The record, kept honestly.
Every published pick is written down before kickoff with its timestamp, its price, its book and the model probability behind it. The table rejects a row dated after kickoff, and rejects any edit to one already written.
—
Published picks
none running
—
Record
n = 0
—
Units
nothing settled yet
—
Closing line value
n = 0
Nothing published yet
Vigil publishes a pick only from a market that has already beaten the closing line over a measured sample. No market has yet cleared that gate, so there is nothing on this page — and there will be nothing until one does.
This is the state the product is designed to be honest about. A record page with numbers on it before any pick had been published would tell you nothing except that we were willing to make numbers up.
What is and is not on this page
- Every pick published, before kickoff, at the price and book that were available. The database rejects a row dated after kickoff, and rejects any edit to one already written.
- Losers and pushes, in the same table and the same totals as the winners.
- Every rate with its sample size — and no rate at all where there is no sample, because an unmeasured figure and a zero look identical and are not.
- No backtests, no paper trading, no simulated results. They are not mixed into these numbers and they are not shown beside them — a record is only evidence if it is one kind of thing.