Say a strategy buys the coin with the biggest gain over the previous week. To replay that choice, every market eligible on that date has to be allowed into the ranking, including ones that have since disappeared from the exchange.

Start with today's list instead, and a former contender might never get considered. Its price history could be available and perfectly accurate. The rule still can't choose it if it isn't on the list. Coins listed later have the opposite problem: they must wait until they were actually available on that venue.

This is where I'd separate two experiments. Testing a chosen basket of coins is a reasonable question in its own right. Testing whether a rule could have selected that basket years ago requires evidence of which markets were eligible at each selection date. A current market directory doesn't supply that history; Bybit describes its spot lookup as returning Trading symbols.

Without the dated membership, the result belongs to the chosen basket. It doesn't establish which coins the rule would have picked from the markets available then.