The screening problem
If you have ever used a stock screener, you know the workflow. You set some criteria—recent momentum, volume spike, news catalyst, whatever—and get a list of 200 names. Then what? You manually check each one, read the news, look at the chart, check the sentiment. By the time you are done with 50 names, the market is open and the fast moves have already happened.
The problem gets worse the more tickers you include. Scan the entire S&P 500 and you get hundreds of matches. Scan the full Alpaca tradeable universe—12,627 tickers including penny stocks, micro-caps, and options underlyings—and you might get thousands. At some point, the list becomes noise. You cannot act on 500 candidates.
Institutions solve this with a two-tier system. Tier 1 is fast: disqualify 95% of the universe with simple, high-speed filters. Momentum, volume, recent news, obvious disqualifications. Take the top 50 candidates from Tier 1. Then Tier 2 is deep: run the slow, expensive models on only those 50. Technical analysis, sentiment fusion, risk-rule checks. No top-20 trading list is sold.
GodFin does not sell predictions or signals. It checks orders against your own risk rules before they reach your broker.
Tier 1: The fast pass
Speed is the priority here. We need to score 12,627 tickers in under 10 minutes. The filters are simple and vectorized:
- Momentum score: Rolling 5-day return trend normalized to -1 (worst) to +1 (best)
- Volume ratio: Today's volume divided by 20-day average volume. We disqualify anything under 1.2x
- StockTwits sentiment: The previous day's StockTwits mention sentiment polarity, scored -1 to +1
- Intraday % change: Sorted descending, we filter for movers above 2% to eliminate sleepy names
- Liquidity gates: Minimum 500k shares and $5M average daily volume. Disqualifies illiquid micro-caps
- News freshness: Did this name get mentioned in any news source in the last 24 hours? Binary flag, helps catch catalysts
We combine these into a simple Tier 1 composite score:
tier1_score = (momentum × 0.35)
+ (min(vol_ratio, 5) / 5 × 0.25)
+ (stocktwits_sentiment × 0.20)
+ (pct_change_normalized × 0.20)
Result: top 50 candidates
This Tier 1 pass typically identifies 50–100 candidates worth deep diving. From our 12,627-ticker universe, that is already a 250x reduction. The rest of your morning focus can happen on only the truly interesting names.
Tier 2: Deep analysis on the finalists
Once we have top 50, we run the expensive models. These are slower but far more accurate:
- Technical analysis: RSI, MACD, Bollinger Band position, trend strength, support/resistance proximity
- Risk-rule checks: Our bias-corrected sequence model does not predict direction with confidence estimation
- Sentiment fusion: Reddit sentiment + news + Google Trends + FinViz + options PCR aggregated into single -1 to +1 score
- Options flow analysis: Unusual options activity, implied volatility skew, put/call flow imbalance
- Google Trends momentum: Spike detection and direction; high-volume searches often precede volatility expansion
These are checked against your risk rules:
risk_rule_check = (risk_rule_status × 0.50)
+ (sentiment_composite × 0.20)
+ (technical_score × 0.15)
+ (confidence_adjustment × 0.15)
No predictions, no signals.
What triggers an entry: the decision rule
Not every high-scoring candidate becomes a trade. Two additional gates must pass:
Gate 1: Signal strength. Tier 2 signal must be above +0.40 for a BUY or below -0.40 for a SELL. Signals between -0.40 and +0.40 are marked HOLD or WATCH. This eliminates ambiguous setups.
Gate 2: Confidence threshold. Risk-rule status (derived from risk-rule checks across our ensemble) must be above 60%. An order that fails a risk rule does not reach your broker. An order is checked against your risk rules before it reaches your broker.
Only orders that pass all risk rules reach your broker. On most mornings, that is 15–35 actionable candidates out of the original 12,627.
Real data from today's scan
To show what this looks like in practice, here are the top 10 Tier 1 candidates from the scan on March 26, 2026:
| Rank | Ticker | Momentum | StockTwits | Vol Ratio | % Change | Tags |
|---|---|---|---|---|---|---|
| 1 | NEXT | +0.978 | +0.890 | 2.20x | +11.04% | yahoo_trending |
| 2 | AIFF | +0.972 | +0.860 | 4.06x | +79.76% | yahoo_trending |
| 3 | MDGL | +0.966 | +0.830 | 4.44x | +12.30% | stocktwits, yahoo |
| 4 | EEIQ | +0.964 | +0.821 | 2267.49x | +209.16% | yahoo_trending |
| 5 | FCHL | +0.959 | +0.792 | 42.21x | +107.60% | yahoo_trending |
Notice the volume ratios. EEIQ has 2267x normal volume—that is not normal, that is a catalyst or a meme stock moment. FCHL is 42x normal. These are the kinds of explosive moves that create intraday trading opportunities, but only if you catch them in the pre-market scan and are ready at the open.
Now, not all of these make it to Tier 2. EEIQ's extreme volume ratio is a red flag—could be a reverse split, a short squeeze, or manipulation. Our Tier 2 gates would evaluate this more carefully: do the risk rules pass? Is sentiment supporting the move or hedging it? Is there real news or just noise? By the time the final watchlist is published, EEIQ might be marked WATCH instead of BUY—flagged as interesting but needing more due diligence before committing capital.
Why this gives retail traders institutional coverage
The math is simple: a human trader can deep-dive maybe 20 stocks before market open. GodFin analyzes 12,627. Even if the quality of analysis per stock is 10% lower than human-level due diligence, you are covering 600x more opportunities. Statistically, you will find the best setups.
More importantly, the two-tier system is brutally efficient about filtering noise. You only get alerts on names that pass both fast and slow gates. This dramatically reduces false positives compared to single-layer screeners. A casual momentum screener might alert you to 200 names a day. A two-tier system alerts you to 20 really good ones. That is actionable.
Institutional traders have been doing this for years. The difference is they have a team of 10 analysts and $10 million in infrastructure. GodFin does it for you, automatically, every morning, for the cost of a subscription.
What happens after the scan
GodFin does not sell a top-20 watchlist. Orders are checked against your own risk rules before they reach your broker.
You decide the size and timing. GodFin does not force trades. It forces decisions. Every day you get a pre-filtered list of the most interesting opportunities in the entire tradeable universe. What you do with that information is up to you.
One more thing: continuous monitoring
The scan is not a one-time event. Throughout the day, our system monitors risk-rule status. If sentiment turns, if risk rules fail, if volume breaks down, you get an alert. Orders that pass all risk rules reach your broker. The universe scanner is not just a morning tool. It is a continuous filter on 12,627 tickers, focused only on the ones worth your attention.
Read Next: How GodFin Reads Market Sentiment — Understand the five data sources that power Tier 2 deep analysis.