BNB Taker-Selling Exhaustion — 4h
Status: Regime entry: a record only — worked in one test window, not the other · Regime #10 · Listed 2026‑10‑14 Instruments: BNBUSDT perpetual only · Timeframe: 4h · Direction: long only Frequency: ~6 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 April 2023 to October 2026 (24 trades). From April 2020 to April 2023 it made money too, but fails the third, and its 18 trades are under the fourth's 20. Everything below the setup is disclosure, not a pass.
Where it fails the full bar:
- Random entry, 2020–23: +26.2% from April 2020 to April 2023 (PF 2.11), but no more than random entries with the same exit made (80.3rd percentile), and +8.0% without its best trade.
- Cross‑asset: BTC lost in both windows (−5.5%, −2.2%). ETH made +15.7% in 2020–23 (76th percentile) and lost 33.7% in 2023–26 (1.5th percentile).
- It is mostly BNB: on the same bars the RSI turn alone, with no flow gate, made +47.6% in 2020–23 and +45.1% in 2023–26. In the research log's words, BNB's 4h results owe more to BNB than to either gate.
It came out of sweep #14, pre‑registered on 2026‑10‑11: four RSI setups on crypto at 1h and 4h, two of them behind this aggressor‑flow gate, on BTC, ETH and BNB in two windows. That is 24 window tests. With a 95th‑percentile bar, about one window test in twenty clears by luck alone. One cleared: this one.
Why the app can't run it
Its gate reads the taker‑buy share of volume: how much of the last 24 hours' volume was bought by traders crossing the spread, which Binance prints with every candle, ranked against the last 28 days. The strategy language reads total volume but not its taker‑buy part, and has no rolling percentile, so the gate can't be written; the RSI trigger and the trailing exit are ordinary.
The setup
Market: crypto BNBUSDT perpetual, 4h trading timeframe. Longs only. Gate: the taker‑buy share of volume over the last 24 hours (six 4h bars) at or below its trailing 28‑day 10th percentile, a day of one‑sided aggressive selling. Entry (4h):
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
| Window | Trades | PF | Return | Strip‑best | Sum DD | Account DD | Random‑entry percentile |
|---|---|---|---|---|---|---|---|
| Apr 2020 → Apr 2023 | 18 | 2.11 | +26.2% | +8.0% | 8.6 | 7.1% | 80.3rd |
| Apr 2023 → Oct 2026 | 24 | 5.02 | +50.7% | +33.1% | 4.3 | 3.7% | 100th |
Returns are per‑trade percentages summed, like every return in the research log. Holding BNB made +2432.7% and +144.2% over the same windows.
- Random entries, 2023–26: 400 runs of the same exit on random bars, with the real funding billed, had a median of −3.9% and a 95th percentile of +23.3%. None beat it.
- Random entries, 2020–23: a median of −1.4% and a 95th percentile of +78.3%.
The record by year
| Year | 2020* | 2021 | 2022 | 2023 | 2024 | 2025 | 2026* |
|---|---|---|---|---|---|---|---|
| Trades | 6 | 1 | 10 | 4 | 7 | 7 | 7 |
| Return | +5.4% | +13.4% | +8.3% | −0.5% | +18.3% | +13.7% | +18.3% |
*2020 from April, 2026 through September. Each trade counts in the year it was entered.
- Four trades made most of it: +18.1% (July 2022), +17.5% (March 2024), +13.4% (October 2021) and +13.3% (May 2026), 62.4 of its 76.8 summed points. The other 38 trades made +14.5%.
- Last high: June 2026, at +78.5% summed since April 2020. It ended September 2026 at +76.8%, 1.6 points below.
- Last twelve months (October 2025 to September 2026): 8 trades, +19.7%.
What the result depends on
Nothing else was run: sweep #14 ran its fee stress, its flow‑lag check and its neighbors only for a strategy that cleared gates 1–5 on two assets, and none did (research log, sweep #14, “Results”).
Before you rely on it
- Much of it is BNB, not the gate: on the same bars the 4h RSI turn alone made +47.6% in 2020–23 and +45.1% in 2023–26.
- A few big trades: 24 in its good window, and four of its 42 made 62.4 of its 76.8 summed points.
- In 2020–23 it made +26.2%, but no more than random entries did (80.3rd percentile). On BTC and ETH the same rule lost money in 2023–26.
Every number here can be re‑run: npx tsx scripts/research/sweep14-rsi-crypto.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.