Do RSI strategies actually work in crypto?

Sometimes, in some regimes — which is precisely the problem with the question. Classic RSI mean-reversion (buy under 30, sell over 70) does well in ranging markets and gets run over in strong trends, where 'oversold' keeps getting more oversold. On daily crypto bars our monthly re-tests show RSI reversion surviving out-of-sample on some coins and failing on others, with results that shift as regimes shift.

Treat any RSI rule as a hypothesis to test on your coin and timeframe, not a truth. It takes about a minute to backtest an RSI strategy on real data with an out-of-sample check — that beats any generic answer.