Gap Trading

Gap trading is a family of short-term strategies built around a single observable event: a market opens at a price meaningfully different from where it last traded. That discontinuity — the gap — is not random noise. It is the first public attempt to reprice everything that happened while the book was closed or thin: earnings and guidance, macro releases, overnight moves in correlated markets, or simply an imbalance of orders accumulated outside regular trading hours.

Because the concept depends on a session break, it lives naturally in instruments whose trading is discontinuous: cash equities and ETFs, index and commodity futures when you measure the regular-hours session separately from the overnight one, and FX or crypto around weekends and low-liquidity windows. In a genuinely continuous market, gaps are rare and usually reflect a liquidity vacuum rather than an auction.

## Why a gap can carry information

At the open, resting interest that could not be executed overnight meets liquidity providers who must decide how much of that imbalance to absorb, and at what price. Two competing readings of the same event drive almost every implementation. The mean-reversion reading treats a gap as an overreaction: participants who react to headlines rather than to price push the opening print past fair value, and the market drifts back toward the prior close as the imbalance clears. The continuation reading treats a gap as legitimate repricing: new information has changed what the asset is worth, the open is the start of the move rather than its end, and fading it means standing in front of a trend.

Neither reading is universally right. Which one dominates tends to depend on the size of the gap relative to the instrument's normal range, its direction relative to the prevailing trend, the volume behind it, and whether there is an identifiable catalyst. That conditionality is precisely what a gap strategy encodes.

## Main variants

**Gap fade (gap fill).** Trade against the gap, targeting a partial or full return toward the prior session's close. Usually applied to smaller, catalyst-free gaps, and often restricted to one direction — fading gaps up is structurally different from fading gaps down, since the risk profile and the borrow requirements are not symmetric.

**Gap and go (continuation).** Trade in the direction of the gap, typically after some confirmation such as a break of the opening range or of the pre-market extreme. Favoured for larger gaps with a clear catalyst and expanded volume.

**Classification-based approaches.** Rather than picking a side in advance, some implementations first label the gap — common, breakaway, runaway or exhaustion, in classical technical-analysis terms — and then apply the fade or continuation playbook that matches the label.

**Session-structure variants.** Weekend gaps, overnight futures gaps measured against the regular-hours close, and gaps measured against the prior high or low rather than the prior close are all distinct events even when they share the same name.

## What typically differentiates implementations

Most of the disagreement between two gap strategies is not philosophical, it is definitional:

- **How the gap is measured**: prior close versus prior high or low; regular-hours close versus settlement; adjusted versus unadjusted prices. - **Minimum size threshold**: a fixed percentage, an absolute amount, or a volatility-normalised measure such as a multiple of ATR — the last of which travels much better across instruments. - **Entry timing**: at the opening print, after the first one-, five- or fifteen-minute candle closes, on a break of the opening range, or on a retracement to a reference level. - **Targets and stops**: full fill, partial fill, VWAP, a fixed R multiple, the pre-market extreme, or a time-based exit. - **Filters**: relative volume, gap direction versus longer-term trend, exclusion of earnings or halted names, liquidity and float requirements, day-of-week or seasonality conditions. - **Universe and ranking**: when a scan returns twenty qualifying gaps, the tie-breaking rule is itself part of the strategy. - **Flat-by-close rules**: whether positions can be held overnight — which, for a strategy about overnight risk, is not a detail.

## Common mistakes

The most frequent error is treating gap fill as a property of markets rather than a hypothesis to test on your own data and universe. Related traps: assuming the fade side is low-risk when the loss on a gap that keeps running is open-ended; mixing catalyst-driven and catalyst-free gaps into one sample so the average hides two opposite behaviours; and backtesting on price series that are not split- and dividend-adjusted, which manufactures gaps that never existed.

On the execution side, filling at the exact opening price you used to define the gap is optimistic — that price is often unobtainable, and spreads at the open are at their widest. Futures backtests need to handle contract rollover, or every roll shows up as a tradeable gap. Equity short implementations need to account for borrow availability. And because opening volatility is elevated, constant position sizing across gaps of very different magnitudes quietly changes the risk of the strategy from trade to trade.

## How to evaluate and backtest versions of it

Start by writing the gap definition down unambiguously, then hold the universe fixed and point-in-time so today's index membership does not leak into a test of five years ago. Because gaps are events rather than bars, sample size should be counted in qualifying events, and a strategy that only trades large gaps in liquid names may produce far fewer observations than the calendar length suggests.

Bucket the results before drawing conclusions: by gap size, by direction, by catalyst presence, and by volatility regime. A single aggregate curve can easily average a genuine continuation effect against a genuine reversion effect into nothing. Model costs explicitly — spread at the open, slippage, commissions, and financing or borrow costs where relevant — and check how much of the result survives them. Then look past the equity curve at the distribution: tail losses, maximum adverse excursion, time in trade, and how clustered the trades are, since gap days often arrive in bunches around the same macro events.

Finally, separate parameter sensitivity from parameter luck. If the strategy only works at one specific gap threshold and collapses at the neighbouring values, that is a fitting artifact rather than an edge. Test out-of-sample across time, and where possible across a second instrument or market, before concluding anything.

## Decoded versions

This page links to two decoded implementations of the concept: *Gap Up Short Trading Strategy* (Chart Fanatics) and *GAP Trading Strategy* (El psicólogo del trading). Each strategy page documents its own rules, filters and session context — use them to see how two authors resolve the design questions above in different ways.

Strategies in this concept (24)

Frequently asked questions

What exactly counts as a gap?

A gap is a discontinuity between one session's reference price and the next session's opening price — most often the prior close versus today's open, but some implementations measure against the prior session's high or low instead, which produces a smaller set of stricter events. The definition matters more than it looks: prior close versus settlement price, regular hours versus extended hours, and adjusted versus unadjusted data can each change which days qualify. Fix the definition before comparing two strategies.

Do gaps always get filled?

No, and the fill rates quoted in trading folklore are not portable — they depend on the instrument, the session definition, the gap size threshold, the period studied and how the fill is measured (intraday touch, close, or within N days). Treat gap fill as a hypothesis to measure on your own universe and data, bucketed by gap size and by whether there was a catalyst, rather than as a market property you can assume.

Which markets and instruments suit gap trading?

Anything with a real session break and enough liquidity at the open: cash equities and ETFs, index and commodity futures when you treat regular hours separately from the overnight session, and FX around the weekend. Continuously traded markets such as crypto gap rarely, and when they do it usually signals a liquidity vacuum rather than an information event — a different phenomenon that needs different handling.

How do I decide between fading a gap and trading its continuation?

That decision is the strategy. In practice it is made with conditioning variables rather than intuition: gap size relative to the instrument's normal range, direction relative to the prevailing trend, relative volume at the open, and whether an identifiable catalyst such as earnings or a macro release is present. A robust implementation states these conditions explicitly and can be tested bucket by bucket, so you can see whether the split is doing any real work.

What data do I need to backtest a gap strategy properly?

At minimum, split- and dividend-adjusted daily data so you do not invent gaps that never traded, plus intraday data at the resolution your entry rule requires — a rule based on the first five-minute candle cannot be validated on daily bars. For equities you also want point-in-time universe membership to avoid survivorship bias, and for futures a documented rollover method so contract changes are not counted as trades. Pre-market data is needed if your entries or stops reference pre-market extremes.

How is gap trading different from a general breakout strategy?

A breakout strategy triggers on price clearing a level during continuous trading, with the order book intact on both sides. A gap trades a discontinuity: there is no price history between the prior reference and the open, so there are no intermediate fills, no observed volume in that range, and no opportunity to exit inside it. That structural absence is what makes the opening print both the opportunity and the main execution risk in this family of strategies.

All strategy concepts · Explore all strategies