Diamonds didn't get cheaper this season. The premium on being a good Diamond got cheaper, and it got cheaper to almost exactly zero.

That sounds like the same sentence twice. It isn't, and the difference is worth about 90,000 stubs if you were holding Ketel Marte in March.

The number that starts the argument

Across the first two weeks of the season, the middle half of the 85–87 Live Series tier ran from about 3,150 stubs at the 25th percentile to 13,800 at the 75th. A good 85 cost 4.2× a bad one.

Across the last two weeks: 3,384 to 4,386. 1.25×.

The floor barely moved — 3,148 to 3,384, if anything slightly up. Everything above it fell down onto it.

Those are two-week medians, not single days, and that's deliberate: a one-day endpoint in a 35-card tier is a coin flip. Take the single days instead and you get 4.21× on March 17 and 1.34× on August 8 — same story, noisier telling.

Every Live Series 85–87 card priced that day. The median falls, but the <em>fan-in</em> is the story — the tier stops being a range and becomes a line.

This is what I mean by compression, and it's why "the market went down" is the wrong description. A market going down moves the whole distribution. This one collapsed the distribution onto its own floor.

The method, because the obvious version is wrong

The lazy way to do this is track "the median price of an 85–87 card" week over week. Run that and you get a number that looks like a market index. It's garbage. You're not measuring price — you're measuring composition. The set of cards sitting at 85–87 in August has almost nothing in common with the set that sat there in March. New content ships, ratings shift, the population turns over. Of course the median moved. You changed the cards.

So the headline number above doesn't do that. Percentile spread on a single day is immune to it. I'm comparing the shape of one day's price distribution to the shape of another day's. Whichever cards happen to be in the tier on a given day, I'm asking the same question: how much more does the expensive end cost than the cheap end? Composition can't contaminate that, because I never compare a March card to an August card.

Then, to check the spread collapse is genuine re-pricing and not an artifact, I pulled the cards that never re-rated at all — same OVR every single day, March 17 to August 8 — and looked at them individually.

Nine cards that never changed

Same rating on the last day as the first. The expensive ones fell through the floor. The cheap ones were already on it.

Read that column twice. It's sorted by loss, and it sorted itself by March price on the way. The most expensive card lost the most. The cheapest card gained. There is no card in the middle of that table that breaks the pattern.

Nobody downgraded these players. Marte is a 90 today. Witt is a 91 today. The game never told you they got worse. The market did, quietly, over five months, while the number on the card stayed exactly where it was.

Why the top falls

Power creep, and it isn't complicated.

An 87 in March is a card you build a lineup around. An 87 in August is a card you already have four of, and one of them is a better fielder. Every content drop adds supply at the high end and moves the bar for what counts as elite. The card didn't get worse — it plays exactly the way it played in March. The standard moved out from under it.

What you were actually paying 100,000 stubs for in March was scarcity of quality, not the card. Scarcity of quality is the thing that expires. This is the same reason raw OVR is a misleading number — a 90 doesn't mean in August what it meant in March, in the market or in the box score.

The 88+ tier is the same story: median 82,922 → 9,006, and the spread from 5.0× down to 2.1× (two-week medians again — with only 17 cards up there, single days swing between 1.2× and 2.6×, so I won't pretend to a sharper number).

Why the bottom doesn't

Because there's a floor bolted to it and the top has nothing.

A card can only fall so far before quick-sell props it up, and that's exactly where the cheap end of every tier lives. Peña opened the season basically at the floor — 3,730 against an early-season 25th percentile around 3,150. He had no premium to lose. Everything that happened to Marte simply couldn't happen to him.

Below the diamonds it goes the other way. The sub-65 median went 5 stubs → 8 stubs, up 60%, while months of play pumped stubs into circulation against a pool of cheap cards nobody ships new content for. Not a fortune per card. But if you assumed your bulk was decaying toward zero, the data says the floor rose under it.

There is a limit to how well the floor loves you, mind. On June 24 I paid 1,000 stubs for a 58 OVR Daniel Duarte, a card whose floor is 25. I would like to tell you there was a reason. The floor protects the card. It does not protect you from yourself.

And the middle genuinely didn't move: 80–84 sat at 677 stubs on April 21 and 672 on August 8. I start that one in April on purpose — the 80–84 population goes from 64 priced cards to 88 over March, so a March-to-August comparison there would be measuring the catalog filling in, not the price. From April, when the population settles, it is flat for four months, while both ends of the market ran in opposite directions. Average the tiers together and you'd conclude nothing happened all season.

What this costs you

A premium is a depreciating asset. A floor isn't. That's the whole finding in one line, and it's more useful than "Diamonds go down," because it tells you which Diamonds. The exposure isn't the card, it's the gap between what you paid and what the tier's floor is. Marte cost 100,000 in March against an 88+ floor around 21,900 — about 78,000 stubs of gap, every one of which was at risk. Peña's gap was zero. He was the floor.

Check what you're holding against the floor, not against what you paid. If you own a Diamond currently trading at 3,400 in a tier whose floor is ~3,384, there is nothing left to lose and no reason to panic-sell it. If you own one trading at 30,000, that number is the thing at risk — not the rating.

Selling premium early and buying it late is the direction this season rewarded. Not a guarantee. Not a strategy I'd bet a lineup on. But it's the direction the numbers point, and it's the opposite of how most people play it.

Stop comparing stub prices across months. 5,000 stubs in March and 5,000 stubs in August are not the same statement about a card. This is exactly why we're moving the Diamond Radar price display toward predicted % change instead of leading with an absolute figure. +40% still means something in November. 8,432 stubs doesn't.

Marte on [his card page](https://diamondops.gg/market/a3b52fd737cde5a076ca32fe854d623d): flat at ~8,500 and going nowhere. The interesting part of this chart happened before it starts.

I own both ends of that table

I'm not writing this from the outside. I own nine of the nine cards up there, and I bought five of them with stubs.

On June 20 I bought Bobby Witt Jr. for 39,000. The close that day was 38,888 — I paid the market, I didn't get robbed, there's no story here about me being careless. He's 19,884 now. Five weeks earlier I'd taken Jeremy Peña at 3,441 against a 3,449 close. He's 4,000. Same inventory, same season, same guy making the calls, both cards still sitting in it. The 91 lost half. The 85 made money.

And here's the part that actually stings: 39,000 felt like the discount. Witt opened the season at 143,200. I bought him already down 73% and still lost half again. There was no floor under him yet — that's the whole point of this post, and I paid 19,000 stubs to learn it.

Zack Wheeler, same lesson, slower: 25,497 on May 7, worth 7,797 today. I have never once been wrong about whether these are good baseball cards. I was wrong every time about what the premium was worth.

What I'm not claiming

• Live Series only. Awards, Signature, Spotlight and the rest of the chase series are a different asset class — priced on scarcity rather than rating — and they're excluded deliberately.
• Nine cards is nine cards. The named table is an illustration, not the evidence. The evidence is the percentile spread across the whole tier (29–35 cards priced per day at 85–87, up to ~940 at the bottom). If the nine had disagreed with the spread, I'd have led with that instead.
• March is the noisy end. Fewer cards were priced in the first three weeks and a handful of odd listings drag the 25th percentile down to ~1,000 in early April. That's the other reason the headline uses two-week medians instead of two single days.
• One season, one game year. This is MLB 26, March 17 to August 8. Real pattern, real mechanism, but "this happens every year" needs a second season before I'll say it.
• These are daily closes, through 2026-08-08. I cut the series at the last complete trading day on purpose — the current day is a live, moving number and quoting it would make every figure here wrong by tomorrow. Individual cards spike far harder than any of these lines suggest — a few one-day listings run several times a card's normal price. This describes the tier, not your afternoon.
• I'm describing, not forecasting. Everything here is what already happened. Whether the floor holds through the winter is a different question and I don't have the data for it yet.

The durable version

Judge a card against its tier's floor, not against a stub number, and never against a stub number from four months ago.

The floor is the part that holds. Everything you pay above it is rent on being early, and this season the lease ran about five months. Everyone will tell you Diamonds get cheaper. Nobody's shown you that the amount you lose is just the amount you paid to be ahead.