The rule · fixed on 1 October 2026
How a stock pick is graded.
One rule, written down before any pick existed. A pick opens at a closing price, is frozen, closes on the first of three events, and counts as beating the market only if it did better than the S&P 500 over the same days.
-
Step 1
A medium-or-higher conviction thesis opens a pick
A pick opens at the close of the first trading day on which a published dossier states its thesis at medium, high or supreme conviction, from 1 October 2026, as long as that close is still above the thesis's own kill line. A stock has at most one open pick at a time. The conviction, the kill line and the thesis are frozen at that close.
Low conviction is research coverage, not a pick.
-
Step 2
The pick is frozen
The conviction, the kill line and the thesis text are fixed at the opening close and published with the pick. Later research can change the dossier; it cannot change an open pick.
The kill line is the price that would prove the thesis wrong, written by the model before the pick opens.
-
Step 3
It closes on the first of three events
A pick closes at the first of: a close through the kill line frozen at open (daily or weekly, as the thesis states; a weekly close is the last trading day of the week); conviction cut below medium; or the 63rd trading day after it opened. Once a pick has closed, the next time the research states a fresh medium-or-higher thesis on that stock, a new pick opens.
-
Step 4
It is graded against the S&P 500
A closed pick beat the market if its total return, dividends included, was higher than the S&P 500's (SPY) over exactly the same days; otherwise it trailed. Every eligible pick is in the record.
Returns use split- and dividend-adjusted closes for the stock and for SPY, from the opening close to the closing close.
An example
Illustrative numbers, not a real pick.
A dossier moves to high conviction with a kill line at a weekly close below $90. The pick opens at that day's close of $100. Six weeks later the stock closes a week at $88: the pick closes there, 12% down. Over the same six weeks SPY returned 2%, so the pick trailed the S&P 500 by 14 percentage points.
Had the stock instead finished its 63rd session at $109 while SPY rose 4%, it would have closed on the horizon and beaten the S&P 500 by 5 points.
Common questions
- Why grade against the S&P 500 rather than on gains alone?
- A stock that rises 5% while the market rises 8% was a worse choice than an index fund. Comparing each pick with SPY over exactly the same days measures the research, not the market.
- Why 63 sessions?
- 63 trading sessions is about three months, long enough for a thesis built on the next earnings report or product cycle to play through, short enough that every pick is graded within a quarter.
- What happens when the research changes its mind?
- If conviction is cut below medium the pick closes at that session's close and is graded like any other. Once a pick has closed, the next medium-or-higher thesis on the same stock opens a new pick.
- Are any picks left out?
- No. Every eligible pick from 1 October 2026 on is in the record, and no pick is removed for how it turned out.
- What about picks before October 2026?
- The record starts on 1 October 2026 under this rule. Earlier research used a different method and is not part of this record.