I track weekly resale prices for sneakers, consoles and collectibles at SoldSignal. This week I added a Hype Meter: a label per item saying whether resale demand is heating up, steady or cooling off.
The interesting part was not the label. It was deciding when not to give one.
The number behind the label
Most "hype" charts use asking prices, and asking prices are wishes. The meter uses one number per item per week:
what a seller actually keeps after fees, at the median sold price for a new item, on the marketplace the item's headline price comes from (GOAT or Mercari).
Then it compares that payout with last week:
| Change vs last week (after fees) | Label |
|---|---|
| +3% or more | Heating up |
| between -3% and +3% | Steady |
| -3% or more | Cooling off |
It also shows the after-fee premium over retail, so you can see "still +56% over retail, but falling".
The refusal rules
An item gets no label unless all of these hold:
- Sold prices in both weeks (asking-price items, like watches, are never rated)
- The same source and method both weeks (a GOAT median vs a Mercari median is not a trend)
- 10+ sales both weeks, including 10+ new-condition sales when Mercari is the source
- A verified retail price (from the catalog or a sourced launch price, never guessed)
- No source served stale data that week
This week that leaves 10 of 68 tracked items (20 of them were only added today). The other 58 appear on the page as "not rated" with the reason: "New this week", "Too few sales", "Asking prices only", "No verified retail price".
Week 47 result
- 0 heating up, 7 steady, 3 cooling off
- Biggest move: Air Jordan 11 'Space Jam' (2026), seller payout on GOAT -14.8% in one week ($431 to $367 after fees), still about +56% over its $235 retail
- Air Jordan 1 'Chicago Lost & Found' -4.1%, Travis Scott x Jordan 1 Low 'Mocha' -3.3%
- Pokemon Prismatic Evolutions ETB: steady (-0.5%), still about +96% over retail after fees
Full table: Hype Meter ยท weekly Hype Report
The core of it (Python)
BAND = 3.0 # % week-over-week, smaller = noise
MIN_N = 10 # sales needed in both weeks
def label(cur, prev):
reason = why_not(cur, prev) # the 5 rules above
if reason:
return None, reason
now, before = after_fee(cur), after_fee(prev)
change = (now / before - 1) * 100
kind = "Heating up" if change >= BAND else "Cooling off" if change <= -BAND else "Steady"
return kind, round(change, 1)
after_fee() applies each marketplace's 2026 seller fees (published at /data/fees.json).
Free JSON feed
No key, CORS open, CC BY 4.0:
curl -s https://www.aiplannermail.com/api/v1/hype.json | jq '.fallers[] | {name, label, weekChangePct, afterFeePayout}'
Each item has label, weekChangePct, afterFeePayout (and last week's), premiumAfterFeesPct, sampleSize, plus a notLabelled list with reasons.
Limits
- One week is noisy. The 3% band hides most of it, not all of it.
- GOAT's number is the median of the last sale in each size, men's 7 to 13, not a full transaction log.
- It describes past sales. It is not a forecast and not advice.
20 more items were added today (Jordan 4 Black Cat 2025, Kobe 6 Reverse Grinch, New Balance 9060/550/2002R, Pop Mart Hirono/Skullpanda/Crybaby...). They get a label after next Monday, once there are two weeks to compare.
Top comments (0)