Internet Computer · ICP/USDT · daily
What happens to Internet Computer after a capitulation day?
Since 2021, Internet Computer has shown a capitulation day 12 times. In the 12 cases old enough to score, the price was higher 7 days later 67% of the time, with a median move of +4.8%. On an ordinary day the median 7-day move for ICP is −1.3%.
Weak evidence43
There is a gap, but once you allow for how many studies were run, a gap this large is luck about 52% of the time, so it should not be relied on.
By holding period
| After | Cases | Higher | Median move | Middle half | Ordinary median | Gap | By luck | Proof score |
|---|---|---|---|---|---|---|---|---|
| 1 day | 12 | 42% | −0.6% | −3.4% to +3.3% | −0.2% | −0.3% | 91% | 8 |
| 3 days | 12 | 67% | +6.4% | −1.7% to +16.4% | −0.9% | +7.3% | 7.4% | 81 |
| 7 days | 12 | 67% | +4.8% | −2.9% to +9.4% | −1.3% | +6.1% | 52% | 43 |
| 14 days | 12 | 58% | +4.0% | −5.6% to +15.8% | −2.3% | +6.3% | 74% | 15 |
| 30 days | 12 | 42% | −3.4% | −13.8% to +15.5% | −5.3% | +2.0% | 97% | 8 |
“Ordinary median” is the median move of ICP over the same number of days starting from any day. “Gap” is the signal's median minus the ordinary one. “By luck” is how often a gap at least that large comes from randomly chosen days, corrected for the 3,572 studies run at each holding period (for this study taken alone: 10%).
Holding it for 7 days
Dips and best points use each day's low and high, measured from the close on the signal day. Costs assume 0.3% for a round trip: a 0.1% exchange fee each way and 0.05% slippage each way.
Did it hold up over time?
Splitting the cases by date: in the earlier 7 the gap over 7 days was +2.1%, and in the later 5 it was +7.1%. Same direction both times.
Every occurrence
One square per occurrence, oldest first. Green: higher 7 days later. Red: lower. Grey: not yet 7 days old.
| Date | Close | +1d | +7d | +30d |
|---|---|---|---|---|
| 22 Aug 2026 | $2.34 | +4.1% | +6.4% | +29.8% |
| 3 Nov 2025 | $3.93 | +31.9% | +73.3% | −4.9% |
| 10 Oct 2025 | $3.04 | +3.0% | −1.0% | +148.7% |
| 12 Nov 2024 | $8.81 | −3.3% | +3.8% | +55.0% |
| 17 Aug 2023 | $3.41 | +3.0% | +5.8% | −11.8% |
| 10 Jun 2023 | $3.69 | −0.9% | +7.9% | +10.4% |
| 19 Apr 2023 | $6.12 | −3.9% | −13.1% | −13.2% |
| 8 Nov 2022 | $4.36 | −15.5% | −8.7% | −1.8% |
| 11 May 2022 | $7.45 | −3.1% | +0.8% | −18.1% |
| 5 Jan 2022 | $30.64 | −0.2% | +14.2% | −27.4% |
| 8 Aug 2021 | $55.27 | +16.2% | +14.7% | +10.8% |
| 30 Jun 2021 | $49.94 | −8.5% | −9.2% | −15.5% |
The 12 most recent. All 12 are in the open data file.
The rule
Daily close at least 7% below the previous close on volume at least 2.5× the previous 30-day average. A steep fall on very heavy volume is the textbook picture of capitulation, when the last holders give up.
Occurrences closer than 7 days to the previous one are not counted again. The signal is confirmed at the daily close (00:00 UTC) and returns are measured from that close. Full method.
The same signal on other coins
| Coin | Cases | Higher after 7d | Median | Gap | Verdict |
|---|---|---|---|---|---|
 Neo | 24 | 42% | −6.7% | −6.1% | Weak evidence |
 Tellor | 25 | 40% | −5.8% | −5.3% | Weak evidence |
 Ethereum Classic | 29 | 48% | −0.2% | +0.8% | Weak evidence |
 Ravencoin | 20 | 40% | −3.7% | −2.3% | Weak evidence |
 Zcash | 24 | 58% | +6.0% | +5.8% | No better than chance |
 Dogecoin | 21 | 43% | −4.3% | −3.4% | No better than chance |
 Cardano | 27 | 52% | +1.2% | +2.1% | No better than chance |
 Dash | 20 | 55% | +2.9% | +3.4% | No better than chance |
Capitulation: heavy drop on heavy volume across all 94 coins
Other Internet Computer signals
New signals are sealed in the public ledger at the daily close. Members see them immediately; everyone else sees them a day later.
See today's signals