How it works

A mirror is one model portfolio

For each of 58 famous investors we publish a single impersonal model portfolio, their latest disclosed 13F holdings plus the stocks a model predicts they will buy next. It is identical for every reader. We never execute trades, hold assets, or tailor anything to your account.

Predicting the next buy

13F filings disclose what each institution held at the end of a quarter, about six weeks after the fact. The model learns each investor’s historical buying pattern and ranks a broad ~2,000-name candidate universe, known at the rebalance date, by the probability that the name becomes a new buy in the next filing. The ranking is the PriorScore. The published page never changes that ranking after the fact.

Tested leak-free

Every backtest is walk-forward: at each historical rebalance the model only sees data available on that date, holds the top predicted names equal-weight for one quarter, and scores complete quarters only, gross of costs. As-of joins are availability-lagged so no future information leaks in.

What the accuracy number means

Accuracy is measured per investor as AUC: the chance the model ranks a name the investor actually bought above one they did not. It is not one number, and we refuse to quote only the best one. Across the roster it spans about 0.47 to 0.96, and that spread is the honest story:

Median near 0.65 for the names people actually follow (value and concentrated long books). That is roughly three times better than a coin flip at the next-buy task. This is the real, modest edge and the number we lead with.
0.87 to 0.96 for the most systematic funds (Citadel, Millennium, the quants). High, but it reflects breadth, not foresight: these books hold hundreds of names, so ranking a new entry against their own giant book is an easier task, and the lift on a live broad universe is near zero. We never present 0.96 as our headline.
0.47 to 0.50 for a few highly concentrated discretionary investors (Ackman, Einhorn). Near or below chance. Their handful of high conviction bets are driven by private research the model cannot see, so it cannot predict them. We say so on their pages rather than hide it.

So the path to the number is plain: a separate model per investor, trained walk-forward on 218 public features per stock per quarter, calibrated on 72,268 out-of-sample calls, with a label-shuffle canary that collapses it to a coin flip to prove there is no look-ahead leak. The honest claim is a median 0.65, not a single big percentage.

The honest result

The consensus mirror's backtest shows +1.8 pts/quarter net vs the S&P 500 over 47 complete quarters (2014–2026), directional, NOT statistically significant (t≈1.4; 95% CI includes zero). The edge is directional but not statistically significant, at t = 1.45 over 47 quarters the confidence interval includes zero. We publish it anyway, labeled honestly, because hiding a weak result would be the dishonest choice.

A research publication

Prior Moves is a research publication: one impersonal model portfolio per investor, identical for every reader. You place any trades yourself at your own broker. No execution, no custody, no individualised advice. You hold the trades; Prior Moves publishes the playbook. Nothing here is investment advice. Read the full disclaimer.