After 2.5 months of algorithmic stock trading, having days like...

The growth on stocks has been around 20% per month net because I prioritize risk of ruin as close to 0 as possible (my risk of ruin is not 100% of equity but 50%, and I have a 0.21% probability).
On futures, my slaves make around 60% per year.
But more important than the profit is that, in futures I need to keep at least 155K on the account just in case a tail happens.
On stocks, I will be able to make the median 12K after 5 months of trading SC. (x10 faster)
I cannot imagine what the market was like in 2025 or 2020. I can see the backtest, but I have survivor bias; my strategies are created now, for this market, and not for the opportunities you folks saw then.
1. Backtests always on ticks with bid, ask and size.
Whether the strategy is for longs or shorts, you need to know exactly how you fill. Test your safety rules; don't assume them.
And don't you just set commissions to 2% on the backtest and it is positive. That's lying to yourself; be humble.
Sadly, this is a research area in itself, and AI will not help you, since it has quite a lot of requirements (it will, but it will be BS, and you'll think it's gold).
If I don't have enough data to test my strategy on a stock, I create it.
Idk why people tend to say "I will size x5 from january to may on tuesdays at 10".
x+y+z != xyz
Same for filters btw.
2 things here, first, you don't modify anything without coding it perfectly with at least 90% fidelity with real execution.
Think again about it; those things are the ones that induce "edge decay" (usually a skill issue).
NORMALIZATION. ALWAYS.
7. Historically accurate data for everything you use.
8. Include dead tickers
9. Know which prints are real
10. Careful with lookahead
11. Limit size and orders to liquidity.
12. Live days vs backtest fidelity > 90%
13. docs and version control
All this is the institutional standard basics (and I am just commenting a few) trying to bring value.
I am tired of writing for this year, bye
