prediction markets (Polymarket) → Apply · makes money in Equities & futures ⤓ Build / Buy · tool needed Datasets & tools