What the scorecard's split means, and what it doesn't

The scorecard scores disagreements only. When a market price and the Scopes Fair Value diverge past the flag threshold, the flag is recorded and tracked to resolution, and it resolves to whichever of the two numbers proved closer to what happened. The split is the running tally of those verdicts.

So the split is a measure of whose probabilities have been closer to reality at the moments the two disagreed. It is a calibration record, not a tally of trades, and not a score for either institution.

A reasonable question follows immediately: “If the market side wins six times in ten, couldn’t one simply take the market’s side of every divergence and come out ahead?”

The answer is no, and the reason is structural: a price can only be transacted at the price.

Consider an invented case: Team A versus Team B, no real game. An event contract on Team A is priced at 60 cents (a 60% implied probability). The independent benchmark reads 50%. That is a ten-point divergence, so the flag fires. Suppose the market’s 60% turns out to be the accurate probability, which is exactly what “the market was right” means when the flag resolves.

Taking the market’s side means paying 60 cents for a contract that pays $1 if Team A wins and nothing if it does not:

The expected value is (0.60 × +0.40) + (0.40 × −0.60) = 0, exactly zero, before exchange fees. Taking the other side at 40 cents is identically zero under the same assumption.

The arithmetic is the same at any price: whenever a price equals the true probability, every position at that price has zero expected value before fees.

An accurate price pays nobody. The market’s accuracy is not a reward available to its followers, it is the price of admission.

Everyone knows the stronger team is stronger. Nobody profits from knowing it, because the line already says so. Rewards attach to being more right than the price, not to being right.

Flags fire at moments of disagreement, and those are disproportionately moments when one side has just absorbed newer information. A flagged divergence has, across the record so far, more often proven to be information than error. That is a statement about when divergences occur, not an invitation to act on them, for the arithmetic above.

So far, that side has more often been the market's.

Profitability would require a systematic gap between prices and outcome frequencies, for instance, contracts priced at 40 cents resolving true far more than 40% of the time, again and again. The calibration page tests exactly this.

The record to date shows prices broadly earning their probabilities, which is what a well-functioning market should produce, and what an independent scorekeeper exists to verify.