BNB Donchian Breakout Above the 200-Day Average — 4h

Status: Regime entry: beat random entries in one test window, not the other · Regime #4 · Listed 2026‑10‑14 Instruments: BNBUSDT perpetual only · Timeframe: 4h, trend read on the daily · 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 August 2018 to August 2022 (61 trades). From August 2022 to August 2026 it made +56.4%, still +25.0% without its best trade, but random entries with the same exit did nearly as well: it sits at the 91st percentile. Everything below the setup is disclosure, not a pass.

Where it fails the full bar:

• Random‑entry null: it cleared the null on BNB from August 2018 to August 2022 only. On BTC and ETH all four windows made money (+36.1% to +56.3%) and none cleared it (53.8th to 81.8th percentile): the filtered breakout mostly earned what being long earned.
• Drawdown: 47.3 summed points in its good window and 22.6 in the other, against a bar of 20. Five of its six coin‑and‑window cells breach it; BTC's second window, at 17.8, is the exception.

It came out of sweep #4, pre‑registered on 2026‑08‑06 as the one allowed revisit of sweep #1's 4h Donchian breakout: the same breakout and 3×ATR trail, plus a daily SMA‑200 trend filter. The sweep killed it on drawdown: the first window breached the bar on all three coins (35.1, 64.1 and 47.3 summed points). That sweep ran before the random‑entry null joined the bar. The re‑audit of sweeps #1–#10, pre‑registered on 2026‑10‑14, ran the null on every window test from those sweeps that made money, made money without its best trade and had at least 20 trades: thirty in all. With a 95th‑percentile bar, about one window test in twenty clears by luck alone. Of the thirty, one or two would by chance; eight did, this one at the 97.5th percentile.

The setup (compilable spec)

Market: crypto BNBUSDT perpetual, 4h trading timeframe. Longs only. Trend: the 4h close > SMA(200) of the daily closes, as of the last closed daily bar. Entry (4h): close crosses above highest(55), the high of the 55 bars before the breakout bar. Exit: trailing stop 3×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 spot candles

Returns are per‑trade percentages summed, like every return in the research log. The price path is Binance spot candles, as in every August sweep; the funding is the real USDT‑M perpetual series. Holding BNB made +1973.2% and +104.2% over the same windows.

• Random entries, 2018–22: 400 runs of the same exit on random bars had a median of +45.7% and a 95th percentile of +182.8%. The random entries keep the 3×ATR trail but not the trend filter — they are drawn from every bar of the window — so the test measures the breakout and the filter together against random timing.
• Random entries, 2022–26: a median of +11.1% and a 95th percentile of +65.4%; the real +56.4% sits at the 91st.
• Fees and fills: priced at the perpetual's 0.05% a side only. No fee‑stress or next‑open run exists for it.

The record by year

*2026 through July. Each trade counts in the year it was entered. The window opens in August 2018; the first trade came in February 2019.

• Two quarters made most of its money: January to March 2021, +171.9% on 7 trades, including the best (+67.4% in February 2021), and January to March 2024, +45.5% on 5. The other 24 quarters with trades netted +38.9 points.
• The filter kept it mostly out of 2018, 2022 and 2026: no trades in 2018, four in 2022 and three in 2026 to July, where sweep #1's unfiltered run of the same breakout traded 8, 18 and 14 times.
• Last high: October 2025, at +271.0% summed since August 2018. It ended July 2026 at +256.3%, 14.6 points below.
• Last twelve months (August 2025 to July 2026): 10 trades, +13.9%.

What the result depends on

Sweep #4 pre‑registered it as a single variant, so it has no neighbors of its own. The nearest evidence is sweep #1's runs of the same breakout on BNB without the filter (the 4h table in the log's H1 section, and the re‑audit's results table):

The filter removed about a third of the trades. The unfiltered runs paid twice this entry's fee, so their returns are not a like‑for‑like comparison, and no run measured the filter at equal fees. The 2×ATR version without it sat at the 89th percentile of random entries in 2018–22; the 3×ATR version without it was not null‑tested.

Before you adopt it

• Two quarters made most of its money: January to March 2021 (+171.9%) and January to March 2024 (+45.5%). The other 24 quarters with trades netted +38.9 points.
• It beat random entries in one window on one coin. Everywhere else — BNB since 2022, BTC and ETH in both windows — the same rules made money, but not convincingly more than random entries with the same trail (53.8th to 91st percentile).
• Size it for its drawdowns, 47 and 23 summed points, not for the library's 20.
• Paper‑trade it first. Adopting it watches BNBUSDT on the 4h chart and alerts every setup, about 15 a year.

Every number here can be re‑run: npx tsx scripts/research/replicate-sweep4-corrected-costs.ts for both windows and every trade (the record by year sums them), npx tsx scripts/research/replicate-sweep1-lowtf.ts for the unfiltered versions, and npx tsx scripts/research/null-reaudit-sweeps-1-10.ts --only "#4 P3" for the null.