▲ Work in progress. This site is built and researched in the open: pages, numbers and findings may change or be wrong, and the record has gaps. Not advice. More
The Lab
Ideas traders use, put to the record
← the Lab · notes · dated note
#18 Bitcoin after the CPI flush: where the liquidations sit
A session note, 14 Oct 2026, odds as of 14:33 UTC. Dated and frozen: it is never refreshed. Public prices only.
Last checked: 16 October 2026.
US inflation (CPI) came in higher than expected at 12:30 UTC, and Bitcoin dropped. This note shows where traders on leverage get liquidated, and how likely the price was to reach each level, as of 14:33 UTC.
- The candle before CPI (12:15) closed at 83,837.
- The dump candle (12:30) closed at 82,806.
- The low by 14:15 was 81,850.
- At 14:33 the price was 81,961. Bitcoin had been moving about 1.6% a day.
The levels nearest the price
The chance of a touch from 14:33 UTC, over the next hours and days:
| level | who gets liquidated | away | 1 h | 4 h | 10 h (end of the UTC day) | 1 day | 1 week |
|---|---|---|---|---|---|---|---|
| 84,256 | 100x shorts, opened before CPI | +2.8% | under 0.5% | 1.0% | 3.2% | 10% | 50% |
| 83,220 | 100x shorts, opened into the dump | +1.5% | under 0.5% | 3.7% | 12% | 28% | 70% |
| 83,078 | 50x shorts, opened at the low | +1.4% | 0.6% | 6.4% | 17% | 32% | 70% |
| 82,259 | 100x shorts, opened at the low | +0.4% | 17% | 46% | 65% | 76% | 94% |
| price at 14:33: 81,961 | |||||||
| 81,564 | 50x longs, bought in the dump | −0.5% | 12% | 35% | 57% | 68% | 88% |
| 81,441 | 100x longs, bought at the low | −0.6% | 5.9% | 25% | 42% | 68% | 88% |
| 80,903 | 25x longs, opened before CPI | −1.3% | 0.7% | 6.2% | 17% | 32% | 70% |
| 80,622 | 50x longs, bought at the low | −1.6% | 0.5% | 3.9% | 12% | 25% | 70% |
| 80,064 | 20x longs, opened before CPI | −2.3% | under 0.5% | 1.4% | 5.5% | 14% | 54% |
Already wiped out by 14:33:
- 100x longs opened before CPI (83,418);
- 50x longs opened before CPI (82,579);
- 100x longs bought in the dump (82,392).
The likeliest touches are the closest: 82,259 just above, and 81,564–81,441 just below. Within the UTC day each was better than even, or close to it.
Every level
| level | who | away | 1 h | 2 h | 4 h | 6 h | 10 h | 1 day | 1 week |
|---|---|---|---|---|---|---|---|---|---|
| 90,672 | 10x shorts, into the dump | +10.6% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | 9.2% |
| 89,626 | 10x shorts, at the low | +9.3% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | 11% |
| 87,610 | 20x shorts, before CPI | +6.9% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | 0.9% | 13% |
| 86,771 | 25x shorts, before CPI | +5.9% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | 1.8% | 20% |
| 86,532 | 20x shorts, into the dump | +5.6% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | 1.4% | 24% |
| 85,704 | 25x shorts, into the dump | +4.6% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | 1.0% | 3.1% | 29% |
| 85,533 | 20x shorts, at the low | +4.4% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | 1.0% | 4.1% | 30% |
| 85,095 | 50x shorts, before CPI | +3.8% | under 0.5% | under 0.5% | under 0.5% | 0.6% | 2.1% | 4.5% | 36% |
| 84,715 | 25x shorts, at the low | +3.4% | under 0.5% | under 0.5% | under 0.5% | 0.6% | 1.7% | 7.8% | 43% |
| 84,256 | 100x shorts, before CPI | +2.8% | under 0.5% | under 0.5% | 1.0% | 1.2% | 3.2% | 10% | 50% |
| 84,048 | 50x shorts, into the dump | +2.5% | under 0.5% | 0.6% | 1.2% | 2.2% | 3.9% | 13% | 54% |
| 83,220 | 100x shorts, into the dump | +1.5% | under 0.5% | 1.6% | 3.7% | 7.7% | 12% | 28% | 70% |
| 83,078 | 50x shorts, at the low | +1.4% | 0.6% | 2.9% | 6.4% | 10% | 17% | 32% | 70% |
| 82,259 | 100x shorts, at the low | +0.4% | 17% | 33% | 46% | 55% | 65% | 76% | 94% |
| 81,564 | 50x longs, into the dump | −0.5% | 12% | 21% | 35% | 43% | 57% | 68% | 88% |
| 81,441 | 100x longs, at the low | −0.6% | 5.9% | 13% | 25% | 34% | 42% | 68% | 88% |
| 80,903 | 25x longs, before CPI | −1.3% | 0.7% | 2.1% | 6.2% | 10% | 17% | 32% | 70% |
| 80,622 | 50x longs, at the low | −1.6% | 0.5% | 1.3% | 3.9% | 6.2% | 12% | 25% | 70% |
| 80,064 | 20x longs, before CPI | −2.3% | under 0.5% | 0.6% | 1.4% | 2.4% | 5.5% | 14% | 54% |
| 79,907 | 25x longs, into the dump | −2.5% | under 0.5% | under 0.5% | 1.2% | 2.2% | 3.9% | 13% | 54% |
| 79,079 | 20x longs, into the dump | −3.5% | under 0.5% | under 0.5% | under 0.5% | 0.5% | 1.4% | 6.6% | 36% |
| 78,985 | 25x longs, at the low | −3.6% | under 0.5% | under 0.5% | under 0.5% | 0.7% | 2.1% | 6.6% | 36% |
| 78,167 | 20x longs, at the low | −4.6% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | 1.1% | 2.5% | 22% |
| 75,872 | 10x longs, before CPI | −7.4% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | 0.5% | 11% |
| 74,939 | 10x longs, into the dump | −8.6% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | 11% |
| 74,074 | 10x longs, at the low | −9.6% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | under 0.5% | 9.2% |
The 5x levels (65,900–67,500 below, 97,800–100,200 above) and 10x shorts from before CPI (91,802) are more than 12% away. That is further than the curve was measured, so they get no odds.
The time of day matters
Bitcoin's typical move in each hour of the day (UTC), against a normal hour, over four years:
| hours (UTC) | how busy |
|---|---|
| 13:00–14:00 | 1.3x |
| 14:00–15:00 | 1.6x, the busiest |
| 15:00–16:00 | 1.4x |
| 16:00–18:00 | 1.2x |
| 00:00–02:00 | 1.1–1.15x |
| 04:00–07:00 | about 0.8x, the quietest |
The US open and its data releases fall at 13:00–16:00 UTC, and 14:33 sits inside them. The odds above use the last few days' average move, not the hour, so the hours just after 14:33 were probably somewhat likelier than shown.
The maths, step by step
1. The liquidation price. A trader at L times leverage puts down 1/L of the position as margin. They are liquidated when the loss eats it, less a small cushion the exchange keeps: the maintenance margin, about 0.5% on Bitcoin.
- A long goes at entry × (1 − 1/L + 0.005). A short goes at entry × (1 + 1/L − 0.005).
- Example: a 50x long opened at 83,837 goes at 83,837 × (1 − 0.02 + 0.005) = 83,837 × 0.985 = 82,579.
- The price fell through that within the dump candle.
2. The odds of a touch. Example: the 100x short opened at the low, liquidated at 82,259. Price at 14:33: 81,961. Within 4 hours?
- The distance, as a growth rate: 82,259 ÷ 81,961 is 0.363% higher.
- The volatility, called σ (sigma): how much Bitcoin typically moves in an hour, from its recent hours, with the newest counting most (an "exponentially weighted" average, made popular by J.P. Morgan's RiskMetrics). It was 0.330% an hour.
- Over 4 hours the typical move grows with the square root of the time (a rule as old as Bachelier's 1900 model of prices): 0.330% × √4 = 0.660%. Up and down hours partly cancel.
- z = distance ÷ typical move = 0.363 ÷ 0.660 = 0.55. The level was about half a typical move away.
- The market's own curve: over four years of Bitcoin's hours, levels about 0.55 typical moves away were touched within 4 hours 46% of the time.
- The textbook bell-curve formula says 58%. It overstates near levels a little and badly understates far ones, so this note uses the market's own record. Big moves are more common than a bell curve allows, as Benoit Mandelbrot showed in 1963.
Read with care
- Liquidation levels don't pull the price. An earlier study of ours found no edge in trading them after fees. These odds come only from distance and how much Bitcoin was moving.
- Real traders got in at many prices, not only these three, so their liquidations are spread out. No data on where positions really sat was used.
- No view on direction. Up and down are treated alike. Over the four years, levels below the price were touched a little less often than levels above.
- The odds were as of 14:33 UTC. They went stale as the price moved.
- Cross margin, fees and funding move a real liquidation price a little. These levels assume isolated margin.
How it was made
- Entries: Binance perpetual 15-minute candles.
- The odds: Bybit linear hourly candles, 15 October 2022 to 14 October 2026 13:00 UTC (35,039 hours), a start every 2 hours.
- The volatility: remembers about four days.
- The curve: levels from 0.5% to 12% away, both sides, waits of 1 hour to a week, read in bands of 0.1 in z. Rare far bands are pooled with everything beyond them.
Update, 16 October 2026: a note on the tables
Added after this page was published. The text above is left as it was.
- A few further levels show higher odds than nearer ones, for example 86,771 at 1.8% in a day against 86,532 at 1.4%. The odds should fall with distance. This is noise in our own curve: it is read from history in bands of z, and a thin band takes the rate of every band beyond it. Read neighbouring levels as about the same.
Update, 27 October 2026: where the idea comes from
Added after this page was published. The text above is left as it was.
- The release. The CPI is the US Bureau of Labor Statistics' monthly measure of consumer prices. It comes out at 8:30 New York time, which was 12:30 UTC that day. Markets move on the surprise: the gap between the number and what forecasters expected.
- Why the hour after a release is the busiest. Louis Ederington and Jae Ha Lee found in 1993 that scheduled US releases, the jobs report and the CPI among them, came with most of the biggest moves in interest-rate and currency futures.
- The time of day. Robert Wood, Thomas McInish and Keith Ord found in 1985 that US stocks move most at the open and the close. The busy hours in our table, around the US open, are the same pattern in Bitcoin.
- The odds of a touch come from the first passage problem: when does a wandering price first reach a level? It is explained on our page "First passage: the chance a price touches a level", and worked through for one level above.
- The catch. The odds used the last few days' average move. A release day is busier than that, so the near levels were probably a little likelier than shown (as the time-of-day section says). And no data on where positions really sat was used: the levels come only from leverage and three entry prices.
- Theirs and ours. The release is the Bureau's. The touch formula (Bachelier), the weighted volatility (RiskMetrics), fat tails (Mandelbrot), and the work on releases and time of day are theirs, credited below. The entries chosen, the levels and the market's own touch curve are ours.
Credits
- Louis Bachelier, "Théorie de la spéculation", Annales scientifiques de l'École Normale Supérieure, 1900: prices as Brownian motion, the square-root-of-time rule and the textbook touch formula.
- J.P. Morgan and Reuters, RiskMetrics — Technical Document, 1996: the exponentially weighted volatility.
- Benoit Mandelbrot, "The Variation of Certain Speculative Prices", Journal of Business, 1963: fat tails.
- Louis H. Ederington and Jae Ha Lee, "How Markets Process Information: News Releases and Volatility", Journal of Finance, 1993: the biggest moves come with scheduled releases.
- Robert A. Wood, Thomas H. McInish and J. Keith Ord, "An Investigation of Transactions Data for NYSE Stocks", Journal of Finance, 1985: the busiest hours of the trading day.
- The liquidation levels follow the exchanges' published margin rules, simplified (isolated margin, a 0.5% maintenance margin).
- The market's own touch curve is our own.
- Data: Binance's public perpetual 15-minute prices; Bybit's public linear hourly prices.
- PerpGremlin and Claude, Bitcoin after the CPI flush: where the liquidations sit, TerminalGremlin, 2026: the tests, the code and the words.
Bitcoin after the CPI flush: a day later →
Educational research from public market data; any trading discussed is paper, never real. Nothing here is advice. The full notice.