Weekly research 2026-39
Observed volume spike outcomes from stored exchange data.
Draft: awaiting owner approval
Coverage and frozen source
Week: 2026-09-21T00:00:00.000+00:00 to 2026-09-28T00:00:00.000+00:00 (end exclusive). Event cutoff: 2026-09-27T23:59:59.999+00:00 (inclusive).
Source read: 2026-09-30T11:54:43.351+00:00. 8374 recorded volume spikes, including 1454 mixed-direction spikes.
First observed: 2026-09-27T13:19:00.210+00:00. Last observed: 2026-09-27T23:59:02.503+00:00.
Counts cover these stored observations. Complete coverage of every hour in the week is unverified.
alerts: volume spikes recorded from exchange public feeds; spike_weekly.load_week
Source SHA-256: be5f60d4de2260abeae4ba3c892b6d62132774c5a4873d5e1cd41b4d9f7abddf
Measured follow-through
A hit means price moved with the recorded spike direction. Each horizon uses its own measured sample.
| Horizon | Measured | Hits | Hit rate | Median signed move |
|---|---|---|---|---|
| 5 min | 6920 | 3187 | 46.05% | -0.0087% |
| 15 min | 6920 | 3169 | 45.79% | -0.0196% |
| 60 min | 6920 | 3362 | 48.58% | -0.013% |
Largest recorded volume ratios
Five largest spikes by ratio to normal volume. Quote volume is USD notional as stored by the scanner.
| Coin | Venue | Direction | Volume ratio | Quote volume (USD) |
|---|---|---|---|---|
| ETHFI | binance | buy | 2275.53× | 528022 |
| WOO | binance | buy | 2255.81× | 394149 |
| MORPHO | hyperliquid | buy | 1646.78× | 585978 |
| MAGIC | binance | mixed | 1643.16× | 150450 |
| XAG | bybit | sell | 1367.50× | 1081963 |
Method and limits
- Recorded scanner alerts, not all exchange trades or market-wide volume. Mixed-direction spikes are excluded from hit rates.
- Event cutoff limits spike timestamps. Outcomes reflect database values at extraction, including later measurements; this is not an as-of outcome reconstruction.
- A hit means price moved with spike direction; zero is a miss. Medians are signed price changes, not trading returns or forecasts.
- No frozen funding, OI, liquidation or prior-week series in this issue. Missing observations are not zeros; retention may truncate coverage.