WIN

Winner Selection Component

Selection Criteria

Which versioned selection algorithm the daily Select Winners job runs. This MUST be set. There is no default. While it is unset, Select Winners errors on execution rather than guessing an algorithm.

Selection Criteria Basic is the original pipeline: drop any symbol at or below the Sentiment Veto Threshold, sort the survivors by absolute forecast magnitude, apply the confidence cull with its fallback, and cap at Max Winners.

Selection Criteria 4P runs that same pipeline, then keeps rank 4 onward: the three highest-magnitude names are dropped and the rest of the list is taken. On a day whose list runs fewer than four deep this produces no winners, which is a normal no-buy day rather than an error.

Confidence cull fallback governs what happens when the Minimum Confidence Threshold cull would leave fewer than Max Winners candidates. On (default): the pre-cull top Max Winners is returned instead, so the list is never short. Off: the accuracy floor is strict — only symbols at or above the threshold are kept, even if that yields fewer than Max Winners winners (or none at all).

Sentiment Veto Threshold

Any symbol whose SAP sentiment score is at or below this value is dropped from the winners pool before the magnitude sort runs.

Minimum Confidence Threshold

The minimum MAG accuracy score a symbol must have to survive being filtered out by WIN. Note: There is a fallback mechanism in the selection algorithm such that when the cull would drop the candidate count below the configured number of winners, the pre-cull number of configured winners is returned instead to prevent empty lists.

%

Max Winners

Cap on the number of winners returned per run. Sets the size of the daily buy list BUY will execute on.