BTC Open-Interest Deleveraging Exhaustion — 1h
Status: Regime entry: a record only — worked in one test window, not the other · Regime #9 · Listed 2026‑10‑14 Instruments: BTCUSDT perpetual only · Timeframe: 1h · Direction: long only Frequency: ~15 trades/yr · Fees: 0.05%/side plus real funding
Why it is in the regime section, not the library
It failed the six‑check bar. It is listed under the regime bar, which asks four things of one of the two test windows: a net profit after fees and funding, still a profit without the best trade, a result above 95% of 400 random‑entry runs with the same exit, and at least 20 trades. It clears all four from May 2024 to October 2026 (33 trades). From January 2022 to May 2024 it fails the first three. Everything below the setup is disclosure, not a pass.
Where it fails the full bar:
- Two windows: −21.6% from January 2022 to May 2024 (PF 0.49), at the 3.8th percentile of random entries: it lost more than almost every random run did.
- An older window says the same. In the 2021 bull market, October 2020 to January 2022, a BTC‑only window the sweep ran for information and never gated, it lost 45.3% over 25 trades. In the log's words: three regimes, it loses in two and pays in one.
- Cross‑asset: ETH made +10.7% in 2024–26, no more than random entries did (86th percentile), and lost 26.0% in 2022–24 (2nd percentile). BNB made +5.6% in 2022–24 but lost money without its best trade, and lost 10.4% in 2024–26.
- Drawdown: 23.8 summed points in 2022–24, against a bar of 20.
It came out of sweep #12, pre‑registered on 2026‑10‑08: three open‑interest and positioning gates in front of the library's house trigger, on BTC, ETH and BNB in two windows. That is 18 window tests. With a 95th‑percentile bar, about one window test in twenty clears by luck alone. The only other cell the log reports above the bar is a crowd‑short gate on BNB in 2024–26, at the 99th percentile on five trades, which the log reads as meaning nothing.
Why the app can't run it
Its gate reads Binance's open interest, the number of BTC perpetual contracts open, from the exchange's five‑minute metrics, and ranks its 24‑hour change against the last 28 days. The strategy language has no open‑interest atom or feed and no rolling percentile, so the gate can't be written; the RSI trigger and the trailing exit are ordinary.
The setup
Market: crypto BTCUSDT perpetual, 1h trading timeframe. Longs only. Gate: the 24‑hour change in open interest (contracts, Binance's five‑minute metrics) at or below its trailing 28‑day 10th percentile, a day of positions closing. Entry (1h):
RSI(14) crosses above 30while the gate is open. Exit: trailing stop 2×ATR(14) below the highest price since entry. Fees: 0.05%/side (Binance USDT‑M taker) · funding charged and credited at every settlement held through.
Evidence: two windows on Binance perpetual candles and open interest
| Window | Trades | PF | Return | Strip‑best | Sum DD | Account DD | Random‑entry percentile |
|---|---|---|---|---|---|---|---|
| Jan 2022 → May 2024 | 40 | 0.49 | −21.6% | −26.6% | 23.8 | 23.8% | 3.8th |
| May 2024 → Oct 2026 | 33 | 1.68 | +14.3% | +6.0% | 11.2 | 9.1% | 98.3rd |
Returns are per‑trade percentages summed, like every return in the research log. Holding BTC made +30.0% and +38.8% over the same windows.
- Random entries, 2024–26: 400 runs of the same exit on random bars, with the real funding billed, put it at the 98.3rd percentile. The null's median and 95th percentile for this window were not recorded; across all of sweep #12's cells the random medians ran from −2.7% to −10.4%.
- Random entries, 2022–24: the real −21.6% sat at the 3.8th percentile.
The record by year
| Year | 2022 | 2023 | 2024 | 2025 | 2026* |
|---|---|---|---|---|---|
| Trades | 20 | 13 | 13 | 13 | 14 |
| Return | −18.7% | −3.4% | +11.2% | +3.7% | 0.0% |
*2026 through September. Each trade counts in the year it was entered.
- Its 2024–26 result is two quarters: six trades in July to September 2024 and January to March 2025 made +23.3% between them; the window's other 27 trades lost 8.9%.
- Last high: February 2025, at +1.1% summed since January 2022. It ended September 2026 at −7.2%, 8.4 points below.
- Last twelve months (October 2025 to September 2026): 16 trades, −2.1%.
What the result depends on
Nothing else was run: sweep #12 ran its fee stress, its one‑bar lag and its neighbors only for a strategy that cleared the full bar, and none did (research log, sweep #12, “Results”).
Before you rely on it
- The windows disagree: the same rule lost 21.6% in 2022–24, at the 3.8th percentile of random entries.
- Its 2024–26 result is two quarters: six trades made +23.3%, and the window's other 27 lost 8.9%. Since its last high in February 2025 it has lost 8.4 points over 25 trades.
- The log offers one reading, not a finding: in a falling market the first flush is the first leg of a longer decline, and in a rising one it is a reset, so a gate whose sign depends on the regime is telling you about the regime, not about the flush.
Every number here can be re‑run: npx tsx scripts/research/sweep12-open-interest.ts for the
windows and the null. The record by year, the last twelve months and the last high are the same
runs' trades, grouped by the date they were entered.