some probably quite useless thoughts...
when it comes to crypto, I think the best strategy is to work on really short periods of time
what I mean is - most of the indicators should be very short sighted
like 3m, 5m, 9m even 2m. perhaps some high pass filtering on 20m, 30m
1. spam confirmations. as indicators are shortsighted - they are likely to align.
indicator combinations yield much more precise control over entries / exits
2. use something trend following like Ehler's even better sinewave to avoid whipsaws.
alternatively - some form of "throttling". like RSIS trick I described.
falling MMI on (!) 4m also qualifies
3. rely on trailing stops and do set a (quite loose) stop loss
4. favor priceClose() over price(). do test both though.
some indicators tend to be more strategic than tactical by their intrinsic nature
5. occasionally - some roofing, filtering, windowing, smoothing can be a game changer
6. favor steadily raising equity curve over flat one with spikes during black swans.
do make sure algo would have survived those in past
7. any change affects everything. rules can become obsolete, hindering, improvable.
parameter optimization can be helpful but it can also lead you astray