Scanner Methodology
Legend AI is built on the frameworks of traders who made fortunes using price and volume. Every scan, every score, every filter maps back to a documented principle from Minervini, O'Neil, Livermore, Weinstein, or Darvas β not a black box, not a generic screener.
The core premise: Most profitable swing trades share the same pre-breakout profile β a leading stock in a leading sector, constructing a tight, low-volume base, with relative strength already building before the move. Legend AI finds that profile every morning so you can focus on execution, not screening.
01 β Trend Template Gatekeeper
Every symbol passes through Minervini's 8-criteria Trend Template before any pattern analysis runs. This ensures we only evaluate stocks already in a Stage 2 uptrend β the only stage where breakouts have positive expectancy.
The 8 Criteria (0β8 score, β₯6 required to advance)
- Price is above the 150-day and 200-day simple moving averages
- The 150-day SMA is above the 200-day SMA
- The 200-day SMA has been trending up for at least 1 month
- The 50-day SMA is above both the 150-day and 200-day SMAs
- Price is above the 50-day SMA
- Price is at least 25% above its 52-week low
- Price is within 25% of its 52-week high
- Relative Strength (vs S&P 500) is in the top 70th percentile or better
Stocks scoring 7β8 are considered "perfect alignment." Stocks scoring 5β6 may appear as Watch candidates. Anything below 5 is excluded entirely regardless of pattern quality.
02 β Relative Strength Leadership
O'Neil's research showed that the best breakout stocks typically have an RS Rating of 80 or above before the breakout occurs. Legend AI uses a 12-month RS Score (0β99) computed against the S&P 500.
RS Score β₯ 80 β Market Leader
Stock is outperforming 80%+ of all scanned symbols on a 12-month price-return basis. O'Neil called this the single most reliable leading indicator for big winners.
RS New High Filter
A stock whose RS line makes a new 52-week high before price does is flagged as a power candidate β the RS line confirming institutional accumulation before the price breakout.
Industry Group Rank
37% of a stock's move comes from its industry group (O'Neil). Each symbol is ranked within its industry β only stocks in the top quartile of their group qualify for full scoring.
RS Rising Signal
Beyond the absolute RS score, we track the direction: is RS accelerating? A stock with RS 75 and rising is often more actionable than RS 85 and flat.
03 β Structural Pattern Recognition
Pattern detection runs on 6β12 months of daily OHLCV data. Each pattern is named after the legendary trader who identified it, not generic TA jargon. The scanner looks for constructive price structure β tight, controlled bases on declining volume that set up explosive breakouts.
Each pattern has specific geometric criteria: base depth, duration, contraction % between pivots, volume profile on the right side, and distance to the structural pivot point (entry trigger).
04 β Readiness: Actionable vs Watch
A well-formed base that is 15% below the pivot is not the same as one that is 2% away. Readiness scoring separates setups by proximity and condition quality.
Actionable
Price is within 5% of the pivot. Volume pattern is constructive (declining volume on base, expansion on right side). Market regime is not in full Distribution. Setup Score β₯ 60. This is a setup ready to trade β entry trigger is near.
Watch
Base is forming correctly but pivot is 5β20% away, or volume confirmation is incomplete. Add to watchlist, set price alerts, and let it develop. Forcing an early entry on a Watch setup is a common error.
Avoid
Trend Template score below threshold, base is too deep or too long, extended more than 5% past pivot, or earnings within 5 trading days. Avoid does not mean the stock is bad β it means conditions are not aligned for a controlled entry now.
05 β The Setup Score (0β100)
A single composite number summarizing six independent signals. It is not a buy signal β it is a quality ranking for comparison within the current scan batch. Think of it as the SEPA checklist distilled to a single score.
Score thresholds: β₯75 = High Conviction Β· 50β74 = Watch Β· <50 = Low priority
06 β Market Regime Awareness
No individual setup exists in isolation. O'Neil's research showed that 3 out of 4 stocks follow the general market. Legend AI tracks the market regime in real time and reduces exposure signals accordingly.
Distribution Day Count
A distribution day is a significant down day on volume higher than the prior session. β₯5 distribution days in 25 sessions triggers a defensive posture β scores are dampened and an "Exposure Restricted" warning appears on the dashboard.
Confirmed Uptrend / Rally Attempt
The scanner uses SPY price vs its 50/200-day SMAs and the distribution day count to classify the current environment. Only in a Confirmed Uptrend does the scanner give full green light. In rally attempts, Watch setups are deprioritized.
07 β Risk Management Framework
Every Actionable setup includes a structural stop and a risk/reward estimate. These are not arbitrary percentages β they derive from the geometry of the base itself.
Stop Placement
The default stop is placed at the most recent structural support level within the base β typically the low of the tightest contraction or the pivot area low. This is not a flat 8% stop. The stop should be just below the level where the thesis is invalidated.
Risk/Reward Ratio
Minimum acceptable R:R is 2:1 β the target must be at least twice the distance to the stop. Setups with R:R below 1.5 are downgraded. Targets are based on the measured move from the base depth or prior resistance level, not arbitrary price projections.
Earnings Proximity Warning
Setups within 5 trading days of an earnings report are flagged with a β warning. Holding through earnings destroys the controlled-risk premise of the setup β a gap down can bypass any stop. The scanner does not recommend avoiding the stock, but it does flag the timing risk clearly.
08 β Data Sources & Scan Frequency
The scan runs once daily after market close, processing ~495 symbols including US equities, ETFs, and 20 major crypto assets. All prices are end-of-day; no intraday data is used in pattern detection.
Universe: ~495 Symbols
Curated to liquid, tradeable names across three asset classes: US equities (S&P 500 + Nasdaq 100), major ETFs (SPY, QQQ, sector ETFs), and 20 large-cap crypto assets (BTC, ETH, SOL and more). Chart patterns β VCP, Cup & Handle, Bull Flag, Darvas Box, and others β apply equally to crypto. CANSLIM fundamental scoring is not applied to crypto (no earnings data); stage/trend/RS analysis applies to all asset classes.
Scan Timing
Full scan completes within 15β30 minutes of market close each trading day. Results are available on the dashboard by 4:30 PM ET on standard sessions. Results persist until the next scan overwrites them.
This methodology is applied systematically every trading day. No discretion, no cherry-picking β every setup in the results table passed every filter above.
View Today's Setups β