Who trades how
Each fund's 13F book placed by turnover (how much of the book changes in a year) and concentration (share of the book in its 10 largest companies). Bubble size is the 13F book. Patient, concentrated owners sit top-left; fast, diversified books sit bottom-right. Click a fund to open its page.
Turnover vs concentration
All 25 funds
Latest 13F. Click a column to sort, a row to open the fund.13F book
$bn, each quarterTop-10 share of book
each quarterTurnover per year
lesser of buys and sells ÷ bookSector mix by year
Profile
Latest fast signals
13D, 13G, Form 3/4Holdings
20 largest companiesBiggest long-run holdings
ranked by weight × quarters heldFast signals
Stake and insider filings by the 25 funds. Activist 13Ds and Form 4 trades arrive weeks to months before the matching 13F; passive 13Gs usually arrive with it. "Head start" is the days between this filing and the 13F that covers the same quarter.
Filings per month
Congress trades
Stock trades reported by House and Senate members (STOCK Act reports), grouped by the month they were filed, which is when the public could first see them. Blue = more buying than selling, red = more selling. Counts are used instead of dollars because amounts are reported as ranges and a few large trades dominate.
Net buying by sector
Does copying Congress pay?
Buy − sell spread: the return vs SPY of stocks members bought minus stocks they sold, bought the day after the filingMost active members
Recent trades
Lag & flows
What the 45-day 13F delay costs someone copying these funds, and whether the funds' combined sector moves point to next quarter's winning sectors. All returns include dividends and are compared with SPY over the same dates.
Copycat vs no-lag clone vs SPY
Growth of $1, all eligible funds equally weighted
Return per year by style group
By fund
Click a column to sort, a row to open the fund.These 25 funds were picked knowing how they did, which flatters every copycat-vs-SPY number. Use the table to choose which books to follow, and 13D and Form 4 filings for timing. Funds with fewer than 12 usable quarters are dimmed: too short to judge.
Sector flows
Context, not a trigger: these flows have barely predicted next quarter's sector returns (see the test below). Breadth = (funds adding − funds cutting) ÷ funds. Flow = average change in sector weight beyond price moves, with patient funds counting more. Crowding = average weight vs its own last 12 quarters (z-score).
Do flows predict next quarter's sector returns?
Rank correlation (IC) between each signal and sector return minus SPYIC runs from −1 to 1, and 0 means no skill. |t| ≥ 2 is shown in bold. The "from quarter-end" rows show what the 45-day wait costs at the sector level; they can't be traded.
Insider buying
Open-market buys and sales that company officers, directors and 10% owners report on SEC Form 4, usually within two business days, so this is the earliest of our sources. The one signal worth keeping is the cluster buy: three or more different officers or directors buying the same stock within 30 days. Sector and market-wide insider buying have shown no usable edge, but they tend to lead the funds' 13F moves.
Recent cluster buys
Clean leaves out new filers (IPOs, spin-offs and newly reporting foreign companies, which have lost money afterwards) and shells, funds and unlisted companies. These have no forward returns yet: they are the list to forward-test, not buy signals. Amber rows: one of our 25 funds is among the buyers. Hover the insider count for names.
Large holders and funds buying
10% owners, companies and funds (including our 25) buying in the last 90 days, apart from management clusters; largest firstStrategic holders and investors, not management: worth knowing, but not the cluster signal. "New listing" = the buy came within 60 days of the stock's first Yahoo price, usually an IPO allocation.
Net buying by sector
Insiders always buy banks, energy and REITs and sell tech, so raw breadth is red almost everywhere; compare each sector with its own past. Buying peaks in the open windows after earnings (Mar, May, Aug, Nov). Hover a cell for the counts.
Does copying insiders pay?
Clean cluster buys since 2013, bought the trading day after the signal; alpha = return beyond the stock's own normal (beta to SPY)Survivor bias: Yahoo has no delisted stocks, so companies that failed drop out and flatter every number here (only about 64% of clusters are priced). The headline is the typical stock (log alpha); the equal-weight basket gains less.
t is clustered by signal month; |t| ≥ 2 is shown in bold. "Beat own normal" = share of clusters with positive 3-month alpha. Left-out groups are dimmed.