There are algos and algos.
Another approach to algo trading is to have a quiver of different algos. Different algos for different market fingerprints.
Often, like yesterday, its possible to see the probabilities. Designing the algo for a specific fingerprint can increase the profitability considerably. I don't have to pick too many fingerprint types. Its just a matter of having 22 or 3 algos, depending on the market, that is suitable for days with different expected ranges. Its the range of the day that determines which algo will perform the best. If I'm wrong, then I'll have a less profitable or a losing day. I can change algos during the day as I see what has developed. In Europe, I can run a different algo in the morning and switch to a different algo for the U.S. day. Market Profile is a great tool for assessing the probabilities. Sometimes I have to wait and not turn my algo on until I see what is developing.
Algos, as I said in the previous post, are not a machine that has data in one end and money out the other but the skills required to be a successful algo trader are different and take less time to acquire than for discretionary trading.