Your edge before
the bell.
AI scans 12,000+ stocks at 6 AM — 3.5 hours before the bell. 21 signals. 80.3% accuracy. Strong Buy to Watch, every trading day.
What a scan looks like
| # | Ticker | Score | Gap | Expected | Result |
|---|---|---|---|---|---|
| 1 | NVTX | 50.97 | +19.6% | $62.28 | +12.5% |
| 2 | VELO | 57.81 | +4.5% | $12.92 | +25.5% |
| 3 | LABX | 48.23 | +15.1% | $48.63 | +2.4% |
What powers Prismo
Nine systems working together, every trading day.
Gradient Boosted Trees with 21 engineered features. 80.3% out-of-sample accuracy. Self-learning — retrains weekly on fresh ground truth.
Institutional momentum curves with 6 multipliers. Computes expected price, stop-loss, targets, and re-entry zones for every alert.
6 AM scan, 3.5 hours before the bell. Scores yesterday's movers for continuation probability. HOT / WARM / MONITOR signals.
Strong Buy, Buy, Hold, Watch — computed from 6 factors including score, expected gain, risk/reward, volume, and ticker history.
Every prediction is tracked against actual results. Per-ticker, per-signal breakdown. 71.5% win rate across 414 validated trades.
4 scheduled jobs run automatically Mon-Fri. Scan, predict, track, learn. No manual intervention needed. Skips weekends and holidays.
How it works
Scan
12,000+ tickers ingested daily. Volume profiles, gap analysis, momentum signals.
Score
21 features through Gradient Boosted Trees. Each weighted by predictive power.
Predict
Expected price, stop-loss, targets. Institutional momentum curves.
Learn
Every result labels training data. Retrains weekly. Gets smarter with every trade.
21 signals
Every edge counted. From core gap-up analysis to sector-wide rotation detection.
Under the hood
How Surgepulse Prismo identifies explosive stocks before the crowd — from raw market data to actionable trade plans.
Scan Engine
Every trading morning, Surgepulse Prismo pulls grouped daily bars for 12,000+ US equities via Polygon.io. The engine compares the latest trading day against the prior day to detect gap-ups — stocks that opened significantly higher than the previous close.
An automatic fallback chain tries three data strategies: all-tickers snapshot, gainers endpoint, or grouped daily comparison — ensuring scans run reliably regardless of API tier.
Scoring Algorithm
Each candidate is scored 0-100 using a hybrid engine: a rule-based static scorer provides the baseline, while a Gradient Boosted Trees ML model overlays probability-weighted predictions. As the model proves itself, its influence increases from 0% to 70% via an adaptive alpha ramp.
The static scorer adds bonus points for volume-gap synergy (when both gap AND volume confirm), dollar volume (institutional interest), 52W high breakouts, and sector-wide momentum.
Self-Learning Engine
After market close, the accuracy tracker fetches actual closing prices and labels every alert as a win or loss. These labeled FeatureVectors (21 features each) become training data for the ML model.
The model auto-upgrades as data grows: Logistic Regression at 50 samples, Gradient Boosted Trees at 200, and an Ensemble (LR + GBT) at 500+. Weekly retraining on Fridays. Currently at 414 samples with 80.3% out-of-sample accuracy.
Prediction Engine
Expected price isn't a guess — it's computed from institutional-grade momentum curves. Decades of market data show distinct gap-and-go patterns: 5-10% gaps have ~55% follow-through, 10-20% gaps have ~65% with +8-15% avg continuation.
Six multipliers refine each prediction: Volume confirmation (the #1 signal), dollar volume (institutional interest), float rotation (squeeze potential), ATR-relative gap (abnormality score), ticker history, and score-based confidence.
Key indicators driving predictions
Opening price vs previous close. The primary catalyst signal — large gaps with volume rarely fade.
Absolute stock price filters liquidity. $2-$500 range, sweet spot at $5-$200 for tradeable small/mid caps.
Gap size divided by average daily range. A 10% gap on a 2% ATR stock = 5x abnormality = highly significant.
Yesterday's high-low range as % of close. Higher volatility = stock is "in play" and more likely to continue.
5-day avg volume / 20-day avg volume. Rising volume trend confirms institutional accumulation building.
Current volume vs 20-day average. RVOL > 2x = confirmed move. > 5x = extreme event, highest continuation rate.
Stop-Loss
Calculated from ATR (Average True Range) — at least 1.5x daily range below entry, capped at 8%. Respects the stock's natural volatility instead of using arbitrary fixed percentages.
Target Price
Two targets: conservative (expected close from momentum curves) and stretch (1.8x expected gain, capped at gap-range upside). Self-calibrates against historical actuals as data accumulates.
Re-Entry Zone
40% pullback of the gap from opening price. If a stock gaps up $10 and pulls back $4, that's the second-chance entry. Momentum stocks that hold re-entry zones typically resume the move.
Plans
Frequently asked questions
What exactly does Surgepulse Prismo do?
How accurate are the predictions?
Is this financial advice?
How does the AI learn and improve?
What data source do you use?
What stocks does Prismo focus on?
What's the difference between plans?
Can I get alerts on my phone?
The next surge is forming.
71.5% win rate. 414 validated trades. Start detecting surges today.
Create Account