When we flagged Yordan Alvarez as a roster-update riser, someone in the comments did the exact right thing: they checked his splits and pushed back.

He's been brutal against lefties this update period. If SDS trims his L splits, there's no path to 93.

That's a good question. It's the question. And the facts behind it were dead-on — over that stretch Yordan had zero home runs and a .210 average against left-handed pitching in 27 plate appearances. If you're reading his splits by eye, that number jumps off the page and screams downgrade risk. Anybody doing the homework would flag it.

He went to 93 anyway. Not because the question was wrong — it was a sharp, correct read of half the picture — but because of something that's genuinely hard to hold in your head when you're grading splits by hand. Getting that one thing right is how you catch risers a week before their price moves. So let's get it right.

Yordan Alvarez's live 93 DiamondOps card — buy/sell prices, flip ROI, and quirks

Your card has two Contact ratings. And two Power ratings.

Most people never notice this, so look at an actual card for a second. Every hitter in MLB The Show doesn't have one Contact rating and one Power rating — they have two of each. Contact vs. Left, Contact vs. Right. Power vs. Left, Power vs. Right.

The Batting attributes on Yordan's DiamondOps card: Contact R 104, Contact L 100, Power R 110, Power L 95

These are platoon splits, and SDS sets them from real-life platoon splits. Aaron Judge, a righty, is rated higher for Contact vs. Left than Contact vs. Right — because in real life he eats lefties alive. The card is telling you how good the player is against each hand, separately.

Here's why that matters for your stub balance: those split ratings lag real life. A player goes on a two-week tear against righties, and his Power vs. Right rating hasn't caught up yet. The gap between what the card says and what the player is actually doing — that's the room for an upgrade. Catch the gap early and you're buying the card before the update closes it and the price jumps.

That's the whole trick. Reading it right is where it gets interesting.

The window that matters is the gap between updates

Before the examples, one thing that trips everybody up, including the people who do the homework: the window.

When you check a player's splits to guess a roster update, you can't use his season line — too much old data drowns out what's happening now. And you can't use his last box score — too small. What SDS actually responds to is everything the player has done since the last update. For the July 16 update, that's roughly June 12 through July 16 — the ~34 days since the June update landed. Not the season. Not last night. That month.

Everything below is that window, pulled straight from our own data. Watch what SDS did with it.

Yordan Alvarez: all eight homers, one side of the plate

Here's Yordan's real production between the June and July updates:

Yordan Alvarez's production split, June to July: all 8 homers and a 1.133 OPS against righties, 0 homers and a .432 OPS against lefties

Every one of his eight homers came against right-handers. Against lefties he barely played, and when he did, nothing. So SDS did exactly what the split-lag idea predicts:

• Contact vs. R: 92 → 104 (+12)
• Power vs. R: 100 → 110 (+10)
• Contact vs. L: 106 → 100 (−6)

How SDS responded: Yordan's Contact vs R +12, Power vs R +10, Clutch +6, Contact vs L −6, overall 91 to 93

Reward the side he mashed, trim the cold side. The platoon split, in one card. And notice it's a recent-form read — season-long, Yordan hits lefties just fine. This bump was about the last month, which is the entire point. He also happened to be the single biggest call the Diamond Radar made that window: +10,880 stubs if you bought the flag and sold the pop.

The commenter had the lefty half of this exactly right. Hold that thought.

Junior Caminero: two dials, not one

Now the case that breaks the simple version of the rule. Same window:

Eleven homers against righties — huge power — but only a .246 average there. And scorching against lefties in a small sample. Watch SDS pull these apart:

• Power vs. R: 86 → 96 (+10) and Power vs. L: 83 → 91 (+8) — 14 homers total, it rewarded the pop on both sides
• Contact vs. L: 78 → 88 (+10) — the .414 against lefties
• Contact vs. R: 77 → 73 (−4) — the .246 average against righties, even as his power there went up

How SDS responded: Caminero's Power vs R +10, Power vs L +8, Contact vs L +10, Contact vs R −4, Vision −6, overall 85 to 87

Look at that righty side. Power up ten points. Contact down four. On the same hand, at the same time. That's not a bug — it's the resolution. SDS isn't grading "how good is he against righties" as one number. It's grading how much power and how much contact against righties as two separate dials. Eleven bombs and a .246 average is exactly what "power up, contact down" looks like translated into a card.

Once you see the dials as independent, roster updates stop looking random.

So why did "no path to 93" miss?

Back to the pushback, because this is the part worth internalizing.

It was a completely valid read, and I want to be clear about that — it got the facts right, it asked the right question, and that kind of thinking is what makes you money on the marketplace. It missed for a reason that has nothing to do with being careless and everything to do with what's hard to compute in your head. The take reasoned in one dimension: bad weak side, so they cut the weak side, so no upgrade. Three things that logic can't easily weigh:

One — OVR is a net, not a floor set by your worst split. SDS's actual lefty adjustment was small: Contact vs. L down six, Power vs. L untouched. A single −6 doesn't come close to cancelling out what was happening on the other side.

Two — the strong side had enormous room to grow, and SDS rewarded it hard. Yordan wasn't just good against righties, he was historic: .342, .678 slugging, all eight homers. SDS answered with +12, +10, and a clutch bump on top. That's the OVR. The lefty trim was a rounding error against it.

Three — "terrible against lefties" was terrible by his standard, not the card's. His Contact vs. L was 106. Even after the cut, it's 100 — still elite. Poor recent production against lefties is not the same as a card-crippling rating. He was never going to fall off a cliff on a side he's rated in the triple digits on.

The lesson isn't "the split doesn't matter." It's that a cold weak side which earns a small trim can't outrun a hot strong side earning a big boost. Net, not floor. That's the whole thing — and it's genuinely hard to eyeball, which is why smart people miss it.

Doing it yourself

You can run this by hand, and it's a real edge:

1. Find a card whose real player is hot against one hand — Baseball Savant, or any splits page.
2. Compare that to the card's split rating for the same hand.
3. Big gap, and the player still getting regular reps against that hand? That's a candidate.

The inverse is the warning label: a card rated high on a side the player has gone cold against is a downgrade risk. Don't overpay for a rating the next update is going to walk back.

Honest caveat, because I'm not going to pretend this is a cheat code: it's an edge, not a lock. SDS doesn't close every gap, and other stuff moves too — discipline, vision, clutch. Reading one split right doesn't mean you've read the card right. Which is exactly the trap from the top of this post, one level up.

Or let the Diamond Radar do it 1,300 times a night

Nobody's hand-checking every hitter's left/right splits against their card every night. That's the job I built the Diamond Radar to do. It looks at how a player is actually producing against lefties and against righties, compares that to what the card's split ratings say he should be, and turns the gap into a signal for who's trending toward an upgrade — across the whole card pool, every night.

And here's the part that connects back to the comment. The reason a model earns its keep isn't that it reads splits better than a sharp human. It's that it weighs the gap on every attribute at once and nets them out, instead of anchoring on the one scary number and stopping. Same split data the pushback had. The difference is the Radar didn't let a cold lefty line carry the whole verdict — it saw the giant right-side gaps swamp the small left-side one and called Yordan the biggest riser of the window. Human grading is great. This is just the spot where it slips, and where running the math on all of it at once pays for itself.

It's a forecast, not a prophecy. That same window, the Radar flagged 226 upgrades and 124 of them actually happened — right a little better than half the time it makes a call. It caught the +10 OVR jump on Samy Natera Jr., 53 to 63, before it landed. It also whiffs plenty. Think weather forecast: it gives you the odds, not a lock. But odds, one week early, is the entire game on the marketplace.

The players who win aren't guessing. They're reading the same signals SDS reads — one update ahead.

See tonight's top split-driven risers on the Diamond Radar.