How long before a trading bot shows results?

The constraint is statistical rather than motivational. Thirty trades is enough to notice a very large effect and nothing else; a hundred begins to be informative; several hundred is where the numbers can be treated as estimates. Nothing compresses that, because more backtests of the same period do not create more independent observations.

This is genuinely awkward for low-frequency strategies and it should be stated rather than glossed over. Ten trades a month is 120 a year, so demonstrating an edge with reasonable confidence takes years. Any vendor showing you a track record measured in weeks is showing you variance.

It is also why out-of-sample testing matters so much. It is the only way to get evidence faster than real time, and it is only worth anything if the held-out period was genuinely used once.

What to watch in the meantime

Judge behaviour rather than profit and loss in the first months. Is it taking the trades it said it would take, at the sizes it said, refusing the conditions it said it would refuse? A system behaving as documented is the thing you can actually verify early.

Track the drawdown against the tested figure too. A decline within the historical range tells you nothing is wrong; one well outside it is worth investigating — see why is my trading bot losing money.