Renko bars are constructed by removing information from the price curve. You can see a trend, but not its acceleration and deceleration. So your algorithm has less information available and consequently will be less profitable than with a real price curve.
Renko bars are meant as a visual aid, with the idea of not getting distracted by the dents and ripples in a curve with real bars. They are not meant for automated trading. A computer won't get distracted by dents and ripples and thus has no use for Renko bars.
If you don't mind a little constructive discussion, and while I agree with you on most parts of your reply, I could also claim that OHLC bars/candlesticks are not suitable for algorithmic trading because they too remove lots of price information (also as a visual aid). Would I be wrong?
For example, consider Renko with 1 pips brick size and H4 OHLC, I'm sure you'll agree that the former (Renko) carries a lot more information than the latter. Sampling every 4 hours does a much stronger data reduction than waiting for 1 pips move with bricks.
Renko and other CRB (constant range bar) types most powerful (and sometimes most troublesome) feature is that they're not time based, IMHO. So powerful that I'm still seriously researching it and definitely will consider it for automated trading. Because, in both cases we're speaking of data reduction, we take lots of price ticks and collapse it to either an OHLC candlestick or a brick, the difference is that OHLC bars are sampled by time (periods), where bricks are sampled by price (difference). Because profits come from price differences, bricks could even be more suitable for trading. Strong words, I know. The research on this interesting subject is ongoing...
