Methodology, stated plainly

Prior Moves is an independent research publication. It predicts the next stock each tracked investor is likely to buy, about six weeks before their SEC 13F filing makes it public. This page is the one to read if you are deciding whether to trust the rest of the site, or to cite it.

Who we track, and why these investors

We track 44 investors across five markets. The US roster is the set of discretionary long-horizon managers whose 13F filings carry real conviction: value and quality investors, activists, concentrated stock-pickers. We deliberately exclude high-turnover quant market-makers whose filings are thousands of churning positions, because their disclosures say little about what they believe. Beyond the US we cover Japan (EDINET large-shareholding, filed within five business days), the UK (FCA, two trading days at a 3% threshold), Korea (OpenDART), and the Netherlands (AFM). A US-only tracker cannot see Buffett’s Tokyo trading houses, the Murakami funds, Oasis, or Cevian. We can.

How the prediction works

There is a separate machine-learning model for each investor, trained only on that investor’s own filing history plus roughly 290 public signals per stock per quarter: SEC filings, insider and congressional trades, activist stakes, fundamentals, price behaviour, and news tone. Each model estimates the probability that its investor adds a given name as a new position next quarter. Training is walk-forward: every quarter is predicted using only data from strictly earlier quarters, with no look-ahead. A leakage check confirms it: when we shuffle the labels, the model’s accuracy collapses to a coin flip, which is what should happen if the signal is real and not leaked.

The honest numbers

Across 47 complete quarters (2014 to 2026), the top-15 consensus basket beat the S&P 500 by about 1.8 percentage points per quarter, net of costs. It beat the index in 26 of those 47 quarters. And the 95% confidence interval on that edge includes zero: the t-statistic is about 1.45, so this is a small, directional research signal, not a proven edge. One honest caveat: the edge is stronger in the model’s earlier years and weaker in the most recent quarters, so the full-window figure leans on that early performance. A separate factor attribution shows the edge is idiosyncratic rather than a hidden bet on the market, size, value, or momentum. We publish the losing quarters at full strength and the confidence interval on the same line as the headline, because a number that only survives when you hide its error bars is not worth showing.

What this is not

It is not investment advice, and Prior Moves is not a registered investment adviser. It is a general-circulation research publication: one impersonal model portfolio per investor, identical for every reader, the way model-portfolio newsletters have worked for decades. We never execute trades, never hold your money, and never tailor anything to your personal situation. You decide and place everything yourself at your own broker. A predicted probability is a statement about a filing pattern, never a claim that the investor has bought a name, or will, or that you should.

Who runs this

Prior Moves is built by a solo founder in Tokyo, bootstrapped. The core research is free and stays free. If it becomes a business, it will be through a paid alerts tier and a data feed, never by charging for the predictions themselves or by touching anyone’s money.

Verify it yourself: the live boards are on top picks, the running accuracy is on the prediction leaderboard and track record, the score components are on how it works, and the calibration of the probabilities is on calibration. Forward quarter: 2026-06-30. Research, not investment advice, see the full disclaimer.