Trading-strategy family
REGIME_BULL_VOLTARGET_V1Volatility-targeted size up to 1.5× inside the 200-day filterFAIL
tested Oct 28, 2026 1 variants tried
What was tested
Long-only BTC, held only while the price is above its 200-day average. Position size = 1.5× at most, otherwise 0.60 divided by the recent 60-day volatility: bigger when the market is calm, smaller when it is wild. It keeps the 20% drawdown brake of our running V3 paper test unchanged. One rule, fixed in advance, no variants.
Judged on Bitstamp BTC/USD daily prices from Jan 1, 2013 to Aug 16, 2017, a period no earlier Anchor test had used. Costs 15 bp per side plus an assumed funding charge. Binance 2018–2026 was shown for information only, because earlier tests had already looked at it; it could only veto, never rescue.
Pre-registered pass bar and result
The bar was written and hashed before the run. Results are on data the selection step never saw (out-of-sample).
| Criterion | Bar | Result | Verdict |
|---|---|---|---|
| G1 Fee gate: net return ≥ 50% of gross | ≥ 0.5 | 0.93 (78.7% / 84.8%) | PASS |
| C0 No liquidation | 0 | 0 | PASS |
| C1 Max drawdown no worse than −30% and shallower than buy-and-hold | ≥ −30%, shallower than −84.9% | −37.1% | FAIL |
| C2 Bull-market capture | ≥ 50% | 26.7% (2 of 13 bull legs ≥ 50%) | FAIL |
| C3 CAGR ≥ 70% of buy-and-hold, or Calmar above buy-and-hold | ≥ 175.2% or Calmar > 2.95 | 78.7%; Calmar 2.12 | FAIL |
| C4 Beat the V2 lines: Calmar above V2(b) and V2(a), CAGR above V2(b), drawdown within 5 points of V2(b) | Calmar > 3.81 and > 3.62; CAGR > 155.5% | Calmar 2.12; CAGR 78.7% | FAIL |
| C5 Calmar above random timing 95th percentile (1000 seeds) | > 2.08 | 2.12 (95.0th percentile, borderline) | PASS |
| C6 Window contains a real bull and a real bear | ≥ 1 each | 13 bull / 13 bear legs | PASS |
| Veto check on Binance 2018–2026 (information): drawdown no worse than −40% | ≥ −40% | −55.1% | INFO |
Why it failed
It survived bears on paper (flat in 12 of 13 bear legs, never liquidated) but missed the bulls: it was in the market only 17% of the time on average, captured 26.7% of the bull-market gains, and earned 78.7% a year against 250.3% for buy-and-hold and 155.5% for the plain 1× V2(b) rule. Two brake exits in a row still took the drawdown to −37.1%, past the −30% line.
The main lesson: the 20% drawdown brake inherited from V3 is what killed bull participation. After the brake fires, the rule only buys back once the price has dropped below its 200-day average and risen above it again. In 2016–2017 the price stayed above the average for months, so the line sat in cash through all five bull legs of those years. Without the brake (information only) the same sizing averaged 0.68 exposure and 237% a year, but with a −57.6% drawdown. V3 keeps running under its own locked rules; nothing about it was changed.
Variants tried
1 rule, fixed in advance (target volatility 0.60, cap 1.5×). Other targets, caps or brake levels were not tested, so we cannot say whether the failure is specific to these numbers. No re-tuning and no re-run after the result. The first run stopped on a one-line programming error before producing any number; it was fixed, logged with before/after file hashes, and the whole test was run again from scratch once.
What this test cannot tell
One rule on one 4.6-year window of a very different, thin market (2013–2017), whose history is public and broadly known. There were no perpetual futures then: spot prices stand in and funding is an assumption. Daily bars hide intraday moves; the worst intraday-low drawdown was −48.7%. Nothing here says anything about the Kraken evaluation account's own rules.
Rules lock
Rules file SHA-256:
c5870c116583d786f9728fe0ae08c716dadd2c2e449259dafdf5c4bb8c51a9d6
Locked:
“not recorded” means the number or time was not written down at the time, and we have not filled it in afterwards.
Submit a rule
Have a rule with numbers? Send it to us on X @ProjectAnchorHQ or Telegram t.me/ProjectAnchorHQ (market, entry, exit, stop). Free; we only need the written rule, never keys, accounts or funds. We publish the result pass or fail, and credit you only if you say yes.