600+ liquid US names run through a four-condition trend gate every night. What survives is ranked by a single backtested Trend Score, and every row carries its 20-day range, ATR and volume structure so you can read the setup yourself. Nothing here is a prediction.
By Gexi · GEXDesk's digital analystRefreshes nightly after the closeA nightly pass over a universe of liquid names (price above $15, 20-day average volume above a million shares) that first applies a hard trend gate — price above the 21-day EMA and 50-day SMA, the 21 above the 50, the 50 above the 200 — and drops anything that fails. Roughly one name in six survives on a normal night.
What survives is ranked by one number: the Trend Score (60–100), how far price sits above its 21-day EMA and how far the 21 sits above the 50. It is the only input that held up when we tested each piece of the original eight-stage composite on five years of data, so it is the only thing that scores. Everything else on the card — range, ATR, volume, sector — is structure to read, not points.
Ranked by Trend Score inside the gate. Structure shown, not scored. Read, not predicted.
Each card shows the sector its name trades with (best-correlated ETF) and where that sector ranks tonight. Rank is context only — it didn't pass as a scoring input.
Standing rule here: no component scores until it has been tested on its own. We ran each piece of the original composite on 5 years × 625 names (208 weekly scoring dates, forward return vs SPY, t+1 entry, bootstrap confidence intervals). This is the result.
| Component | 10-day IC (95% CI) | Verdict | On the page as |
|---|---|---|---|
| Trend score (extension + stack) | +0.038 (+0.005 to +0.073) | Passed — ranks | The score |
| Relative strength (20/60d vs SPY, QQQ) | −0.013 (−0.043 to +0.016) | Not supported | 20d vs SPY, context |
| Sector rank | −0.002 (−0.033 to +0.029) | Not supported | Sector + rank, context |
| Volume vs 20-day | +0.014 (+0.002 to +0.026) | Weak at 10d only, not robust | Volume ratio, context |
| ATR expansion | −0.007 (−0.025 to +0.010) | Not supported | ATR14 / expansion, context |
| Breakout label | −0.023 (−0.044 to −0.004) | Negative — fresh breakouts lagged | Label with a caveat |
Inside the gate, the top 10 by Trend Score ran about +1.7% vs SPY over the following 21 trading days on average (95% CI +0.1% to +3.5%). That edge was concentrated in trending tape (Sep 2024 → Sep 2026) and was flat in the Sep 2022 → Sep 2024 chop — which is why the market-context strip sits above the list. The gate itself is a setup definition, not an edge: gated names as a group did not beat SPY. Full study: IMS component validation, 2026-09-18. Re-tested quarterly; components that later pass get added with a logged weight.
Pro members get every name that clears the gate — typically 50–90 — ranked by Trend Score with the same range, ATR and volume structure on each. Waitlist members get first access and locked-in early pricing.
GEX