What is NDCG@30 for stock ranking?

NDCG@30 (Normalized Discounted Cumulative Gain at rank 30) scores how well the top thirty names in a ranked list align with realized outcomes for that horizon. It rewards putting better-performing names higher. When fewer than 30 names were evaluated, TradeWell reports NDCG@20 instead.

Why does TradeWell use NDCG@30 instead of win rate?

Win-rate and hit-rate percentages are easy to invent and easy to misread. NDCG@30 measures ranking quality at the depth users actually see. TradeWell does not invent or publish fabricated win-rate, hit-rate, or IC percentages as product claims on marketing pages.

How does walk-forward testing relate to ranking accuracy?

Live rankers are scored with NDCG@30 (or NDCG@20 when the evaluated list is shorter than 30). Research architectures use walk-forward out-of-sample testing before any promotion to the public board. Past model performance does not guarantee future results.

Is TradeWell investment advice?

No. TradeWell publishes research scores and forecasts, not investment advice. Nothing on the site or in the app is a recommendation to buy or sell any security.

Stock ranking accuracy: how TradeWell uses NDCG@30

TradeWell publishes a separate TW Quant Score, expected return, and model-implied price target for each of 2,800+ equities on each of ten horizons (1D, 3D, 1W, 2W, 1M, 3M, 6M, 1Y, 3Y, 5Y). Ranking quality matters more than a single invented win rate — so live rankers are evaluated with NDCG@30.

Get the iOS app Open the web app

What NDCG@30 measures

NDCG@30 (Normalized Discounted Cumulative Gain at rank 30) scores how well the top thirty names in a list align with realized outcomes for that horizon. Better-performing names higher in the list raise the score; burying them lowers it. When the evaluated list is shorter than 30 names, TradeWell reports NDCG@20 instead. NDCG@10 may still appear as a secondary field, but it is not the headline metric.

Live rankers vs research models

Live public rankers: NDCG@30 (floor NDCG@20 when the list is shorter than 30).
Research architectures: walk-forward out-of-sample testing before any promotion to the public board.
Full methodology: how TW Quant Scores are built.
Measurement detail: how TradeWell accuracy is measured.

Why we refuse fabricated win rates

TradeWell does not invent or publish fabricated win-rate, hit-rate, or IC percentages as product claims on marketing pages. Ranking metrics like NDCG describe list quality for a horizon; they are not a promise that any single name will rise. Past model performance does not guarantee future results.

Related research

Research, not investment advice. Nothing on this site or in the TradeWell app is a recommendation to buy or sell any security. Scores and forecasts are model estimates for research use. Past model performance does not guarantee future results.