A Signal Is Not a Good Trade
“The AI says tomorrow has a 62% chance of going up.”
That sounds useful, but it is not a strategy.
You still do not know:
- How much to buy
- When to enter
- When to exit if wrong
- What costs and slippage look like
- Who stops the model if it fails
Without those answers, a signal is just an opinion.
Four Layers From Signal to Strategy
| Layer | Question |
|---|---|
| Signal | What direction, probability, or score does the model output? |
| Position | How much loss can this trade take? |
| Execution | What order type, timing, and fill price? |
| Risk control | When do we reduce size, stop, or review? |
Many beginners lose money at the first layer: they receive a bullish signal and go all in.
Why High Probability Can Still Lose
Suppose a model wins 60% of the time, but gains 1% on winners and loses 4% on losers.
Expected value = 60% × 1% - 40% × 4%
= 0.6% - 1.6%
= -1.0%
High win rate, negative expected value.
You need win rate, payoff ratio, costs, and drawdown together.
Common Misuse of AI Tools
1. Treating explanation as prediction
LLMs can explain news well. That does not mean they can predict price.
2. Treating historical correlation as future causality
A feature that worked before may stop working.
3. Treating one signal as a system
Without sizing and risk control, a signal is unlikely to survive live trading.
Pre-Launch Checklist
- Do you know historical win rate and payoff ratio?
- Are fees, spreads, and slippage included?
- Is single-trade loss capped?
- What happens after 5 consecutive losses?
- Does the strategy force trades when there is no signal?
- Who can stop the strategy immediately?
Quiz
Q1. Is an AI signal a full trading strategy?
A. Yes B. No
Q2. Why can a high win rate still lose money?
A. Payoff ratio and costs may be worse B. Win rate never matters
Q3. What must be added to a signal?
A. Position sizing, execution, risk control B. More slogans
Answer Key
Q1: B Q2: A Q3: A
Further reading: Investopedia — Expected Value · Investopedia — Position Sizing · IBM — Large Language Models
For education only. Automated strategies should be tested at small scale first.
Get the pre-trade checklist.
We are turning these guides into a searchable checklist for checking terms, rules and risk before you trade.