The crypto regime indicator

Read these figures as one whole, not in isolation. A single strong number says little without the drawdown and the period next to it.

The strategy has only two states: invested in Bitcoin, or on the sidelines in cash. No leverage, no shorting (long-flat). A stress thermometer built on public on-chain data sets the state: invested when calm, to cash as stress rises. The question is simple: can you largely avoid the two deep crypto winters without giving up your return along the way?

Test period January 2015 to June 2026 (11.4 years), deliberately started at the trough after the 2014 crash, the most favorable starting point for buy & hold, and therefore the toughest test for the strategy. This is a backtest: a measurement of the past, not a prediction. All headline figures include 0.15% trading costs per switch.

We test it this way because a strategy is only worth something if you can actually stick with it. Readability and the psychology of holding on matter as much as the numbers: a mechanical rule with few switches is easier to follow than a gut feeling, and we show the downsides as prominently as the upsides.

The stress thermometer combines four signals. What each one watches, from the heaviest to the lightest in the model:

The four are combined into a single stress reading from 0 to 100 and lightly smoothed. Above a stress threshold the strategy moves to cash; only once the thermometer falls back to calm does it step in again. That wide margin between out and in prevents nervous back-and-forth switching, hence 2 to 3 positions per year. The position lags one day: the signal lands at today's close, you trade the next day, so there is no same-day look-ahead. The exact weights, the normalisation and the precise thresholds are the recipe of the model and are shown in the full version.

Strategy (orange) and Bitcoin buy & hold (gray), both starting at 1, logarithmic scale. The band beneath the curve shows when the strategy was invested (orange) and when it sat in cash (grey); the difference arises almost entirely in the two deep winters, not from hitting tops or bottoms.

Over these 11.4 years the strategy took 28 positions and was invested 57.5% of the time. Final result ×412 against ×196 for buy & hold, with a deepest drawdown of −57.5% against −83.6%. Read the combination, not the individual figures: the higher return comes almost entirely from not having fully lived through the two winters.

Every study ends in exactly one of four outcomes and is published regardless of which, including a null result. This is not a cherry-picked success story. For this study we see, on this data and without look-ahead, less drawdown and a higher return than the benchmark (Bitcoin buy & hold).

This is a conclusion of the test, not a value judgment and not a prediction. It does not mean you will by definition win or lose with it; only that this pattern was measured in our setup, on this data. A higher return advantage for the future is therefore not demonstrated, which is why we keep testing each quarter on new data.