You might want to have a look at Van Tharp's "Trade Your Way To Financial Freedom". Don't be put off by the title. He covers position sizing and trade expectancy pretty thoroughly.
If you vary % risk of trades, then you aren't letting probability work to your advantage. If a setup is great, but stop is too far for your risk management to adjust for average participation risk, then just find entry on a lower time frame or wait for a pullback. It's the relation between a potential target vs actual risk that is mucho importante. Trades that are put on near cycle exhaustion have low probability, have to find a way to be in near the beginning of a cycle.
That's fine, but your average return will be smaller than mine percentage wise in relation to trading capital due to significant drawdowns that you are bound to incur when not using stops. Some folks look at a stop as if it was reducing chances of maximising equity growth, but I think otherwise, it increases as a stop eliminates occasions of massive drawdowns and let's you maximise return potential by not having a get out 50% below current valuations. Glass is either half full or half empty, a matter of perspective. If anyone could warrant that XYZ will recover then OK stops would be unnecessary, but there are never any guarantees in financial forecasts.