--- title: "Bitcoin CME gap: does it always fill? Weekend data since 2019" language: ja type: empirical canonical: "https://whycoinmoves.com/ja/research/bitcoin-cme-gap-fill" published: 2026-10-20 reviewed: 2026-10-20 --- # Bitcoin CME gap: does it always fill? Weekend data since 2019 Test the Bitcoin CME gap on 404 weekends: how often it fills, what happened after the Sunday reopen, and why CME’s move to 24/7 trading in 2026 changes it. WhyCoinMoves Research · 2026-10-20 Observation / evidence scope: 2019-01-06/2026-10-04 ## BTC/USDT hourly prices at CME’s Friday 16:00 and Sunday 17:00 Chicago times; gaps of at least 0.5%; return in the gap’s direction from the Sunday reopen, before costs | Period | Gaps N | Mean 4h move in gap direction (bp) | t-stat | Moves in gap direction (%) | Mean 24h (bp) | Mean 72h (bp) | | --- | --- | --- | --- | --- | --- | --- | | 2019–2021 | 122 | -3.4 | -0.15 | 54.1 | +14.1 | +48.9 | | 2022 – 28 May 2026 | 151 | +52.3 | +4.52 | 58.3 | +82.3 | +41.2 | | From 29 May 2026 (CME 24/7) | 8 | +9.0 | +0.26 | 62.5 | +42.0 | -162.4 | ## What a CME gap is, and why it existed Bitcoin trades every hour of the week, but CME’s regulated Bitcoin futures historically closed on Friday afternoon Chicago time and reopened on Sunday evening. Any weekend move in spot Bitcoin therefore appeared on the CME chart as a jump between Friday’s last price and Sunday’s first: the “CME gap”. A popular claim says the price always returns to fill it. We reconstructed the gap from spot BTC/USDT hourly prices at 16:00 Chicago time on Friday and 17:00 on Sunday for 404 weekends from January 2019 to October 2026. 281 of them, 69.6%, moved at least 0.5%; the median absolute gap was 1.67%. Using spot prices at CME’s clock times captures the weekend move itself; it is not CME’s own futures settlement, which also includes a futures basis. ## Gaps fill often, but so do ordinary price levels From April 2022, where we have hourly highs and lows, until CME’s schedule changed, 106 of 142 gaps, 74.6%, saw the price revisit the Friday level within five days. Smaller gaps (under 1%) filled 87.5% of the time and larger ones 69.6%. That sounds like support for the folklore, but a comparison matters: after a weekday move of the same size (Tuesday to Thursday, 17:00 against the previous day’s 16:00), the price revisited the earlier level within five days 64.0% of the time across 530 cases. Prices that wander up and down will revisit a nearby level often. The weekend gap filled somewhat more often than that baseline, but “always” is wrong: about one gap in four was still open after five days. A gap can also fill only after the price first moves further away, which matters to anyone holding a position against it. ## Betting on the fill at the reopen lost money in this sample Taking a position toward the Friday level at the Sunday reopen, across all 273 gaps before the schedule change, averaged −51.8 basis points over 24 hours (t −2.07) and −44.7 bp over 72 hours (t −1.18), before costs. In other words, the reopen move tended to keep going for a while rather than snap back. The continuation was strongest from 2022 to May 2026: over the first four hours after the reopen, BTC moved in the direction of the gap by an average of 52.3 bp, and did so 58.3% of the time. From 2019 to 2021 the same measure averaged −3.4 bp. One period does not establish a durable pattern, and the yearly averages ranged from −39.6 bp (2021) to +84.5 bp (2023). ## CME’s 24/7 trading changes the question CME began offering cryptocurrency futures trading around the clock from 29 May 2026, apart from brief maintenance windows. Without a weekend close there is no weekend gap on CME’s own chart, so the folklore’s premise largely disappears. Since then, the eight weekends with a 0.5% move at the old clock times showed a four-hour average of +9.0 bp: too few to estimate anything, but no sign of the earlier pattern. Charts and indicators that still draw “CME gaps” after that date are describing the old trading schedule, or a different futures contract, rather than an actual gap in CME’s trading. ## How the numbers were calculated Weekend returns use hourly open prices of Binance BTC/USDT from January 2019, converted to Chicago time so that daylight saving is respected. A gap is the Sunday 17:00 price against the Friday 16:00 price. Directional returns are measured from the Sunday 17:00 price and signed by the gap. Fill checks use OKX BTC/USDT hourly highs and lows, available from April 2022, over the five days after the reopen. Weekends with missing hours at the required times are skipped. Holiday schedules, when CME did not reopen on Sunday, are not removed, and the analysis does not use CME futures prices or include a futures basis. t-statistics treat each weekend as one observation. ## Does the Bitcoin CME gap always fill? No. About three in four gaps from April 2022 to May 2026 saw the Friday price revisited within five days, but ordinary weekday moves of the same size were revisited 64% of the time, and one in four weekend gaps stayed open after five days. ## Are there still CME gaps after CME moved to 24/7 trading? CME’s cryptocurrency futures have traded around the clock since 29 May 2026, so there is no weekend close to create a gap on CME’s chart. Measured at the old clock times, the few weekend moves since then show none of the earlier pattern. ## Limitations Spot prices at CME clock times rather than CME futures prices; one exchange per measurement; hourly data; holidays not removed. The post-May 2026 sample has eight observations. No fees, spreads, funding or slippage are included, and the results are not a trading rule. ## Sources and data access - [CME Group: 24/7 cryptocurrency futures and options trading]() - [CME Group: Bitcoin futures contract specifications]() - [Binance public data and archive formats]() - [OKX market data documentation]() Aggregate data only; raw archives are not redistributed. Source fingerprints identify the files used, not independent verification. [JSON]() ## Related live archives - [BTC]() - [価格急変動]() ## Live prior 30-day hourly baseline — separate from the historical study Only the current watch list is shown, up to 12 assets. Missing or stale estimates remain unavailable. Venue and quote currency are kept separate; beta needs matching BTC data. Hourly statistics are not minute-move probabilities. Cutoff (UTC): 24/10/2026, 15:37 | Asset | Source | Bars / expected | Median \|r\| (%) | P99 \|r\| (%) | BTC Beta | Matched N | Last close (UTC) | Status | | --- | --- | --- | --- | --- | --- | --- | --- | --- | | BTC | OKX · USDT | 720/720 | 0.1409 | 1.0082 | 1.0000 | 719 | 24/10/2026, 15:00 | Ready | | HYPE | OKX · USDT | 720/720 | 0.3477 | 1.8803 | 1.2391 | 719 | 24/10/2026, 15:00 | Ready | | ZEC | OKX · USDT | 720/720 | 0.5432 | 4.0682 | 2.2811 | 719 | 24/10/2026, 15:00 | Ready | | SUI | OKX · USDT | 720/720 | 0.4833 | 3.5596 | 2.1690 | 719 | 24/10/2026, 15:00 | Ready | | ONDO | OKX · USDT | 720/720 | 0.4999 | 3.9817 | 2.2012 | 719 | 24/10/2026, 15:00 | Ready | | SOL | OKX · USDT | 720/720 | 0.2713 | 2.0082 | 1.4853 | 719 | 24/10/2026, 15:00 | Ready | | ADA | OKX · USDT | 720/720 | 0.4336 | 2.9570 | 2.0101 | 719 | 24/10/2026, 15:00 | Ready | | AVAX | OKX · USDT | 720/720 | 0.4243 | 2.9471 | 1.8254 | 719 | 24/10/2026, 15:00 | Ready | | BCH | OKX · USDT | 720/720 | 0.3521 | 2.5564 | 1.5484 | 719 | 24/10/2026, 15:00 | Ready | | BNB | OKX · USDT | 720/720 | 0.1759 | 1.0513 | 0.8846 | 719 | 24/10/2026, 15:00 | Ready | | DOGE | OKX · USDT | 720/720 | 0.3019 | 1.9830 | 1.5542 | 719 | 24/10/2026, 15:00 | Ready | | ETH | OKX · USDT | 720/720 | 0.1896 | 1.4105 | 1.2140 | 719 | 24/10/2026, 15:00 | Ready | [Research]() [Methods and limitations]()