Sorted by monthly Google searches for the strategy's usual name (worldwide). Click a column to re-sort, click a row for how it works, the evidence and the sources.
A rule that only wins in one era is luck, not an edge. Green beats holding in that era, red loses to it.
Each family was re-run with every common setting (for example the 20, 50, 100 and 200-day average). If most settings beat holding on both coins, the idea is real. If only one does, it was fitted to the past.
Two of crypto's favourite timing rules. Both look great on a chart of the past, and both rest on very few data points.
In our backtest the only rules that held up were trend-following ones: holding while Bitcoin is above its 200-day average, a MACD crossover, or a 20-day breakout. They beat holding on a risk-adjusted basis on both Bitcoin and Ethereum in almost every setting we tried, mainly by cutting the worst fall from about −83% to between −52% and −68%. For most people, plain buy-and-hold or a weekly DCA is simpler and did nearly as well. This is research on past prices, not investment advice.
As a way to build a position, yes. $100 of Bitcoin every Monday since January 2014 made about 53% a year on the money put in. Putting everything in on day one would have made more, as it usually does in a rising market. DCA's real benefit is that a bad start hurts less: started at the 2017 top it is up about 290% today and from the 2021 top about 49%, while from the October 2025 top it is still down about 18%.
They often show a profit, but they lose to simply holding. Our monthly BTC grid (±15%, 30 levels) made money in 31 of 46 months since January 2023, yet $1 grew to about $1.26 while holding Bitcoin grew to about $4.69. The grid sells its coins as the price rises, so it lost to holding in every strong month. It only did better in sideways and falling months.
Not in our tests. None of the 7 RSI settings and none of the 8 Bollinger settings beat holding Bitcoin, even though they win most of their individual trades. In crypto, "oversold" can stay oversold for months, and selling when it looks "overbought" means missing the biggest rallies.
We found no rigorous study of crypto day traders specifically. The closest evidence is clear, though. In Brazil, 97% of people who day-traded index futures for more than 300 days lost money. In Taiwan, less than 1% of day traders earned reliable profits after fees. In Australia, 69% of retail clients trading crypto CFDs lost money in 2023-24.
No. It removes price risk, not other risks. Binance BTC funding paid about 11% a year on average since 2019, but the yearly figure ranged from about 3% to 31%. Funding is sometimes negative, so you pay. The short can be liquidated in a sudden spike if margin is thin, and the exchange holding it can fail, as FTX did in 2022.
Nobody can know from four data points. Past cycle highs came 371 to 546 days after each halving, and the gain shrank every time. When we held Bitcoin only for a fixed window after each halving, just one window length (18 months) beat holding's return; 12, 15, 21 and 24 months did not. That pattern is what fitting the past looks like.
Our tests say it is a poor trading rule. Buying Extreme Fear and selling Extreme Greed made roughly nothing on Bitcoin since 2018, while holding made over 20% a year, and none of 15 threshold combinations beat holding. Fear tends to last through bear markets, and greed shows up early in bull runs.
Because its few winners are huge and its losers are small. The 200-day rule won only about 1 in 5 trades on Bitcoin. Each loss was a small whipsaw, while each win rode a multi-month bull run, and being in cash during the 2018 and 2022 crashes avoided most of those falls.