An AI does not make a strategy profitable. It executes a strategy consistently. Whether the result is green depends on three things that exist with or without AI.

1. A real edge

You need a reason the strategy makes money more often than not, or wins bigger than it loses. Automation amplifies whatever edge exists — and just as faithfully amplifies a negative one. Run a losing rule 500 times perfectly and you lose 500 times. See win rate vs profit factor for what “edge” actually means.

2. Risk sizing

The fastest way to turn a winning edge into a blown account is oversizing. Position sizing, not entry timing, is the survivable variable — being right on direction and wrong on size still bleeds you out. See position sizing and risk of ruin.

A profitable edge, oversized, is a losing account with extra steps. Sizing is the difference between surviving variance and being erased by it.

3. Costs

Slippage, spreads, and fees quietly eat returns, especially at high frequency or on illiquid contracts. A strategy that looks profitable before costs can be underwater after them — see slippage. This is a place good execution actually helps, by getting better fills than a human clicking manually.

The honest bottom line

AI trading can be profitable when a genuine edge is sized sanely and executed cheaply. It is never profitable by default, and no software can promise a return. If a pitch guarantees profit, that alone tells you to walk (how to spot an AI trading scam).