How to Read the CME Gap

For eight years, one of the most quoted lines in crypto was "CME gaps always fill." You'd see it every Monday. Someone posts a CME chart with an empty space from the weekend, draws a box around it, and says price will come back to close it.

It's easy to test whether gaps fill. It's harder to test whether that means anything. So we took every CME weekend from December 2017 to the last one in May 2026, 427 in all, and checked two things. How often did bitcoin go back to the Friday price? And how often did it reach a level the same distance away in the other direction, where there was no gap at all?

What a CME gap is

CME is the Chicago exchange where big funds trade bitcoin futures. Until this year it kept stock-market hours. Trading stopped at 4pm Chicago time on Friday and started again at 5pm on Sunday.

Bitcoin doesn't stop. It trades all weekend on every crypto exchange. So when CME reopened on Sunday, its price jumped straight to wherever bitcoin had got to. On the CME chart that leaves an empty space between Friday's last candle and Sunday's first. That space is the gap.

Here's a big one. On Friday 28 February 2025, bitcoin was at $84,258 when CME closed. News of a planned US crypto reserve landed over the weekend, and when CME reopened on Sunday bitcoin was at $94,247. That's a gap of +11.9%. The rule says price comes back to $84,258. It did, about a day later.

Most gaps are much smaller than that. The median weekend moved bitcoin 1.2%. 186 of the 427 gaps were under 1%, and 43 were 5% or more. The biggest was 13.1%, in May 2019.

Finding 1: gaps did fill

The rule's supporters have the basic fact right. Over half the gaps, 55%, filled within a day of the reopen. 80% filled within a week and 89% within a month. By today, 97% have.

We measured this on Binance's bitcoin price at CME's closing and opening times. CME's own chart tells the same story. On its daily bars, 56% of gaps filled in the first session back, against 55% on our measure.

So if you'd read those numbers alone, you'd believe the rule. Here's the problem.

Finding 2: so did levels with no gap

The Friday price is always close to where bitcoin reopens. The median gap is 1.2%, and bitcoin often moves that much in an hour. Any price that close gets reached a lot. The fair question is whether the gap level gets reached more than a price with no gap behind it.

So for every weekend we made up a second level. It's the same distance from the reopen as the gap, but on the other side. In the example above, bitcoin reopened $9,989 above the Friday price, so the made-up level was $9,989 above the reopen, at $104,236. If gaps really pull price back, the gap level should be reached more often.

It wasn't. The made-up level was reached as often or more at every point we checked. Within a day it was 65% against the gap's 55%. Within a week, 83% against 80%. By today, 99% against 97%.

Then we raced them. For each weekend we checked which level bitcoin reached first. The gap level won 161 times. The made-up level won 231. In the other 35, both were hit in the same hour, so we can't tell. Among the races with a winner, the gap won 41%. A coin would win 50%.

That's the trade the rule implies. Short a gap up at the reopen, aim for the Friday price, and put your stop the same distance away. It lost more often than it won, before fees.

There's one size where the gap did better. For gaps of 3% to 5%, the gap level came first 33 times to 23, and filled within a week 71% of the time against 52%. That's 56 weekends. Either side of it the gap lost: under 1% it lost 61 races to 91, and at 5% or more it lost 14 to 29. One band out of four, with a small sample, isn't a rule you can lean on.

It didn't get better over time, either. Split the weekends at March 2022. Before that, the gap won 90 races and lost 106. After it, it won 71 and lost 125.

Finding 3: the ones that never filled are below

"Always" isn't quite right either. When CME went 24/7 on 29 May 2026, 16 weekend gaps had never filled. Two have filled since. The last big one, at $84,274 from January 2026, filled on 21 September 2026, four months after CME stopped closing.

Look where the open ones sit. 13 of the 14 are below today's price, all the way down to $3,198 from December 2018. They're gaps up that bitcoin never came back down to, because it went from a few thousand dollars to six figures. The one above today's price is a $25 gap at $95,524, from January 2026.

The same drift shows up in the whole sample. Gaps down leave their level above the price, and 99.4% of them have filled. Gaps up leave it below, and 94.8% have. Over eight years, "gaps fill" mostly means "bitcoin came back to prices near where it had been". A rising market reaches the levels above it. It leaves some of the ones below behind.

Finding 4: the gap is gone

On Friday 29 May 2026, CME didn't close. Its crypto futures and options moved to 24/7 trading, with one short maintenance pause early on Saturday. So since then there's been no weekend gap to draw.

We checked the pause too. Across the 18 Saturdays since the switch, bitcoin moved a median 0.09% over that window. The median weekend gap before it was 1.2%. The biggest move across the pause was 1.4%.

You'll still see old gaps posted as targets. The ones left are the 14 in the table, and 13 of them are thousands of dollars below the price.

How to read it now

• Treat a gap level like any nearby price. It'll probably get touched, because bitcoin moves a lot. 55% of gaps filled within a day, and a level with no gap was reached within a day 65% of the time.
• Don't fade a gap because it's a gap. Aiming for the Friday price with a stop the same distance away lost more races than it won: 161 to 231.
• Be careful with old open gaps. 13 of the 14 left are below today's price. They're a record of how far bitcoin has risen, not a forecast.
• Check the date on any CME chart you're shown. New weekend gaps stopped forming when CME went 24/7 in May 2026.
• Watch the other closed markets instead. Stock index futures and bitcoin ETFs still close for the weekend, so their charts still jump on Monday.

Method, and what this cannot tell you

We took every weekend CME's bitcoin futures closed, from December 2017 to the last one in May 2026. For each, we read bitcoin's price on Binance (BTCUSDT) at the hour CME closed on Friday, 4pm Chicago time, and at the hour it reopened on Sunday, 5pm Chicago time. The difference is the gap. It's filled when bitcoin trades back to the Friday price. As a control, we also tracked a made-up level: the same distance from the reopen price, but on the other side. Then we checked how often each was reached within an hour, a day, a week, a month, and by today, and which one came first.

• We measured on Binance's spot price, not CME's futures chart. CME trades at a small premium to spot that shifts over time, so a gap drawn on the CME chart isn't exactly the same size. Against CME's own daily bars the two measures line up closely: the gap sizes have a correlation of 0.90, and the share filled in the first session back matches within a point.
• Only ordinary weekends are counted. Holiday closures that started on a Friday, like Good Friday and some Christmases, are left out, and so is the Christmas 2020 gap people still talk about. Early closes are handled only for the day after Thanksgiving.
• Fills are checked on hourly bars. A level touched and left within the same hour counts as filled. When both levels were touched in the same hour, the race is a tie.
• The made-up level is one control, not the only possible one. It tests whether the gap level gets reached more than a level with no gap behind it. It doesn't test anything about what happens after a fill.
• Binance's hourly record has a few short holes from maintenance, mostly in 2018-2021. No weekend's close or reopen fell in one. No fees are counted anywhere.