An opening gap is the difference between a market's last price in one session and its first price in the next; it is a description of a price change, not a promise that the gap will close. The open is a new price-formation event, and evidence about overnight returns does not turn every gap into a standalone trading edge.
Define the gap before giving it a story
For a market with a scheduled session, an opening gap compares the previous session's close with the next session's opening price. The exact prices matter: an exchange's official opening auction price is not automatically interchangeable with a broker's first tick, a candle open, or a CFD quote.
Nasdaq says its Opening Cross sets the Nasdaq Official Opening Price when a cross occurs; when it does not, a qualifying first trade supplies the official opening price. That is one exchange's documented mechanism, not a universal rule for every venue. See Nasdaq's Opening and Closing Cross FAQ and the governing Nasdaq Equity 4 opening-process rules.
What overnight-return research can—and cannot—say
Research separates returns earned while a market is closed from returns during the regular session. Berkman, Koch, Tuttle and Zhang report that, for stocks attracting retail attention, positive overnight returns were often followed by intraday reversals; their proposed mechanism concerns elevated opening prices and retail trading. Their result is a finding about a defined stock sample and setting, not a rule that all gaps reverse. See the Journal of Financial and Quantitative Analysis study.
A separate line of work asks how overnight and intraday returns relate at the broader market level. Lou, Polk and Skouras describe a “tug of war”: return patterns can persist within one part of the day and offset across another, with different investor clienteles as a possible explanation. The authors' LSE research listing summarizes that paper and a later extension. These studies examine return decomposition and investor behaviour; neither tests the simple chart rule “a gap must fill.”
That distinction is the useful one. An overnight move may reflect information arriving while the main session is closed, trading by different participants, or the mechanics of the opening process. A chart alone cannot identify which explanation applies to a particular instrument or day.
Turn a gap idea into a falsifiable rule
“Fade the gap” is not yet a strategy. Before looking at outcomes, write down a complete rule set:
| Decision | Specify before the test |
|---|---|
| Instrument and venue | Name the traded contract or symbol and the session whose close and open define the gap. |
| Opening price | Choose the official auction, first eligible trade, or a precisely defined broker quote; do not switch between them. |
| Gap measure | Record signed size and a fixed normalization, such as a prior volatility measure, if that is part of the hypothesis. |
| Entry and expiry | State when a trade may begin, what confirms it, and when an untriggered setup expires. |
| Exit and costs | Define the exit, spread, commission, slippage, and treatment of a jump through an order level. |
| Comparison | Compare the rule with a no-trade baseline and a simple direction-neutral alternative on the same observations. |
A hypothetical test might classify each session's gap using only prices available at the open, then ask whether a predeclared reversal condition occurs before a fixed expiry. The classification, entry, and expiry must be set without using later bars. If you tune several thresholds, record all tried variants and reserve untouched data for the final check. A result on one venue or asset class does not automatically transfer to another.
For a related example of turning an opening-session idea into explicit clocks, triggers and fills, see our Opening Range Breakout backtesting explainer. The pattern differs, but the discipline is the same: define the event before measuring its outcome.
Where the method is weak
“Gap” can mean different things across markets with different trading hours. A continuous FX quote, a futures session, an equity auction and a broker CFD do not share one universal close or opening print. Corporate actions and instrument-specific calendars can also make raw price differences misleading unless the data is adjusted and session boundaries are documented.
The cited equity studies do not establish profitability after trading costs for a retail gap strategy, and their results are not a forecast. A reversal in an aggregate sample can coexist with continuation on particular days. The practical question is whether a precisely defined rule survives its own instrument's data, execution assumptions and an unseen period.
How you'd actually test it
Keep a session ledger with the venue, timezone, calendar, close definition and opening-price source. Save the raw observations so another person can reconstruct each gap. Build the entry and exit logic using information available at the time, then charge the same realistic costs to every variant.
Report the number of eligible sessions, triggered trades, skipped setups, net outcome after costs, adverse movement before exit and results by a predeclared gap bucket. Keep the full sample visible; do not remove awkward openings after seeing the trades. Compare in-sample choices with an untouched period and state what the test still cannot establish. That is the standard realbacktesting uses for an inspectable methodology: a reader should be able to reproduce assumptions rather than trust a smooth chart. See the backtest verification methodology.
Takeaway
An opening gap is a boundary between sessions, not a forecast. Test a rule that names the instrument, opening print, trigger, costs and expiry—or leave the chart annotation as a description.