185 crypto strategies tested. One rule held up.
Trend following, momentum rotation, breakouts with trailing stops, funding-rate filters, a machine-learning model: we tested the strategy families systematic crypto researchers publish, on 8.7 years of exchange data with fees and slippage. Chosen on 2018-2022 only and judged on 2023 to today, one published rule beat simply holding Bitcoin on every risk-adjusted measure. It kept pace with BTC, 53 % a year against 55 %, while its worst fall was -30 % instead of -53 %.
The rule against holding BTC, since 2018
log scale · after costs · hypotheticalRule (BTC+ETH, spot)BTC, bought and heldout-of-sample from 2023
Simulated results on Binance daily candles, 0.15 % per side in fees and slippage, spot only, no leverage. They are not trades anyone made. The forward record below is the real thing: it started on 2026-09-24 and is sealed every day.
I.Out-of-sample results hypothetical
Judged only on years the rule had never seen.
Every strategy was chosen on 2018-2022. The table shows 2023-01-01 to 2026-09-25, which played no part in any choice. The bar was Bitcoin itself: a strategy had to beat holding BTC on Sharpe, Sortino and MAR, with a smaller worst drawdown, and still beat it at twice the costs.
| Out of sample, 2023-01-01 to 2026-09-25 | Per year | Sharpe | Sortino | Worst drawdown | MAR |
|---|---|---|---|---|---|
| Selected rule (spot, BTC+ETH) | +52.8 % | 1.41 | 2.34 | -30 % | 1.75 |
| BTC, bought and held | +54.5 % | 1.17 | 1.83 | -53 % | 1.03 |
| BTC+ETH, bought and held | +40.4 % | 0.91 | 1.38 | -60 % | 0.67 |
| Top-30 coins, held (point in time) | -21.2 % | 0.03 | 0.04 | -90 % | -0.24 |
| Same rule at 1.5x gross cap (research variant) | +75.2 % | 1.35 | 2.22 | -44 % | 1.73 |
Per year = compound annual growth. MAR = yearly growth divided by the worst drawdown. The 1.5x variant uses the same entries and exits with the book allowed up to 1.5 times its size through perpetual futures, charged their real historical funding; it is shown as research, and the published record runs the spot rule.
Calendar years, rule against BTC held
| Year | Selected rule | BTC held | Difference |
|---|---|---|---|
| 2018 | -22 % | -72 % | +51 % |
| 2019 | +64 % | +94 % | -30 % |
| 2020 | +267 % | +302 % | -35 % |
| 2021 | +105 % | +60 % | +45 % |
| 2022 | -39 % | -64 % | +25 % |
| 2023 OOS | +91 % | +156 % | -65 % |
| 2024 OOS | +50 % | +121 % | -72 % |
| 2025 OOS | +35 % | -6 % | +42 % |
| 2026 (to 09-25) OOS | +26 % | -4 % | +30 % |
What the numbers support
The rule's edge is what it avoids. It stepped aside for most of 2018 and 2022, when holding BTC went -72 % and -64 % (the rule: -22 % and -39 %), and it rode the uptrends at a volatility-capped size. Over the whole test that doubled BTC's yearly growth with a much shallower worst fall.
What it does not do: it trails holding in the strongest bull years, it still had losing years, and since 2023 its yearly growth was about the same as BTC's - the gain was the smaller drawdown, not a higher return. 185 strategies were tried, so some of any winner's lead is luck; the method page shows the deflated Sharpe and every rival.
II.The rule, in full
Five lines. That is the whole method.
- After each daily close (00:00 UTC), take each coin's closing price.
- If the close is not above its 50-day simple average, the coin is risk-off: weight 0, the share sits in cash.
- If it is above, the coin is risk-on, sized to a 60 % volatility cap: weight = min(1, 0.60 / annualised volatility of the last 30 daily returns).
- The weight is held for the next day. Moving into it is charged 0.15 % per side of the change.
- Books: BTC+ETH (the core, the tested book) and a 10-coin basket (BTC, ETH, SOL, XRP, BNB, ADA, DOGE, LINK, AVAX, LTC), equal-weighted.
Rule id sma-50-volcap · spot only, long or cash, no leverage, no shorting · forward data: Kraken daily USD candles.
III.Charts
Every entry and exit, on every coin.
Candles since 2018 with the rule's 50-day line, each entry and exit marked, the days in a position shaded, and the full trade list. Days from the forward record appear once their contents are revealed.
IV.The forward record
Sealed every day, so nobody can edit a bad one away.
Every day's result is fingerprinted the moment it is computed and revealed seven days later - including by us. Your browser can check the whole chain.
- 1--Loading the record...-
Your browser re-checks every link of both chains - the record and its settlements - with SHA-256, and recomputes every revealed line.
- Next line
- after the 00:00 UTC close, about 00:07 UTC
- Service
- checking
Each line is written right after the daily close and chained to the line before it. Its contents stay sealed for seven days, then open here. This record started on 2026-09-24; it is short, and it is real. Offline check: download verify.mjs and run node verify.mjs.
Scorecard: did the rule beat just holding?
Every day of the record is settled automatically against Kraken's daily closes 1, 7 and 30 days later, and every settlement is chained and sealed like the record itself (/api/settlements). The scores below count only days that are already revealed, and they are recomputed each day after the close - losses included.
Loading the scorecard...
V.Why this exists
A track record nobody can check is not a track record.
Sealed, then revealed
The most common complaint about paid crypto channels is a record nobody can check. Here every day is fingerprinted at once and opened after seven days, good or bad.
Tested against 184 rivals
Trend, momentum rotation, breakouts, funding and a machine-learning model all ran under the same costs and the same out-of-sample test. The losers are published next to the winner.
Built to cut drawdowns
One rule, five lines, fully public. No black box and no secret indicator: you pay for the daily computation, the alerts and the discipline.
VI.Free weekly summary
The record, in your inbox once a week.
Once a week: where each coin stands against its rule, what changed, and the record's running result - losses included. Free, no card.
- The full forward record, revealed after 7 days
- The rule, the charts and the research, in full
- Every day's fingerprint, so you can verify the record
VII.Pro
Today's weights, the day they are issued.
The record above shows each day's fingerprint at once and its contents after 7 days. Pro shows you the contents the morning they are issued: every coin's weight, yesterday's weight and the close it was computed from. Each day's fingerprint is public at the same moment, so what you received can be checked against the record later. When a coin moves in or out of the market, or its weight moves by 10 points or more, you get an email alert that morning. Cancel any time.
Monthly
US$19 / month
Yearly
US$149 / year
Payment by Stripe. Renews until you cancel; cancelling on the billing page keeps access until the end of the paid period.
| Coin | Weight | Yesterday | Close (US$) |
|---|
VIII.Risk and what this is not
Read this before anything else on the page.
Not investment advice. The regime is one general, published rule, identical for every reader. It does not know your finances, goals or taxes, and it is not a recommendation to buy or sell anything. This site publishes research, data and software output.
Crypto assets are highly volatile and can lose most or all of their value. A rule that cut drawdowns in the past can fail in the future. Backtested results are hypothetical, were produced with the benefit of hindsight, and 185 strategies were tried to find this one; the forward record is short.
Only use money you can afford to lose, and check the rules for crypto assets where you live.