Methodology
How every number on this site is computed. For a running log of changes, see Updates.
What this site is
13F Research surfaces stocks that may warrant further research by aggregating the quarterly disclosed positions of a curated universe of prominent hedge funds and institutional investors. The premise: when several unrelated, well-resourced managers independently converge on the same stock in the same quarter, that convergence is worth knowing about — especially if the price hasn't already run away.
It is a starting point for research, not a substitute for it. Nothing here is investment advice, and no signal on this site is a recommendation to buy or sell anything.
The core signal: convergence
Every quarter, institutional managers with over $100M in US equities must disclose their long positions on SEC Form 13F-HR, due 45 days after quarter-end (most funds file on the deadline). We parse the filings of a curated universe of funds — value, macro, activist, long/short, and tech/growth managers — and compare each fund's positions to its prior quarter. A small number of firms are tracked for their 13D/13G filings only and excluded from the convergence signals: either their 13F no longer reflects a live portfolio, or their book spans thousands of largely market-making positions that would dilute a signal built on multi-fund agreement. The Funds page lists both tiers.
A long signal requires 3 or more funds buying the same stock in the same quarter — a new position or any share-count increase. A short signal requires 3+ funds each reducing a position by more than 50% or exiting entirely. Stocks with both a long and a short signal appear in Mixed Signals — divergent conviction, shown separately.
Share classes of the same company (e.g. GOOG/GOOGL) are merged into one signal — a fund holding both classes counts once, and whether it is buying or selling is judged on its combined position across classes, not class by class. When a company re-issues its stock under a new identifier with no trade behind it — a change of domicile, or depositary receipts converted into ordinary shares — we compare against the old line, so holders are not counted as having sold one security and bought another. Bonds a fund reports on its 13F — mostly convertible notes — and warrants, rights, and SPAC units are left out entirely, both from the signals and from fund holdings: a convertible is frequently held against a short position in the same company, and warrants are leveraged instruments we treat like options, so neither says much about the fund's view of the shares. ETFs are excluded entirely: funds hold them for beta and hedging, not stock selection.
Conviction score
Each buying fund contributes to the stock's score:
- Base: 1.0 per fund
- New position: +0.5
- Large add (shares more than doubled): +0.25
- × size multiplier: 0.5 + min(position %, 3) / 3— from 0.5× for a tiny sliver up to 1.5× when the position is 3%+ of the fund's equity book
The size multiplier is the heart of the score: a manager putting 4% of their portfolio into a name is making a categorically different statement than a multi-manager platform holding a 0.03% position. Each fund's "% of Fund" is shown on the stock detail pages.
Short signals are scored the same way (base 1.0 × size multiplier), weighted by the fund's prior-quarter position — dumping a large position is a louder exit signal than trimming a small one. Long Ideas with a score of 4.0 or higher appear in the main table; lower-scoring signals move to the collapsed Watchlist.
The price filter
13F data is stale by design — holdings are as of quarter-end, published up to 45 days later. The price filter separates ideas where the market has already reacted: a long signal passes while the stock has outperformed SPY by lessthan 10 points since quarter-end; a short signal passes while it has underperformed by less than 10 points. Signals that fail move to the "Already Moved" sections — the thesis may be intact, but the easy move likely already happened.
Prices refresh daily, so stocks migrate between sections as returns change. A stock that was Already Moved and pulls back into range gets a ↩ badge for 14 days — the funds were right once, and the price has reverted.
Corroborating signals
Five independent data layers are shown alongside the core signal. None of them affect the conviction score — they are context for your own research, each linked to its primary source on the detail pages.
Insider activity — open-market buys and sells by officers and directors, from SEC Form 4 filings since quarter-end. An insider buying alongside institutional accumulation is a different fact pattern than insiders selling into it. Refreshed daily; each row links to the filing on EDGAR.
Beneficial ownership (13D/13G) — filed when any investor crosses 5% ownership. SC 13D signals intent to influence control (activists) and drives the purple 13D badge; SC 13G is passive accumulation, shown for context only. Refreshed weekly.
Congressional trading— House and Senate members' stock transactions disclosed under the STOCK Act, sourced directly from each chamber's official disclosure system. Amounts are disclosed as ranges, not exact figures. The Congress column counts distinct members, not trades, who bought or sold since the quarter being shown began, placed by trade date rather than filing date. Members can file months late, and one member reporting a whole new portfolio would otherwise mark dozens of stocks at once. Older trades still appear on each stock's page, listed separately. Refreshed weekly; rows link to the original PTR filing.
Short interest— FINRA's consolidated biweekly short interest: the latest days-to-cover, and the change in shares sold short since the quarter-end report, the same starting point as the price filter, insider trades and news. Rising short interest alongside hedge fund buying can set up a squeeze; alongside hedge fund selling it corroborates the exit thesis (marked with a ✓ on Short Ideas). Checked daily; FINRA publishes roughly twice a month.
News — AI-generated summaries of coverage since quarter-end, with sentiment and red flags. Informational only, refreshed weekly.
Fund flows (Trends)
The Trends page measures how much the tracked funds bought and sold between two quarter ends, by stock, by investment theme, and by fund. It includes only funds that filed for both quarters, so a fund joining the universe never shows up as buying its whole portfolio.
Net buying, not value change.A position's value moves with both trading and price. We value the change in shares at the midpoint of the two quarter-end prices, so a holding that simply rose in price does not count as buying. Whatever value change is left over is reported as the price effect.
New listings are separated. Positions in a company that first traded during the quarter, such as an IPO, spin-off or direct listing, are reported separately. A first line in a new listing may be an allocation or a pre-IPO stake converting into public shares, not a purchase in the market.
Stock splits.A split multiplies every holder's share count without any trade. We detect splits from the filings themselves: the implied price per share and the share counts held move together by a clean ratio. Each split is confirmed against the split record, and the prior quarter's shares are restated onto the post-split basis before any comparison. Reported values are never adjusted, because a split does not change what a position is worth. The same adjustment applies to signals and fund pages.
Themes and returns.Stocks are grouped into investment themes drawn from what the funds themselves concentrate in. Assignments are fixed once made, so a published quarter's theme totals never shift. The return column runs from quarter end and updates daily until the next quarter closes. After that it is frozen, so each archived quarter shows how its buying and selling did over the following three months. A theme's return is weighted by the funds' combined holdings at quarter end.
Each quarter's figures are computed once when its filings are complete and are not revised afterward, like the signals.
Data freshness
| Data | Source | Refresh |
|---|---|---|
| Holdings & signals | SEC 13F-HR | Quarterly (45-day filing lag) |
| Prices & returns | Yahoo Finance | Daily (weekdays) |
| Insider trades | SEC Form 4 | Daily (weekdays) |
| Beneficial ownership | SEC SC 13D/13G | Weekly |
| Congressional trades | House Clerk / Senate eFD | Weekly |
| Short interest | FINRA | Checked daily; published ~2×/month |
| Fund flows (Trends) | SEC 13F-HR | Quarterly; returns daily until the next quarter end |
| News summaries | Web search + AI summary | Weekly |
Exclusions & limitations
- 13F filings only cover US-listed long positions — short books, swaps, and non-US holdings are invisible.
- Managers below $100M AUM are exempt from filing; positions can change materially between quarter-end and the filing date.
- Options positions (calls/puts) are tracked on a separate tab with a lower 2-fund threshold — 13F reports notional share equivalents, not capital at risk, so they never mix with equity signals.
- ETFs are excluded from equity signals.
- Congressional amounts are ranges; Senate filing links require accepting the Senate's access agreement.
- This site is informational only and is not investment advice.
Methodology changes are logged on the Updates page, including any that retroactively affect existing ratings.