How SneakerPulse numbers are made
Every price on this site is a recorded resale sale on GOAT or StockX, and a day's sales only count once that day has finished loading. Today that is Oct 28 on both marketplaces.
This page says where the data comes from, which sales are left out, how far behind the marketplaces our record runs, and which statistics we use. It is rebuilt every day; this copy is from November 5, 2026.
Where the sales come from
- Completed sales only. We record sales that closed on GOAT and StockX, through the kicks.dev data service, once a day. Asks, bids and listings are not prices and are never used as prices.
- Prices are the US view. StockX reports a sale made outside the US at the price a US viewer would see, so a few StockX prices include import costs the buyer paid elsewhere. GOAT sells new and used pairs and we cannot always tell them apart; StockX sells new pairs only.
- Our record is a sample of the market, not the whole of it. The feed covers more products over time, so a count of sales recorded is never a measure of market size, and we do not compare sales volume year over year as a demand measure. Since about September 1, 2026 the feed reports sales for fewer distinct listings than in August (checked at the source on October 24, 2026: the provider has no newer sales for most of the missing ones), so no page compares volumes or listing counts across that date.
- Retail prices come from StockX's listed retail price at release, refreshed daily for every listing with 50 or more sales. The catalog was last refreshed on Nov 5, 2026 and holds 48,377 listings with a retail price. Long-running shoes may be compared with an older, lower retail price; posts that use retail say so and check shoes released since 2024 separately.
The completeness gate
The marketplaces reach us late: the newest two or three days are still filling, and a shoe discovered today has its history loaded over the following days. So a day only counts in published volumes, shares and weekly movers once it passes our completeness check, which compares how many different shoes sold that day with what the same weekday a year earlier implies (scaled by Aug 10-Aug 23, 2026). Apparel is left out of the check. A shoe group's week counts only when the group's listings that week are close to the group's own peak; otherwise the page says the history is still loading. Same-shoe price comparisons may use the sales we hold on a filling day, with the sample size shown, but no volume trend is drawn over one. Over the last seven complete days the feed covered GOAT 92% and StockX 93% of the distinct shoes the same weekdays a year earlier imply.
How complete is the data today?
Sneaker sales on both marketplaces are loaded through Oct 28; the 8 newest days are still filling, so their counts may still rise.
Each bar is one day: how many different shoes sold that day, as a share of what the same weekday a year earlier implies (scaled by August 10-23, 2026 against the same weeks of 2025). Teal = loaded past our completeness check; grey = not yet: the newest days are still arriving from our data provider, and a day can stay grey for longer while its long tail loads. Apparel is left out. We only count a day in published sneaker volumes, shares and weekly movers once it passes. Rebuilt daily.
Weekly movers and the market line on every page use the latest complete week against the complete week before it, per marketplace and per day, and leave out fee-free promotion days, which the gate flags as spikes. The homepage and /trending name those days when they fall inside either week.
What is left out
- Apparel, trading cards, bags and collectibles are excluded from brand, model and market figures where the product category says so.
- Sales under $10 or over $5,000, and days where one shoe's feed misreported every sale at a fraction of its usual price, are skipped from movers and the price index.
- Football, soccer and baseball cleats never appear in recommendations, gift lists or tool results; the price checker still finds them.
- StockX sales carry no size, so size curves use GOAT sales only, and size mix inside a StockX shoe is not controlled.
The statistics we use
- Medians, not averages, with the band half of all pairs sold in (25th to 75th percentile), the number of sales and the window on every answer box.
- Same-shoe comparisons. A price change is each shoe's median against its own median in the earlier window, summarised as the median of those changes with a bootstrap 95% interval and an exact sign test on rose versus fell. Pooled medians mix colorways and are not used for change.
- Changes, not levels. Correlations between series use log changes, with p-values from the effective sample size of autocorrelated series, and every family of tests behind a headline is corrected with Benjamini-Hochberg; we call a result clear only at q below 0.05, and say plainly when it is null or the sample is too small.
- Shares carry Wilson 95% intervals, directional hit rates use exact binomial tests, and brand-versus-brand shares use two-proportion tests.
- Past sales, not a forecast. Nothing here is investment advice or a price prediction. Call It records the crowd's guess and grades it against the sales that follow; we never add a prediction of our own.
Corrections
When a later check changes a published figure, the post keeps its date, gains a dated "Updated" paragraph in its first screen with the old and new numbers, and keeps the original figures in its methodology note. Re-checked so far: 3,770 releases vs retail (unchanged), Jordan 11s vs retail and Converse vs retail (both corrected on October 30, 2026 after the retail catalog gained about a thousand listings), and the highest above retail list (October 29, 2026).
Download the data on the free data page (CC BY 4.0), read who makes it on About, or see this week's movers on Trending.