DropForecast forecasts what upcoming collectibles will be worth after they drop — trading cards, sneakers, sealed product, exclusives — and grades every call it makes in public, including the ones it gets wrong.
DropForecast is built and run by Hayden Simerly, an active collectibles reseller based in Houston, Texas. Not a media company and not a marketplace: the models here were built to answer questions that came up while actually buying and selling this stuff — is this box worth preordering, and when should I sell what I already have.
That matters for one reason. Every forecast on this site is a bet the operator is also taking. There is no incentive to publish a number we do not believe.
The hobby news cycle, refreshed hourly. Two engines: sixteen publisher feeds read directly, plus an open-web sweep that runs a search for every product already on our board. Each story is typed as an event — preorder, delayed, restock, checklist, pricing, announced, release — and carries a date only when the source states one. We never infer a date to fill a field.
A 0–100 score for how hard a drop will cop and flip, built from five signal families: scarcity, demand heat, product intrinsics, market structure, and catalysts. The component scores are printed on the product's own page, so the number can be argued with rather than taken on faith.
Projects a resale multiplier for a drop before it happens, from public pre-drop signals — search interest, video coverage, breadth of retailer listings — run through a discriminator refit nightly on resolved outcomes. It outputs a multiplier, a COP/WATCH/PASS call, and a confidence out of ten. Low confidence is labelled low confidence.
A dated sell-by call for product you already hold, from a hype-decay curve: when the resale premium peaks, and what holding past that costs. Where a product has real price history in our comp index the call is fitted to that history. Where it does not, the call falls back to a category prior and is marked low confidence on the page — most current calls are in that state, and the page says so rather than dressing up a guess.
Everything is gathered from free, public sources under a self-imposed rate limit. Where a source blocks automated access, we say so on the page rather than working around it.
These are forecasts, and forecasts are wrong on a schedule. The honest limits, stated plainly:
Thin history is the big one. A product with no sold-comp history gets a category prior, which is a considerably weaker claim than a fitted curve. Those pages are labelled low confidence. Treat them as a starting point, not a signal.
Resale markets move on things no model sees. A reprint announcement, a licence change, an injury, a lawsuit. A dated call made before the news is a call made without it.
Correlation is labelled as correlation. The catalyst feed links dated events to comp movement. It does not prove the event caused the movement, and the page says so.
Nothing here is financial advice. Collectible resale carries real risk of loss, including total loss. Verify live comps before you buy or sell anything.
Every forecast that resolves gets graded in public on the track record page — hits and misses both, with the original call preserved. A forecasting site that only shows its wins is a marketing site. If our accuracy falls, that page is where you will see it first.
If a number here is wrong, we want to know and we will fix it. Reach us at the contact page or hello@dropforecast.com. Corrections to a published forecast are made on the page itself, with the original call left visible.
DropForecast is independent. We are not affiliated with, endorsed by, or sponsored by Topps, Panini, Upper Deck, Nike, Pokémon, Wizards of the Coast, LEGO, or any manufacturer or retailer named on this site. Product names and images belong to their respective owners and appear here for identification and commentary. Last updated September 4, 2026.
