The Hot & Cold Index is one 0–100 reading of how warm the whole Pokémon card market is: how fast the market is moving against its own recent pace, how many cards are above their 30-day average, and how many are breaking to new highs. Below it, the same temperature for each era group, and the sets moving most over the trailing 30 days — what the reading is actually made of.
Below 50: more of the market is falling than rising.
Neutral — the warmest of the four era groups right now.
The index last crossed 50 going down — a cooling turn.
Market temperature
Cool on Sep 17, 2026, smoothed over 5 days.
365 daily readings since Sep 18, 2025.
The sets moving most in each direction over the window — the moves the index above is made of.
The index is a weighted sum of three 0–100 measurements — where the all-cards market composite sits in its 90-day range (40%), the share of priced cards above their own 30-day average (40%), and new 90-day highs minus new lows (20%) — then smoothed over five days. Its bands are fixed constants; the board’s are not.
Every board figure is our own evaluated raw, near-mint, English price for each set (or card), compared between the first and last day of the trailing 30-day window. The temperature band is cut at THIS cohort's own percentiles on every build — never a fixed "up 5% is hot" rule.
A row needs at least 3 priced instruments to appear by default, and carries a Weak badge below 5 — a real reading on a thin basket, never a fabricated confidence figure. The full method below.
The index is precomputed WEEKLY (Sunday night) by research:hot-cold-index — evidence as of 2026-09-17.
The board is precomputed nightly by research:hot-cold — evidence as of 2026-09-27, built 2026-09-27T22:33:59.989Z.
One 0–100 reading of how warm the whole raw, near-mint Pokémon card market is on a given day, computed daily from our own price history. It reads 44 — Cool — as of 2026-09-17. Above 50 more cards are rising than falling; below 50 more are falling.
Trend (40%): the all-cards market composite's 30-day rate of change, placed in the range it covered over the last 90 days. Breadth (40%): the share of priced cards above their own 30-day average. Breakouts (20%): new 90-day highs minus new lows.
Cold is below 30; Cool is 30–45; Neutral is 45–55; Warm is 55–70; Hot is 70 and above. The bands are fixed constants calibrated against the index's own two-year history, so today's 62 means what last year's 62 meant.
A 5-day exponential moving average. The raw reading jumps two or three points on a day when the price guide reprices a large block of cards, which is a fact about the guide's schedule rather than about the market. Both the smoothed reading and the raw components are published.
No, and that is the point of the era strip: the same formula runs on each era group's own composite and its own priced cards. Modern is warmest right now at 48. A market-wide reading can hide a Vintage rally happening inside a flat Modern market.
Which sets moved most over the trailing 30 days — what the index is made of. Its colours are cut at that cohort's own percentiles, so they rank sets against each other, not against the index's fixed scale. Rows need 3 priced instruments; under 5 is flagged weak. 194 sets clear it as of 2026-09-27.
One 0–100 reading of how warm the whole raw, near-mint, English market is, computed for every calendar day since 2024-02-08 — the day our price history hands over from a monthly figure to a daily one, and therefore the first day a 30-day moving average means anything. It is a weighted sum of three measurements, each of which is itself a 0–100 number, and each of which a reader can check independently:
The three are then smoothed with a 5-day exponential moving average. The raw reading moves two or three points on a day when the price guide reprices a large block of cards, and a temperature nobody can act on daily should not flicker daily. Both the smoothed reading and the three raw components are published, which is what the strip under the chart shows.
Because the obvious version does not work, and we measured it. Reading the composite's LEVEL against its own 90-day range says "the market is at a three-month high", which on this market was true on most days between 2024 and 2026 — card prices on the TCGplayer guide rose almost monotonically. That component pinned at 100 and put 58% of all days in the top band, which is a scale that cannot tell you anything. Ranking how FAST the market is moving against how fast it has recently moved removes the trend without inventing a model of it.
The cut points are constants, not percentiles recomputed nightly: a temperature whose meaning moved every night could not be compared to last month's. They were checked against the index's own two-year history so that each band holds a meaningful share of the days the market has actually had — Cold below 30, Cool 30–45, Neutral 45–55, Warm 55–70, Hot 70 and above. Over 888 daily readings to September 2026 those five bands held 4%, 28%, 17%, 36% and 15% of days, and the index crossed 50 nineteen times.
Same index, same formula, over each group's own composite and its own priced cards. The right-hand figure is the change over seven days; the tick is 50.
The index is rebuilt weekly, Sunday night — a 90-day range and a 30-day average do not change overnight, and computing them across the whole catalogue is this platform's heaviest read. The board below is rebuilt nightly. Both come from precomputed tables. Evidence as of 2026-09-27.
Weekly, on Sunday night. The one statement behind breadth and breakouts is a 30-day moving average and a 90-day high and low for every priced card in the catalogue — the heaviest single read this platform makes — and what it measures does not become a different number overnight. The movers board underneath is rebuilt nightly. Neither is ever computed while you wait: the page reads a precomputed table.
A component with no evidence on a given day is absent, not zero and not a neutral 50: its weight is redistributed over the components that do have evidence. A day where the whole market had fewer than 200 priced cards behind a reading has no temperature at all, and the line draws a gap rather than an interpolation. The same rule runs the four era-group readings, on each group's own composite and its own cards.
For every set (or, on the card toggle, every card printing) the board reads the raw, near-mint, English price on the first and last day of the trailing 30-day window, from our own evaluated prices — TCGplayer's market guide first, the sold low/high midpoint or the sold-median bridge where the guide is silent, never a delivered or a graded price. The percent change between those two levels is the figure the board ranks on; a set with no price on both days has no figure at all and is left out rather than guessed.
Its Cold/Cool/Neutral/Warm/Hot colours are a DIFFERENT scale from the index's, and deliberately so: they are cut at this cohort's own 10th, 25th, 75th and 90th percentiles, recomputed on every build, so the board always shows which sets are hot RELATIVE TO EACH OTHER. The index answers how warm the market is; the board answers which parts of it.
"Priced instruments" counts the distinct raw/near-mint/English instruments with a usable price somewhere in the window. The board's default filter hides anything under 3; a row that clears the filter but sits under 5 carries a Weak badge instead of being hidden or assigned a fabricated confidence number — the figure is real, the basket behind it is thin, and both facts are shown.
The index is a reading of the PRICE GUIDE's view of the market, not of executed sales: a guide that reprices a block of cards on one day moves breadth on that day whether or not anything traded. Breadth and breakouts are unweighted — a five-dollar common counts as much as a five-hundred-dollar chase card — which is why the trend component reads a composite instead. A single large sale can move a thin basket's average on the board below more than it moves that set's real market. Auction and fixed-price listings are never mixed into one floor, and a vanished listing is never counted as a sale.
Research classification generated from observable market data with stated confidence and assumptions. Not personalized financial advice.