The simulated portfolio at The Five, which we introduced last week, answers one question: if you had rebalanced annually into the five most-undervalued names in our coverage universe, what would you have earned? It is a useful question, and the answer is on the page. It is also a slightly misleading question, because almost no patient owner actually rebalances on a calendar. A more honest question is: what did Mr. Market pick on this specific day, and what have those names done since? The Cohorts, which we are publishing today, answers exactly that question, on every quarterly and annual creation date going back five years.

A “cohort,” in our terminology, is a frozen portfolio. The constituents are the top N names in our universe — ranked by Δ% to intrinsic value — on a specific historical day. The portfolio is never rebalanced. The buyer of, say, the Q1 2024 Top 5cohort bought five businesses on the first day of January 2024 and has held them since, equal-weighted, full stop. We do not update the constituents when the protocol's reckoning changes. The cohort is what was picked, locked in.

The buyer of the Q1 2024 Top 5 bought five businesses on the first day of January 2024 and has held them since, equal-weighted, full stop.

We publish three sizes — Top 5, Top 10, Top 20 — at every cohort date, in both quarterly and annual cadences. That gives the reader a way to ask several questions at once:

• Does concentration help or hurt? (Top 5 vs Top 20 at the same date.)
• Does rebalancing cadence matter? (A Q1 cohort with a year of forward performance vs the annual cohort starting the same January.)
• How stable is the protocol's view? (Do the same names appear across consecutive quarters?)
• Does the protocol's read of value persist? (Do the cohorts created during the 2022 sell-off, when prices were depressed across the board, hold up better than the cohorts created at the 2024 highs?)

We do not, here, claim to have answered any of these questions. We have, instead, published the data. The Cohorts page indexes every vintage with its size, cadence, inception date, total return since inception, S&P 500 return over the same period, and the spread. Click into any one and the page shows the full constituent list — rank, entry price, entry intrinsic value, entry Δ%, current price, return since entry, contribution to the portfolio. The biggest winners and the biggest losers are highlighted. There is no flinching from the bad ones.

A note on what the Cohorts are not. They are not buy lists. The cohort from January of 2022, for example, includes a name or two that the protocol now flags as fairly priced or even slightly overpriced; the constituents have been held through the change in mood. They are also not predictions about the future — the vintage from last quarter has had three months of forward performance, which is not enough data to draw any conclusions about. They are, simply, the record. We publish the record because a publication that does not is worth less than one that does.

The archive will grow by three vintages every quarter (Top 5, Top 10, Top 20 at the new quarterly date) and three more every January (the annual cohorts at the new year). When subscribers come online — which is soon — the archive will get a calibration toggle that lets you see the protocol with sector premiums set differently, with the discount rate raised or lowered, with the margin of safety tightened. For now, it is the canonical protocol, applied identically to every cohort, on every date, no exceptions.

Open The Cohorts and have a look. The data is the data.

— The editors