Forensic screening
Short screener: methodology
How Screener ranks each day's biggest gainers as short candidates: impending dilution, toxic financing, share-count growth, reverse splits, NAV premium and pump-and-dump history, all scored from public SEC filings.
Every trading day, a handful of small companies jump 20%, 50%, or 100% on little or no news. Many of these moves fade. The ones that fade hardest usually have the same traits: a history of reverse splits, a steady stream of share offerings, warrants or convertibles about to unlock, and a stock price far above what the balance sheet supports. Screener is the desk's forensic short-side analyst. After the close, it ranks the day's biggest gainers as short (fade) candidates by checking those traits against public filings.
It is a research filter, not a trading signal. Its output is a ranked list of names worth a closer look, with every score linked back to the filing or headline behind it.
1. Build the universe
- Collect the day's top US gainers from several free sources. Each source's status is logged, so a broken feed is visible.
- Drop warrants, rights, units and SPAC shells.
- Re-check every candidate against its own daily price bar. Percent change, volume and price are recomputed, and stale quotes are dropped.
- Filters: up at least 20% on the day (falling back to 15%, then 10% on quiet days), at least 500,000 shares traded and $500,000 in dollar volume. At most 25 names.
2. Score the red flags
Each factor produces a sub-score between 0 and 1, which is multiplied by its weight. The total is ranked from most to least suspicious. The main factor families:
| Factor family | What it looks for | Source |
|---|---|---|
| Impending dilution (largest single weight) | Warrants, convertibles or preferred shares that unlock on a stated future date. Bigger, sooner and cheaper unlocks score higher, and "toxic" variable-price terms score highest | SEC filings (8-K, 10-Q/10-K, 424B, S-1/S-3) |
| Financing history | Registered offerings, ATM programs, registered directs, equity lines, unregistered sales and a fresh shelf the company could sell into the spike | SEC EDGAR filing index and full-text search |
| Share-count growth | Split-adjusted growth in shares outstanding over about two years (4x growth earns full weight) | SEC XBRL company facts |
| Reverse splits and listing stress | Recent and repeated reverse splits, a pending split vote, days spent under $1, exchange deficiency notices | Split history, proxies, 8-K Item 3.01 |
| NAV premium | Price versus tangible book value, negative equity, and a credit when the company trades below its net cash | SEC XBRL balance sheet |
| Pump-and-dump history | Past spike days (high at least 40% above the prior 7-day VWAP on 3x normal volume), how far they faded, and how fast they round-tripped | Split-adjusted daily bars |
| Catalyst quality | No news or weak news (partnerships, pivots, letters) adds to the score. A substantive catalyst (FDA approval, buyout, earnings beat, large contract) subtracts heavily | Headlines and same-day 8-K/6-K text |
| Size | Micro-caps under $50M score highest | Market data |
3. Show the evidence
For each name, the report lists every factor, the score breakdown, and a plain-English "why" with links to the exact SEC filings and headlines used. The point is that a human can check the work in minutes.
Known limits
- Balance sheets for these companies are often months old, and serial issuers raise cash after them. NAV figures are a starting point, not a verdict.
- Warrant and conversion terms are read from filing text on a best-effort basis. Always open the cited document.
- News classification is keyword-based and only sees headlines and same-day filings.
- Borrow availability and cost are not included. A name can look like a great short and still be impossible or expensive to borrow.
- Free data sources can be stale or wrong for micro-caps. The screener flags mismatches it can detect, but it cannot catch everything.
Screener does not publish a live list of names here. The methodology is shown so the process can be judged on its own terms.