Every call, including the bad ones.

This is not a return figure. It measures whether the position a tracked wallet opened resolved in their favour. It does not measure what you would have made, because that depends on when you entered, at what price, and at what size — three things this page cannot know.

Exits are excluded. When a leader sells, we record it but never score it. Whether an exit was “right” depends on their cost basis, which is not something we can see. Counting exits as wins would flatter the number, so they sit outside it entirely.

High-conviction signals only, and nothing else is filtered. The engine scores far more trades than it recommends; this record covers those scoring 65 or above, which is the set it presents as worth acting on. That threshold is applied to the score, which is fixed before anyone can know the outcome, so it cannot select for winners. Beyond it there is no filtering: a signal enters this record the moment it is generated and there is no step at which a bad call could be removed.

Void markets are excluded. A market that closes without paying out either side — cancelled, delisted, or never resolved — is recorded and then set aside. Counting those as losses would drag the number down with markets that resolved for nobody, which is as misleading as dropping the losers would be.

Small samples show no percentage. Below 30 settled signals, a band shows its raw counts and no rate. Four correct calls out of four is a 100% hit rate and tells you nothing.