Are they right?
682 convergences · 463 macro · 219 single names
Every time three or more of the investors we track landed on the same call, we wrote it down and priced what happened next — entering at the following session's open and measuring against SPY over the same window. The ledger below is every one of those events. Read the median next to every average: the averages here are carried by a short tail of large winners, not by being right consistently.
Returns are the instrument minus SPY over the window, entered at the next session's open. A filled dot after a name means they said they hold it; a plain name argued the view. A bearish call is scored as a short. The Source column is the appearance and quote behind each name — it is what The Brief carries.
The distribution
Where the matching events actually landed against SPY. The shape is the point: a dense middle around zero with a long right tail is a lottery ticket, not an edge — and it is what this data looks like at every setting.
Every combination we pre-registered
The same arithmetic on fixed filters, shown against SPY with the median and sample size on every cell. It ships whole, in a fixed order, because the best-looking cell in a grid this size is what noise produces — picking one out and calling it a finding is the mistake this table exists to avoid.
| Filter | 5 sessions | 20 sessions | 60 sessions | 120 sessions |
|---|---|---|---|---|
| Any 2+ voices | +0.12% med +0.00% · n=682 | +0.11% med +0.00% · n=658 | +0.51% med +0.00% · n=559 | +3.12% med +0.00% · n=473 |
| Any 3+ voices | +0.16% med +0.00% · n=254 | +0.46% med +0.00% · n=242 | +1.51% med +0.00% · n=201 | +5.34% med +1.77% · n=160 |
| Any 4+ voices | +0.27% med +0.00% · n=53 | +0.70% med +0.00% · n=46 | +3.68% med +0.00% · n=39 | +4.38% med +3.19% · n=30 |
| 3+ voices, 2+ money managers | -0.29% med +0.00% · n=121 | -0.66% med +0.00% · n=117 | +0.93% med +0.00% · n=97 | +4.12% med +0.00% · n=75 |
| 3+ voices, at least one holds a position | +1.29% med +0.75% · n=47 | +2.11% med +1.29% · n=42 | +0.88% med +0.00% · n=33 | +6.04% med +1.33% · n=24 |
| 3+ voices, macro vehicles | +0.16% med +0.00% · n=200 | +0.66% med +0.00% · n=190 | +1.75% med +0.00% · n=165 | +5.28% med +0.54% · n=134 |
| 3+ voices, single names | +0.16% med -0.04% · n=54 | -0.24% med +1.69% · n=52 | +0.44% med -0.56% · n=36 | +5.65% med +3.82% · n=26 |
Backward-looking. Each row is what happened after events matching that filter, entered at the next session's open and held for the stated number of sessions, measured against SPY over the same window. It is not a strategy, a forecast, or a claim that any of it continues to work. Read the n beside every number, and read the median beside every average: the median event is flat against SPY at every horizon, so the averages are carried by a small number of large winners rather than by being right consistently.
How an event is defined. Three or more distinct investors making the same directional call on the same instrument inside a ten-day window, de-duplicated so one long argument does not count repeatedly. Entry is the next session's open after the event — the day we could first have acted on it, not the close we detected it at.
“Price going in” is the axis worth playing with. How far the name had already moved in the month before they spoke, adjusted for direction so a short on something that just ran counts the same as a buy on something that just fell. Chasing and fading look different in this data, and the buckets are thin: a thing to look at, not a thing we claim.
Why 60 sessions is the default. These are people running money against an index over long horizons. Scoring them on a five-day move would be scoring them on a question they are not playing.
Not advice. Nothing here is a recommendation, and none of it is a claim that any combination continues to work.