What is DPX (Dynasty Performance Index)?

Data Dictionary · the in-career dynasty value model

DPX, or Dynasty Performance Index, is my in-career score. DMX grades a guy once, on draft day, and then it's done with him. DPX picks him up from there and re-scores him off what he actually does on an NFL field. I run it on every fantasy-relevant skill player — 348 of them for the 2025 season — and publish two deciles apiece: DPX-Next for next season, DPX-Long for three years out. It's my answer to the only question that matters when your phone buzzes with a trade offer…what is this guy worth right now?

Why it matters

Dynasty trade values run on ADP and vibes, and both of them lag. A guy's role changes in Week 3 — his snap share jumps, or his targets quietly dry up — and the market takes weeks to catch up. Sometimes it never does. That lag is where you make your money.

So I built DPX to price it. One number, computed the same way for every guy in every season, that you can hold up next to whatever your leaguemate thinks he's worth. When those two disagree, you've got a trade. It works in both directions, and honestly the sell side is where most folks leave money on the table.

How DPX is calculated

Three pieces, each one a z-score, all of it updated off real game data:

Volume
VOL
Touches, targets, snap share. This is the opportunity baseline, and it's the stickiest thing in football. A guy who's getting the ball holds his dynasty value even through stretches where he's running average.
Scoring
SCR
What he actually put up per game, with half-PPR and PPR pathways. Dynasty value compounds off scoring, so this layer carries the most freight.
Efficiency
EFF
What he does per touch. Catches the guys whose volume bounces around but who produce every time they're on the field, and it rats out the volume merchants living on borrowed time.

Sitting on top of those three is a dynasty modifier for contract status and draft-capital decay. That's the part that makes this a dynasty number instead of a redraft number. Give me a 28-year-old back and a 22-year-old back who just put up the same season, and I'm not paying the same for them. Nobody should. The modifier is where that gets priced.

Same deal as DMX on the weights. How I blend the layers, and what the modifier coefficients are, I keep to myself. Everything else is out in the open on the Model Transparency page — the inputs, the outputs, the validation.

DPX-Next vs DPX-Long

Every player-season gets two deciles, and the gap between them is usually where the interesting stuff lives:

Take Derrick Henry in 2025. He's 31, and DPX-Next still has him D1, because he's one of the best backs you can start. DPX-Long has him D3. Both of those are true at once, and together they tell you what to do: start him, and shop him. Now flip it. Tyler Warren, the rookie tight end in Indy, came out D2 on Next and D1 on Long. The long-horizon number moved first, which is what a young guy looks like right before the market wakes up to him.

My rule of thumb: if you're contending, buy Next and let somebody else hold the decline. If you're rebuilding, buy Long and rent out the present. Most of the genuinely bad dynasty trades I've watched come down to a contender paying for Long or a rebuilder paying for Next.

A worked example

Let me walk one through. Here's Jonathan Taylor off the 2025 season scores:

ComponentZ-scoreReading
Volume+1.84Workhorse usage, elite touch and snap share
Scoring+3.41An outlier — three-plus standard deviations of production per game
Efficiency+1.65Per-touch output backing the volume up
DPX-Next1.93 · D1Elite next-season projection

Live 2025-season scores from the DPX rankings, June 2026.

All three layers point the same direction, which is what a no-doubt RB1 looks like in the data. Fine. You didn't need me for that one.

The guys worth your time are the ones where the layers fight each other — big scoring sitting on thin volume, or monster volume with the efficiency falling out from under it. That's where DPX walks away from the consensus rankings, and that's where the trades are.

Reading DPX like an analyst

Four patterns cover most of what you'll actually use this thing for.

DPX vs DMX

DMXDPX
QuestionWhat should I pay on draft day?What is he worth now?
InputsAthleticism, draft capital, college productionNFL volume, scoring, efficiency
ComputedOnce, at the draft — never updatesUpdated through every NFL season
Covers27 draft classes, 2000–202613 NFL seasons of player-scores

I built the two so they wouldn't overlap. A rookie carries his DMX score into the league as the prior, and from his first NFL season on, DPX takes over as the live read. When a guy's DPX trajectory breaks hard from what his DMX said he'd be, in either direction, that's exactly what the Analytics residual tools are built to chew on. Those breaks are my favorite thing in the whole database.

Using DPX on the platform

Frequently asked questions

What is the difference between DPX and DMX?

DMX is the pre-draft number, frozen the day a guy gets drafted. DPX is the in-career one, rebuilt every season off his actual NFL volume, scoring, and efficiency. DMX prices rookies. DPX prices everybody after that.

What do DPX-Next and DPX-Long mean?

Two deciles per player-season. Next projects the coming season, Long projects three years out. The gap between them is usually the part you can act on.

How is DPX calculated?

Volume, Scoring, and Efficiency each get z-scored and blended into a composite, then a dynasty modifier adjusts for contract status and draft-capital decay. I keep the exact weights private. The inputs and the outputs are public.

How often does DPX update?

It gets recomputed off new game data all season long, so risers and fallers show up while you can still do something about them, not in some tidy end-of-year refresh.

Why does a high-scoring player sometimes have a mediocre DPX?

Two usual suspects. Either the scoring is riding efficiency that won't hold with no volume underneath it, or the modifier is pricing in decline from age, contract, and draft capital. DPX is telling you what the production implies going forward, not what it was worth last week.