Charlie Quant LabThe Weekend Mirage
From Friday's close to Monday's open, COIN, MSTR and IBIT cannot trade — but Bitcoin never stops. Their Monday gap is a delayed reaction to a move everyone already watched. This page measures how much of the gap the weekend Bitcoin move actually explains, and whether big weekend moves produce gaps that fade by Monday's close.
Updated 08 September 2026 · 20:37 UTCEvery weekend, plotted
One dot per weekend: how far Bitcoin moved while the stock market was dark (horizontal) against the Monday opening gap (vertical). The dashed lines are the per-ticker regressions. Sample: 99 weekends, 2024-09-13 → 2026-09-04.
The regression, per ticker
Monday opening gap regressed on the weekend Bitcoin move. Beta reads as gap-percent per weekend-percent; R² is the share of the gap the weekend explains. Intercept is the average gap when Bitcoin went nowhere.
| Ticker | Weekends | Gap beta | R² | Flat-weekend gap |
|---|---|---|---|---|
| COIN | 99 | 0.89× | 0.73 | +0.1% |
| MSTR | 99 | 1.14× | 0.81 | +0.1% |
| IBITtrading since January 2024 — smallest sample | 99 | 0.99× | 0.99 | -0.0% |
Do big gaps fade?
On weekends when Bitcoin moved 3% or more, did the Monday gap overstate? "Rest of Monday" is the average move from open to close on those days; "faded" is how often the day's direction opposed the gap. This is a description of past gaps, not a strategy.
| Ticker | Weekend condition | Cases | Average gap | Average rest of Monday | Gap faded |
|---|---|---|---|---|---|
| COIN | BTC weekend ≥ +3% | 12 | +4.2% | +0.7% | 25% |
| BTC weekend ≤ −3% | 9 | -4.8% | -0.7% | 33% | |
| MSTR | BTC weekend ≥ +3% | 12 | +5.3% | +1.2% | 33% |
| BTC weekend ≤ −3% | 9 | -6.1% | +1.2% | 56% | |
| IBIT | BTC weekend ≥ +3% | 12 | +4.7% | +0.1% | 42% |
| BTC weekend ≤ −3% | 9 | -5.4% | +0.4% | 44% |
The ten loudest weekends
The largest weekend Bitcoin moves in the sample and the gaps they produced. Holiday-Monday cases open on Tuesday.
| Friday → next session | Weekend BTC | COIN gap | MSTR gap | IBIT gap |
|---|---|---|---|---|
| 2025-02-28 → 2025-03-03 | +11.2% | +7.7% | +14.8% | +10.6% |
| 2025-04-04 → 2025-04-07 | -8.0% | -9.0% | -11.4% | -8.4% |
| 2025-01-31 → 2025-02-03 | -6.9% | -6.3% | -7.4% | -6.5% |
| 2026-01-30 → 2026-02-02 | -6.7% | -2.5% | -6.6% | -6.8% |
| 2024-11-08 → 2024-11-11 | +6.7% | +10.9% | +9.8% | +6.8% |
| 2026-06-05 → 2026-06-08 | +5.9% | +2.8% | +4.5% | +5.4% |
| 2025-11-28 → 2025-12-01 | -5.8% | -4.5% | -5.0% | -5.1% |
| 2026-01-16 → 2026-01-20 | -4.8% | -3.5% | -4.9% | -5.0% |
| 2026-06-12 → 2026-06-15 | +4.7% | +6.4% | +7.5% | +5.0% |
| 2025-03-07 → 2025-03-10 | -4.4% | -5.7% | -6.7% | -4.9% |
The most recent weekend
Bitcoin moved -1.7% between the Friday 2026-09-04 close and the Tuesday 2026-09-08 open (nearest hourly prints). The regression said to expect the gaps below; the actual column is what opened.
| Ticker | Gap the weekend implied | Actual Tuesday gap | Rest of Tuesday |
|---|---|---|---|
| COIN | -1.4% | -2.1% | -1.0% |
| MSTR | -1.8% | -3.6% | -0.8% |
| IBIT | -1.7% | -2.0% | +0.2% |
Method, scope and what would make this wrong
What is measured. For every Friday to next-session pair since 2024-09-13, Bitcoin is read at the U.S. market close on Friday and at the U.S. market open on the next session (daylight-saving aware), using the nearest hourly print. The stock gap is the next session's open (Monday, or Tuesday when Monday is a market holiday) against Friday's close. Prices come from Yahoo Finance's public tape; all three stocks and Bitcoin are read from the same provider on the same dates.
Honest edges. IBIT began trading in January 2024, so its sample is the shortest and every row it joins says so. The Bitcoin anchors are the nearest hourly prints — they can differ from the exact closing and opening moments by up to an hour, and one provider's Bitcoin tape is not every venue's. The fade test ignores borrow costs, slippage and the impossibility of trading the open at the open. A gap that fades on average can still run violently on any single Monday.
Missing stays missing. Weekends with no usable hourly Bitcoin print near an anchor are dropped from the sample, not filled. If a price feed fails at export time the affected figure is marked not available below.