The first time I went looking for ideas in quarterly filings I did the obvious thing and screened across every filer for names being bought. The output was garbage, and it took me a while to work out why. The screen was not broken. The population was. The overwhelming majority of firms filing these reports are not trying to beat anything. They are wealth advisors and registered advisory firms allocating client money across model portfolios, and their filings are a description of a business, not of a view.
They still file. They still show up as "institutional buying." And because there are so many of them relative to the handful of firms actually picking stocks, any screen that treats every filer equally is mostly measuring how advisory money got allocated last quarter.
What the typical filer is actually doing
Picture the firm behind a filing you have never heard of. It manages money for a few hundred families. It runs three or four model portfolios, mostly built from index funds and sector funds, with a sleeve of large-cap names that everybody already owns. When a client deposits money, the models buy everything in proportion. When the models drift, they rebalance. Nobody at that firm has a thesis on any individual holding.
That firm's quarterly filing looks busy. Hundreds of lines, dozens of changes, several new positions. Run it through a screen and it produces buy signals all day. Read it as a human and you can see in about thirty seconds that it is a book being administered, not managed.
The reason this matters for a retail screen specifically is dilution. You are looking for the small number of filings where somebody made a hard decision. Every advisory book in the population adds hundreds of rows of noise on top, and because advisory books tend to own the same well-known names, the noise is correlated. Your screen will keep telling you that institutions are buying the biggest companies in the market, which is true, uninteresting, and has nothing to do with anybody's view.
Four checks that take a minute
You do not need a classifier. Open one filing and look at four things, in this order.
- Count the rows. A book with several hundred positions is not being picked. Nobody has a view on four hundred companies at once. A book with twenty to eighty lines is a book where somebody chose each one.
- Look at the top ten weights. Add up the ten largest positions as a share of the total reported value. If those ten are a third of the book or more, someone is betting. If the largest position is under two percent, nothing in the book is a bet, because nothing in it can move the result.
- Count the fund and trust lines. Broad index products, sector funds and bond funds are the fingerprint of an allocation business. A book that is mostly wrappers is telling you the firm's job is asset allocation, and their individual stock lines are decoration.
- Compare two consecutive quarters and see whether every line moved in the same direction by roughly the same percentage. That pattern means money came in or went out and the book was scaled. It is not a hundred separate decisions, it is one deposit.
Any single check can be misleading. A concentrated book can still be a passive holding company. A large book can belong to a genuine manager running a broad quantitative strategy. But a filing that fails all four is not worth another second, and most of them fail all four.
The names that still fail the test
Being well known does not get a filer past this. Three categories catch people out.
Holding companies and permanent-capital vehicles look wonderfully concentrated and turn over almost nothing, which is exactly what you want to see. The problem is horizon. A position they have held for a decade is not an idea you can act on this week, and their reported book usually reflects decisions made years ago at prices that no longer exist.
Insurance companies file a large equity book that behaves like an index, because the actual business risk sits in instruments that never appear on the form. The equity sleeve is a byproduct.
Family offices are the trickiest, because some are genuinely concentrated and active and some are wealth-management operations with a nicer name. The four checks separate them properly, which is the point of running the checks rather than reading the letterhead.
What a followable filer looks like
The profile that survives is fairly specific. Somewhere between twenty and eighty positions. A top ten that is a third or more of the book. Few or no fund wrappers. Positions that move independently of each other quarter to quarter, so that some lines are up while others are down, which is the signature of decisions rather than flows. Names outside the largest handful of companies in the market, because a book made entirely of mega-caps is a book you could replicate with an index fund and lower fees.
You will end up with a small list. That is correct. A follow list of eight to twelve firms is manageable and a list of two hundred is a list you will stop reading by March.
The version of this to run this week
Pick the five filers whose names you would repeat to a friend as investors worth watching. Pull one filing from each. Run the four checks. My honest expectation is that two of the five will fail, and the specific way they fail will tell you something you did not know about a firm you thought you understood.
Then keep the survivors and set them aside until the next filing lands, because the list is the durable part of this exercise and the individual positions are not. A filing is at least forty-five days old when you read it, and a position established early in the quarter can be four and a half months old, so nothing in it is an entry.
What it is good for is deciding which names get an hour of your attention. That is where the pairing matters. Insider Alpha carries institutional holdings alongside the Form 4 feed as cross-reference material, and the reason to run them together is that they fail in opposite directions. A quarterly filing tells you a lot about who owns something and almost nothing about when. A Form 4 has a transaction date on it and tells you nothing about the size of anybody's conviction. A name that a genuinely concentrated manager just opened, where officers are also buying on dates you can see, is a name worth the hour. A name that a four-hundred-line advisory book topped up by eleven percent is not, no matter how the headline is written.