Pokémon Set Value by Era: 142 Sets Measured

Group every English set we price by the era it belongs to and the money is not where the noise is. Across 142 sets carrying both a recorded release date and at least ten priced cards, 17,699 cards in all, the median set from the WotC years of 1999 to 2003 is worth $2,357 with a median card at $3.62, against $643 and $0.27 for the median Sword and Shield set. This is a level, not a return: it is a one-day cross section of asks, and the second half of this piece measures what those era cohorts have been doing.

Set value here is the sum of the last recorded market price of every English card we price in that set. It is a floor rather than a valuation of the set: cards we do not price contribute nothing, and asks are not sales. The median card column is the more robust half of the table, because it is unaffected by how many cards a set has.

The median card falls from $3.62 in the WotC years to $0.20 and $0.19 in the current two eras, a factor of about eighteen, with one step back up at Sun and Moon. The median set value does not follow it down in a straight line, because modern sets are much larger: the Scarlet and Violet cohort averages 176 priced cards a set against 104 for the WotC cohort, and a long tail at twenty cents adds up. The EX row is the era with the dearest median set on the board, at $5,505, which is a level statement about fourteen sets and not a claim about how the era trades. Concentration is flatter across eras than the folklore suggests, with a range of 47.7% to 66.3% in the top five cards; the deeper cross-era treatment of that number is in What 56,000 Cards Say About How Sets Hold Value and its subset companion The Set Inside the Set.

Levels and moves need different populations. Moves come from the recorded set-value tape, restricted to the 100 sets whose series runs continuously from January 27, 2025 to September 2026 with no adjacent step above 15%, the same continuity filter used in Do Pokémon Sets Gain Value With Age?. That filter is much harsher on old sets than on new ones, so the vintage rows below carry three sets and should be treated as anecdotes with arithmetic attached. The modern rows carry twenty-two and seventeen.

The last column is the cleanest split in this study. Pre-2020 eras are sitting on their own recorded highs: 85% to 100% of HGSS, Black and White, Diamond and Pearl and Sun and Moon sets are within one percent of the best value their series has printed. From Sword and Shield onward that collapses to 9% and 12%. Whatever has happened over these twenty months, it happened to sets released before 2020 and it has not happened to sets released after. This is a set-level drawdown measurement; the card-level drawdown census is a separate study with a separate population, and the two will not produce the same number.

The level map falls if a broader pricing pass changes the median card materially in either direction for an era. We price 101 of the 102 cards in Base Set and 271 cards in Cosmic Eclipse, so coverage is not the binding constraint on the big expansions, but it is on promo bins and boxed products. The move map falls if the pre-2020 cohorts stop sitting on their highs while the post-2020 cohorts stay underwater, which would mean the split we measured was a phase rather than a structure. And if a future window shows the Scarlet and Violet cohort recovering to its highs without the older cohorts giving anything back, the framing of this study as an era effect is wrong and it should be rewritten as a supply-cycle effect.

The two halves use different populations and cannot be combined into one number: 142 sets for levels, 100 for moves, with partial overlap. Set value is a sum over the cards we price, so it is a lower bound and it is not comparable to a sealed product price or to a completed-set cost. Every price is an ask. The era boundaries are our own, drawn on release dates rather than on the publisher's series names, and a set released within weeks of a boundary could sit either side. The EX and WotC move rows carry one and three sets respectively, which is not enough to support a claim about those eras and is printed only so the table is not silently truncated. Japanese sets are excluded throughout; the catalog holds 38,965 Japanese cards but only a handful carry a price series long enough to join this window.

Levels: every English catalog set with at least ten cards carrying a last recorded market price and a release date resolved from `report:spread:setsmeta`, 142 sets and 17,699 cards, priced from the daily `price_history` tape as of 2026-09-22, which is the same figure the card pages publish. One catalog row is kept per TCGplayer product id, which drops 240 English rows that are claimed twice (trainer-kit halves, and Trainer Gallery cards carried by both a subset and its parent). Set value is the sum of those prices; median card is the median of them, upper middle on even samples; top-5 share is the five dearest priced cards over the set total. Moves: the continuous cohort described in the linked age study, `sethist2:` series anchored at the first reading between 2025-01-27 and 2025-02-10 and closed at the latest reading on or after 2026-09-01, with sets excluded when any adjacent pair of readings differs by more than 15%. Distance below the recorded high is the latest reading over the maximum reading dated on or after the set's release date; sets whose series begins after release have no pre-release readings to exclude, so for them every reading is post-release by construction and the maximum is a recorded high rather than an all-time high. Era buckets are assigned by release date with the boundaries printed in the tables. Not financial advice.