Wrong attitude. You've never had a negative day. Who the hell cares if you can or cannot outperform the benchmark in the end? And by the way, your algo DID outperform the benchmarks. Can't you read your own chart?!? Look at the March 12th line for example. How is that not outperforming? Scale up fer crying out loud!
You have a dream algo. No losses in a month. Triple your exposure now for a month, and see how it does.
I'm curious to ask, seeing how slowly the PnL is grinding higher, (although the consistency is awesome), do you not wonder about the days where you were making a killing with your manual trading? I realize you had huge swings where you would be up a ton and then down a ton. But it seems to me like if you had the guts to make those huge gains, which takes a certain skill set and psychological profile, was it not perhaps just small tweaks that you need to make in order to lessen the huge drawdowns? Most of us probably cannot have the massive gains you had, so that in itself seems like a skill, and hence I'm of the opinion that to prevent the big losses, it may have just been small tweaks necessary in order to hang onto the winners, no? I doubt it was just straight up gambling, and since you could already make those gains, it just seems much easier to figure out how to prevent the losses than even trying to figure out how to get to your level to begin with.
What I used to do was trading mean reversion and breakout at the same time. I was "groomed" to think mean reversion is how pros trade. I used breakout as a reversal to make up the loss in mean reversion. Most times it worked. Mean reversion setups come quick and easy, and when price trends, I reversed to breakout. Problem is coming up with the rules to indicate when the market switched over. I would come up with a set of conditions that would work for 2-3 months then next thing i know market swings like crazy and I lose on both reversion and breakout setups (that's usually the big losing days). So I decided to stick with breakout, because I feel it's the approach that makes the most sense to me. Price expansion is as clear as day (especially during OR). I just had to come up with the risk profile/timeframe that can yield consistency and acceptable risk of ruin. I'm ok with the slow grind since it's fully automated (besides teh weekly vps reset). Like I said, if I can outdo the benchmarks (+40% annual expected), then I'm good.