Few rules in trading get repeated as often as this one: "price gaps always fill." Hardly a chart tutorial goes by without the claim that price will, sooner or later, return to the level before the gap. It sounds like a dependable law. On the sober, data-driven level it is mostly one thing: a myth with a kernel of truth.

The reason to sort this out right now is concrete. In late May 2026 the CME exchange introduced weekend trading for Bitcoin futures — and in doing so effectively abolished crypto's most famous gap, the CME gap. A good moment to separate two things that often get lumped together: stock gaps, which still exist, and the CME gap, which has just become history.

Market and structural data as of June 2026. Individual price and statistical figures are source- and time-dependent.

How gaps form

A gap forms when trading pauses but information doesn't. Stocks trade only in fixed hours. During the break, earnings come out, news lands, foreign markets and futures keep moving. At the next open, the orders placed beforehand process all of it into a new opening price. If it sits above or below the previous close, the chart shows a gap. As long as exchanges close, these overnight gaps will keep appearing — that doesn't change.

Bitcoin's CME gap worked similarly, but with one decisive difference. Bitcoin itself trades around the clock. CME's Bitcoin futures, by contrast, rested over the weekend. If the Bitcoin spot price moved from Friday evening to Sunday evening, the CME chart showed a gap between Friday's close and Sunday's open. The key point: the actual Bitcoin price never jumped — it ran continuously. The gap formed only on the chart of a single venue that was closed. "Filling" a CME gap was therefore less a mysterious magnetism than the CME chart catching up to what spot had already done.

Do gaps fill? What the evidence shows

This is where folklore parts from evidence — and the picture is more uncomfortable than the rule promises.

On the practitioner side, the answer depends heavily on the size and direction of the gap. Small gaps in broad indices or index ETFs fill often, sometimes within the same trading day. The S&P 500 ETF filled gaps in about 60 percent of cases over a recent six-month window — a slight edge, not a law of nature. Tiny gaps of a fraction of a percent close same-day more than 90 percent of the time. Studies across 25 years of S&P 500 futures show the same pattern: the smaller the gap, the more likely it closes same-day; and down-gaps fill more often than up-gaps, consistent with the market's general upward drift.

For large gaps the picture flips. Very large jumps — beyond two standard deviations — rarely fill same-day, on the order of five percent of cases in some studies. Over longer windows the probability does rise, in one study above half after about 45 trading days — but that's weeks later, not the quick snap-back the rule implies. And news-driven gaps in single stocks — a takeover, a profit warning — frequently never fill, because fair value has shifted permanently.

It gets sharper in academia. A widely cited paper (Plastun and colleagues) on the US stock market finds the opposite of the rule: on a gap day, price tends to keep moving in the direction of the gap — a momentum effect rather than a reliable reversal. A trading strategy built on it was non-random, hinting at market inefficiency; but after transaction costs it did not beat the market return. A more recent study of weekend gaps in the Dow Jones, NASDAQ and DAX confirms the direction: immediate filling in the short window is the exception, and the momentum effect fades quickly and varies by index.

The famous textbook figures — common gaps fill 90 percent of the time, breakaway gaps 35, continuation 45, exhaustion 75 — circulate everywhere but are practitioner heuristics without a clean, traceable source. They shouldn't be sold as evidence.

The CME gap — a chapter that just closed

For the CME gap, the real story isn't the fill rate — it's the end of the phenomenon.

The fill rates themselves were a lesson in disagreement. Depending on source and method, the figures ranged widely.

That range is itself the warning. And in the most-cited figure the catch sits in one word: roughly 77 percent fill "eventually" — and "eventually" does a great deal of heavy lifting. Small gaps closed quickly; in one analysis, gaps under 700 dollars filled more than 90 percent of the time within 30 trading days, while large gaps in strong trends stayed open for a long while.

Since 29 May 2026, the chapter is closed. The CME moved its crypto futures and options to around-the-clock trading; more than 7,200 contracts changed hands on the first weekend alone. Only a short maintenance window remains — sources differ between a two-hour Saturday pause and a one-hour Sunday window — in which a tiny gap could in theory still form. But the core fact stands: new CME weekend gaps can no longer appear. The three gaps still open from earlier in 2026 remain visible on the chart as historical markers, but no new ones join them.

Weekend risk hasn't vanished, though — it now trades on CME's clock. Execution runs around the clock, while back-office settlement stays tied to business days, and right at launch Bitcoin fell below 70,000 dollars and triggered nearly ten billion dollars of long liquidations over a week.

The structural-break lesson

This is where the most important lesson sits — and it has only a little to do with Bitcoin.

A strategy that bets on the CME gap filling would backtest beautifully over 2018 to 2025. The fill rates were high, the logic seemed to work. But its predictive power for the future is now exactly zero — because the cause of the pattern, CME's closed weekends, was abolished. A clean-looking backtest of a regime that no longer exists is worthless, however convincing the curve.

This is a textbook structural break: the condition that produced a pattern disappears, and the pattern goes with it. Anyone still selling "trade the CME gap" as a live strategy in mid- or late 2026 is selling a setup on a structure that produces no new cases. That's not a detail — it's the heart of what methodically honest backtesting warns about. A backtest measures the past. Whether it says anything about the future depends on whether the world that produced the pattern still exists.

Why the myth survives — four traps

So why does "gaps fill" persist so stubbornly? Because almost every statistic about it breaks down at the same places.

The first is the "eventually" tautology. A market that oscillates and doesn't disappear will, given enough time, revisit almost any price level. "Fills eventually" is nearly trivially true for an index or a 24/7 market — and says nothing about a tradeable edge. Only a fixed time window makes the claim testable.

The second is survivorship bias. Single stocks that gapped on permanent news and never returned, or names later delisted, drop out of many datasets — inflating the fill rate.

The third is definition-dependence. Gap size, direction, instrument and horizon shape the result so strongly that without them there is no meaningful "fill rate." That's exactly why the sources contradict each other.

The fourth is the self-fulfilling prophecy, especially for the CME gap. Because everyone sees the same gap and bets on it closing, their orders partly push price there themselves. That's real, but not a fundamental effect — and it can break once too many crowd in.

Above all sits one last truth: even where a pattern is statistically clean, it often eats its own edge after transaction costs. That's exactly what the academic work on stock gaps found.

Frequently asked questions

Do gaps always fill in the end? No. Small gaps in indices fill often; large and news-driven gaps in single stocks frequently never do. Being filled "eventually" over infinite time isn't a tradeable claim — it's nearly a given.

Does the CME gap still exist? Not for new gaps. Since 29 May 2026 the CME trades Bitcoin futures around the clock; new weekend gaps can't form. The few old gaps still open remain visible as markers.

So is Bitcoin's weekend risk over? No, it has only changed location. Instead of showing up as a gap on reopen, it now appears in continuous CME trading. And back-office settlement stays tied to business days.

Does a gap-fill strategy work on stocks? The evidence is sobering. Academic work finds momentum rather than reversal, and after costs an edge rarely remains. It's only testable with a fixed window and realistic costs.

Can I still use the old CME gap backtest? As a historical description, yes; as a forward strategy, no. The regime that produced it is gone — a classic structural break.

What Backtesting Arena contributes

"Gaps fill" is a catchy story — and a good example of how a claim becomes testable only through clean methodology. An honest study doesn't ask "do gaps fill?" (answer: eventually, mostly, which helps no one), but "does a gap of size X fill within Y days — and does a strategy on it survive transaction costs?".

That mindset is built into the platform. If you test a gap strategy on stocks or crypto, you'll see not just a fill rate but the fixed test window, the number of cases, the drawdown, and the result after costs. And for the CME gap, the most important label comes with it: the underlying regime ended in May 2026 — the backtest describes the past and is explicitly not extendable forward.

This is not investment advice. We're not financial advisors — this post frames a common trading topic with sourced numbers and clean methodology.