Foundations EV Backtest: Why Calculators Disagree, and How Ours Held Up to a Real Opening

Box EV is a computed number, not a guess. For Foundations Play Booster Displays, SpellBook Finance starts with Wizards' published Play Booster odds, reads MTGJSON's raw config and sheet weights, prices each exact printing, and sums the expected value contribution from every priced card.

The key audit unit is a single card contribution. If a card is worth $32.48 and the distribution expects about 0.349 copies per 36-pack display, that card contributes about $11.34 to display EV. The same arithmetic is repeated for every pullable priced card.

Expected copies are not set by rarity alone. They come from each printing's weight on the real MTGJSON print sheets, so two mythics need not match. Doubling Season sits on the Foundations rare and mythic sheet one slot heavier than a typical mythic, weight 8 of 1,120 instead of 7, so it appears about 14 percent more often per display. Muldrotha and Progenitus are the only other Foundations mythics on that heavier slot. That is why a card can out-contribute a more expensive card of the same rarity: the EV follows the actual sheet weights, not the sticker price.

Most of the gap between EV calculators is not a factual dispute. It is a choice about how you sell. The same Foundations box is worth very different amounts depending on whether you dump everything to a buylist, sell the meaningful cards yourself, or something in between. Here is our own number at each of those exits.

The second question is whether the model resembles a real opening. This cohort used every public Foundations card row from the implied-box inventory, excluding precon-open rows that were not Play Booster output. Read it honestly: only 97 of 5,847 cards have sold, only 78 of those have matched sale-disposition proceeds, and the remaining 5,746 live cards are valued at current market. So this is mostly a mark to market, not realized cash. The realized portion is broken out first.

The box cost was $127.49. The live market price was $155.17. SpellBook Finance full EV was $245.76. The actual card result was $268.07 per implied box opened.

That strong cost multiple is partly acquisition edge. The boxes averaged 17.8% below the $155.17 market price on 2026-06-25, because the buy came from hunting the EV rankings instead of paying whatever a random sealed listing asked. If every opened box is marked at that market price, full EV is still $245.76 per box, or 1.58x market. The observed card result is $268.07 per box, or 1.73x market.

A fair objection is that market value is not cash. So we also modeled non-realized cards at 96% of market and then applied marketplace fees. Under the conservative version, the fee hits every dollar, including already-realized sales. At a 15% fee rate, the cohort still shows $1070.97 profit, 42% net margin, and 72.4% ROI on implied-box cost.

We used the production Foundations inventory cohort: every card row with set code FDN, plus Special Guests collector numbers 74 to 83, excluding 72 precon-open rows because those were not Play Booster Display output. Each row was priced against the latest production card price for its finish. Matched sold rows were counted as realized sales, sold rows without matched sale-disposition proceeds were valued at current market, unsold rows were counted as remaining card value, and lost rows were surfaced separately.

The implied boxes opened count is derived from card rows divided by a Play Booster Display footprint: 36 packs times 14 playable cards, or 504 cards per display. The owner has 13 Foundations boxes, but not all are opened, so the card-count-implied 11.60 boxes is the right denominator for this backtest.

EV values come from the live Foundations Play Booster Display row on SpellBook Finance: full EV, land-adjusted EV, and vendor-floor EV. The chart above links the result to market price and actual cost because those answer different questions.

It proves that this real Foundations opening landed close to the published full EV and nearly on top of land-adjusted EV, while remaining profitable after ordinary selling friction. It does not prove every box wins. Variance still exists, stale card prices can move, and listed inventory is not cash until it sells.

The useful conclusion is narrower and stronger: the EV estimate was not a hand-wavy number. It was close enough to audit against a real card list, and the published assumptions are visible enough for a skeptical reader to inspect.

Most of the result is unrealized. Only 97 of 5,847 cards have sold, and only 78 have matched sale-disposition proceeds; the rest is valued at current market using the same prices that feed the EV, so the close match is partly expected. Realized cash so far is 0.25x cost. The honest claim is narrow: the modeled EV is auditable, conservative on coverage, and consistent with how this cohort is selling, not a promise of cash in hand.

If the math made you want cardboard instead of caveats, the SpellBook Shopify store is over here. No variance proof required at checkout.