Our track record: we say it, then we keep score

How calls work

A call names a set, a direction, a threshold and a deadline: "this set's index at or above 250.00 by December 9". It is published with the index value of the day it was made, and it cannot be edited afterwards.

On the deadline the call resolves against the published index — no judgement, no rounding in our favour. Hits and misses both stay on this page forever.

Two exceptions, both made in the open. On September 24, 2026 we found that our set indexes had been counting sealed products — boxes, packs, cases — next to single cards, contrary to our own methodology. We rebuilt every index from single cards only. Open calls that still met the rule on the corrected index were restated, with the original value kept next to the new one; the rest were withdrawn and are listed above with the reason.

On September 25, 2026 we found two more problems. A set's index restarted at 100 after a day with no data, which drew jumps that never happened; and some calls had been made on "set indexes" that rested on one to four cards — the price of a card or two, not a set. We fixed the calculation and rebuilt every index. The one open call touched by a restart was restated; calls made on indexes of fewer than five cards were withdrawn, each with the reason. New calls need at least 20 priced cards. Withdrawn calls do not count towards the hit rate.

How calls are chosen

A set becomes a candidate when its index moves more than 15% away from its own 90-day average, and only if that index rests on at least 20 priced cards on every day of the window — below that it is the price of a card or two, not a set. The call then follows the move rather than betting against it.

That rule was measured, not guessed. Across 868,735 historical pairs of "this day versus the same set 90 days later", on sets with at least 20 priced cards, following the move resolved correctly 73.2% of the time upward and 76.7% downward. Betting on a return to the average — the intuitive choice — came out at 23–27%, worse than a coin flip. For reference, simply calling "everything goes up" on the same data scores 58.0%. Set indexes count single cards only; sealed products are tracked separately.

None of that guarantees the next call. Card prices also move on reprints, tournaments and anime releases, and none of those live in our data.