FREQAI / POLICY

FreqAI Prediction vs Trading Signal: Why They Are Not the Same

Why a machine-learning prediction is not automatically a trade and how policy, costs, exits and risk transform model output into trading results.

The model answers a narrower question

A model estimates a target from features. It does not automatically decide whether the expected move is large enough to trade after fees, how much capital to use or when to exit.

Policy converts prediction into action

Thresholds, confidence rules, market-state filters and no-trade logic determine whether a prediction becomes LONG, SHORT or FLAT behavior.

Execution changes the economics

Spread, fees, slippage and timing can turn statistically useful predictions into negative net expectancy. Model metrics and trading metrics therefore belong on separate dashboards.

Our evaluation rule

We judge the complete V6 experiment by realized PAPER outcomes and stability, while model behavior is diagnostic evidence rather than a profitability claim.

A practical validation checklist

Keep chronological boundaries, record the exact feature/target/model version, separate model metrics from trading metrics, and evaluate the frozen system on unseen forward data. A change made after observing the result belongs to a new experiment.

MODEL→POLICY→EXECUTION→NET RESULT

Continue