Two Drafts, Same Time, One Me

Both of my ESPN leagues draft on September 2nd. 6:45 PM and 7:00 PM. Fifteen minutes apart.

I found out about the second one from an email last week. That league doesn't poll anybody about draft dates. It drafts the first Wednesday before the season, every year, and I'm the new guy, so I didn't know. The other league is my home league and I'm the commissioner. I polled everybody and we settled on our date a few weeks ago, so I can't change it now.

My league gives people two minutes a pick. The new one gives you ninety seconds.

Major issue...

Ninety seconds is fine if you have a plan and a good idea of how your league mates draft, and I do for one league. It's nowhere close to enough when a second draft is running at the same time, in a brand new league, with brand new people, and very bizarre rules. You can't build a good team by taking the next name off an 'expert list'. At every pick you're juggling six or seven things at once, which I got into in my newsletter a few weeks ago. Who came off the board, positional runs, who's left in each tier, roster needs, and so on.

On top of all that I now have two clocks. Two scoring systems. Two rosters that have to be built very differently. Ugh...

Last month I was building a little draft assistant because I wanted to. Now it has to be a lot more than that. I planned on a tool that would help me make informed decisions. Now I have to essentially clone myself.

This probably belongs on a fantasy football prep site rather than a LinkedIn newsletter. But I write about the side projects I'm working on in my spare time, and fantasy season is here, so this is what I'm spending my time on.

I wrote about this project last month, but that one was conceptual. What I was thinking about building, plus some strategy. This one is a deep dive into fantasy football analytics in 2026. Once I found out both drafts were happening at the same time, I realized I had to test every theory I've carried around for the fifteen years I've been playing. And research the 2026 board. If I'm building a clone, I need to do it right.

If you don't play fantasy football, you can still read this. I'll explain terms as they come up. The real subject isn't football. It's analytics.

It's long. If you play fantasy football, it's worth it. If you don't... idk. Nobody is forcing you.

Why not just auto-draft one of the leagues?

Um... no thanks.

Two reasons.

One, a fantasy team is a four-ish month commitment. Lineups every week, waivers, injuries, byes, trades, streaming defenses and sometimes tight ends and quarterbacks. If I'm in a league, I'm keeping up with all of that, so I'm not handing away the few hours that actually decide my roster. Plus it's MY team, not ESPN's.

Two, ESPN's rankings are terrible. And if you auto-draft, that ranking system picks your team.

I pulled the ESPN suggested draft order out of my actual leagues and compared it to high-stakes real money draft prices, adjusted for ESPN and Yahoo default settings, which is the closest thing to an honest market signal that exists in fantasy.

ESPN's board sits two and a half times further from that market than Yahoo's does. The worst individual misses run about 50 picks in both directions. That's frankly pathetic.

This might not seem like a big deal. It is massive. A round in a twelve team league is twelve players. Being a round off means the board is telling you to take a guy while twelve better players are still sitting there. Why would anyone do that? And ESPN keeps offering those suggestions the entire draft.

The other problem is that my two ESPN leagues have very different rules. One is full PPR paying 4 points per passing touchdown. The other is half PPR paying 6. PPR means a point every time your guy catches a ball, which moves a lot of players around.

ESPN hands both leagues the identical board. All 250 ranked players, same order, both rooms. Which makes zero sense, and it's more proof you shouldn't lean on their rankings.

Auto-drafting won't automatically lose your season. But it hands you a roster with well under an 8.33% chance of winning the league. (1 of 12 teams.) Drafting well probably gets you somewhere around 15%. Drafting badly drops you closer to 3%.

Before I spent weeks programming a bunch of drafting rules, I needed to know whether my rules were any good.

I've been drafting fantasy teams for fifteen years, and a lot of my decisions come from experience and instinct. Drafts are fluid. Every pick changes the board, and every decision depends on who's still there.

That makes it surprisingly hard to turn a drafting philosophy into a set of clear rules. But before I could even try, I had to find out which of my long-held assumptions the data actually supported.

So I pulled fourteen seasons of real draft prices and joined them to real box scores. Roughly 112,000 simulated 12-team leagues. Then I started digging.

Championships usually aren't won by a roster full of solid picks. They're won by a couple of players who massively outperform what you paid for them.

Draft the middle of the pack every round and you'll probably build a competitive team. You probably won't win a championship. For that you need one or two 'league winners'.

46.5%

Todd Gurley's 2017 took his owners from a baseline 8.3% title chance to 46.5%. He went at pick 19. That's what a league winner is. One guy who roughly quintuples your odds.

The obvious next question is how you spot them in advance.

I tested 108 preseason markers, corrected for luck. Age. Prior finish. Target share. Touches. Draft capital. Team change. Vacated opportunity. Role. Market disagreement. Everything I had.

0 of 108

None of them held up as reliable predictors. Some were better than others. But a model built from the best combination of them beat sorting the board by ADP by 0.024 AUC, 95% CI [-0.001, +0.049], p = 0.066. In plain English, my best model was no better than just drafting in market order.

So... fantasy football is unpredictable. Shocker.

Here's what that means though. When you're prepping for a draft, you are not identifying league winners. You're buying lottery tickets, and the only things you control are the price you pay and where you buy each one.

Which actually simplifies everything. Instead of chasing a predictive model that never beats the market, I can build a machine that does the math on value during the draft itself.

I walked through value over replacement and VONA in last month's newsletter. Value over replacement compares every player to the best undrafted player at his position.

In other words, if you pass on this guy and wait until late for the position, this is what you're giving up.

The number one running back is typically worth more than twice his replacement. The number one quarterback is worth about a third more than his.

That ratio is why the first three rounds are almost always running backs and wide receivers. It's also why paying up for a quarterback is usually a mistake, even when it's the best quarterback in the league. (Josh Allen... not Patrick Mahomes.)

Auto-draft would never...

Replacement level is league-specific, and understanding that is a real edge. Most cheat sheets are built for generic settings, not your league, and they rarely evaluate replacement value at all. The good news is building your own is not complicated. It depends on two things: your roster settings and your scoring. How many players you start at each position, and how those players earn points.

Once you have projections, the rest is a little math. Easily 'Googleable' math... I think that's a word.

Opportunity cost by round. The question at every pick, especially early, is who the best value is relative to everyone else on the board right now.

Measured across fourteen seasons, points over replacement by position by round, the pattern is obvious. Receiver is the best value in round 1 and again in rounds 3 through 5. Running back goes negative from round 4 on. Quarterback spikes in round 6 and again in round 10. Tight end is basically the best position in the draft from round 8 to the end.

Tiers matter, not ranks. The biggest single drop at every position is from first to second. RB1 to RB2 is 49 points. TE1 to TE2 is 39. QB1 to QB2 is 37. WR1 to WR2 is 35. The most valuable thing in fantasy is being right about who finishes first, and nobody can predict that with any certainty.

After that, the positional cliffs matter most, because the gap between the last player in a tier and the first player in the next one is the only good reason to ever reach for somebody.

Below are some of the 2026 cliffs by position for one of my leagues. Every player is a dot. As guys come off the board I can see what's left at each position and how close the next cliff is. I keep this chart open the entire draft. And the cliffs move with your settings. One of my leagues is completely flat at quarterback this year, so I won't take one until the end and I'll spend those picks on higher tier players everywhere else.

The 'dead zone' is the stretch where running backs stop being worth what you have to pay for them.

In my data the average back stops clearing his own replacement level at pick 37. Two full rounds before any other position goes negative. So if I want backs, they have to come early. From picks 37 through 72 a back returned a startable season 38% of the time, against 58% for a receiver at the same price. From round 5 on, running back leads every single round in returning nothing.

That's why every draft strategy is named after running backs. Zero RB, Hero RB, and Double Hero RB.

A lot of analysts argue the dead zone has moved. The market pushed backs around, and the dead range now gets described as rounds 7 through 10 instead of 4 through 6, partly because those backs catch more passes and see more red zone work than they used to. Sharp Football's ADP history tracks the drift. So you can wait a little longer than you could 5 or 10 years ago. I'd still rather not. Mid-tier backs bust constantly.

My capture says the 12th back comes off the board at pick 37 this year. Same number as the one above, and yes, that is a coincidence. One is where backs historically stop clearing replacement, the other is where the 12th back is going in 2026. For nine straight seasons he was gone by pick 31, so the top of the position has stretched later. I still want two of them early if I can, because the elite ones go fast and mid-draft receivers are much easier to hit on.

The real problem with the dead zone is that nobody can tell you where it is before the season starts. But look back at the end of any year and there's always an obvious point where the backs stopped returning value.

The first receiver taken is the safest pick in fantasy football. Backs get hurt far more often. 88% of first round receivers finished WR3 or better against 72% of first round backs, and they bust half as often. (Bust just means they finished well below what you paid.) The median first round finish is WR7 versus RB11.

Inside the first three rounds, one marker is everything. Target share, which is just how many of the team's passes went to that player. A receiver with a 28%+ target share the year before hit a top-24 finish 79% of the time, against 62% for everyone else. The metric stops working after round three, because alpha receivers don't last much longer than that.

Which sounds like a great argument for taking receivers over backs early. But...

Past round three is where it gets interesting.

2026 specific: receivers in rounds 6 through 9 are the most contested ground on the board at any position this year. The experts disagree there more than anywhere else, with a mean standard deviation of 17.0 and a rank range of 91. The single widest disagreement inside the top 100 is a receiver whose expert range runs from pick 51 all the way to 241. And that is the same band where the good backs have already died off. In my opinion this is where the draft gets won and lost.

This is where it pays to read up on offseason moves and figure out which receivers are set up to take on a bigger role and shock everybody.

Middle round receivers can win you your league. Middle round backs can't. So I'm taking running backs early and buying lottery tickets at receiver in the middle.

0 for 55

Fifty-five quarterbacks have been drafted between picks 48 and 95 over fourteen seasons. Not one of them produced a league winner.

Early is defensible, but only if you get Josh Allen, who has skewed these numbers for years. Late is defensible. The middle is where you pay a real price for the smallest premium in the game.

Paying up has a worse record than people think. Quarterbacks drafted in the top three at the position matched or beat their draft slot 22.9% of the time. (Much lower if you pull Allen out.) Quarterbacks drafted QB10 through QB12 did it 43.8% of the time.

2026 specific: Allen is the only quarterback going in the first three rounds. Lamar Jackson is next around pick 40. The position is cheap this year by recent standards, which makes that middle band even harder to defend.

The public consensus landed in roughly the same place. Wait until round 8 or later, because tier three runs flat from about QB7 to QB17. Why take a quarterback in round seven when a guy who scores almost the same is sitting there in round twelve? That is not true at running back or receiver.

It also matters which ones run. In a four point touchdown league, rushing quarterbacks are cheat codes. In a six point league, pocket passers gain value.

Similar shape to quarterback. Get a tier one guy before the cliff or wait until TE10 and later. The middle, roughly TE3 through TE9, is the worst value on the board. In PPR pricing this year there's a 21-pick gap between the second tight end off the board and the third.

Tight ends are the best bet left on the board for a starter you can plug in late. 25.6% of late tight ends cleared replacement level against 14.3% for late backs and receivers combined, p < 0.001. That's the single most actionable number in the whole study.

Late tight ends are not lottery tickets and you're not swinging at a league winner. You're grabbing a usable starter in a round where usable starters don't exist, which frees up an earlier pick to take a real swing somewhere else.

So from round 10 on, take one or two tight ends.

One more thing. Try to take one who had a 20% target share or better last season. Those hit 58% of the time. Under 15% only hits 18% of the time.

The board is the most veteran it's been in fourteen years. Only 4 rookies inside the first 100 picks, against 9 last year. In their place is proven production. 12 receivers inside pick 120 held a 28%+ target share the prior season, against 8 a year ago.

The 'experts' have also basically stopped arguing. Mean expert standard deviation between picks 37 and 100 is 11.9, the tightest board in fourteen years, and running back disagreement is the lowest I have on record. One caveat. My 2026 capture is from August 7 with 87 experts, and every historical season in the archive was captured in early September.

And the two elite tight ends are leaving earlier than usual. The first one goes at pick 17 on ESPN's board against 20 in the money market.

No existing draft tool handles customization well. My three leagues use three different scoring systems and three different roster setups, which means the same player is worth three different amounts depending on which room I'm in. Full PPR with 4 point passing touchdowns. Half PPR with 6 point passing touchdowns. And the weird one, half PPR with one mandatory back, two flex spots, fourteen rounds, and six bonus rules including kicker and defense bonuses I'd never seen before.

Same players. Same projections. Three completely different boards... and one me.

So I built a clone that can keep up with both clocks. It is probably going to draft one of my teams outright.

A lot. It does a lot of cool stuff. It's amazing, still in progress, and I am running out of time.

The homepage looks like a sleazy AI slop paid tool. It's not, and I have no intention of selling it. It'd be hard to sell anyway, considering it's built for three specific leagues.

I had to put a password on it so my league mates stay out. If you want in, message me and I will give it to you.

A value engine prices each league separately, on its own replacement levels.Three leagues, three boards, each built from that league's exact scoring and roster. The engine computes value over replacement off five seasons of real finishes instead of projections, then fits it to a curve that can only go down. Which means it refuses to claim it can tell QB5 from QB8 when five years of evidence can't.

It reads my eleven favorite ranking sources and knows what each one is good for.Expert consensus, real money prices, my own rankings, the platform's order, a YouTuber I trust, a ceiling list built on upside, and about 20 other metrics. ADP never gets a vote, because ADP is the market I'm drafting against, not an opinion about what a player is worth. Sources built for another format, like best ball and high-stakes, can corroborate a pick within a position but never nominate one, because their cross-position pricing is distorted by their own rules.

A verdict box merges thirteen voices and every piece of data on the board into one recommendation, and prints its reasoning.Which sources agree, which ones don't, what the alternative is, and what passing costs me. It also prices injuries, breakout candidates, fades, and late round flyers.

It draws the tiers and cliffs instead of a ranked list. Every position gets its own curve with the drop-offs marked, so a reach is something I can see rather than something I feel. This is the view I look at most. I can see where we are on every curve and what the guys already drafted were worth.

It enforces my rules as hard constraints.Never more than two quarterbacks or two tight ends. No injury designation ever touches a pick suggestion, because August designations are unreliable and I'll know who's hurt on draft day anyway.

It shows who's falling. I can see the players who are badly undervalued in each room and make sure I take them before they land on somebody else's radar. Some of them are going three or more rounds later than they should on ESPN. I'll take a few a round early just to be certain I get them.

It tells me who survives to my next pick.For every player left, the odds he's still there when I'm back on the clock, and the pick after that. So I can pass on the best value in front of me and know what that costs, instead of hoping.

It gives me targets, and targets by round.This is the one I've already memorized. Every player priced below what he's worth, ranked by the size of the gap, in each room separately. The lists barely overlap, because the scoring systems change everything. Some of these guys are going five rounds later than they should.

They're also broken out by round, so I know the round I have to take each one by before somebody else does. This is the most important tab on the board, and it's the one screenshot I can't show you. If a league mate sees that list my whole strategy is gone.

The last piece is still being built. Wiring it to the browser. Right now I mark players off by typing a name or clicking a checkbox. Soon the boards will hook straight into my live drafts, so every player taken in either league comes off both boards the moment it happens, without me lifting a finger.

Every previous year I used spreadsheets and the FantasyPros draft tool, crossing players off as they came off the board. That is not happening this year.

The piece I'm most excited about is Barnabas, my agent, running as my in-draft researcher. Both rosters, the live board state, every decision engine's output, all the research above. He watches for runs at a position or deviations from a normal draft and flags value as it pops up. Reacting to the room is the hardest part of a live draft. Everything moves fast, people make mistakes, and you have to stay fluid. He'll tell me what he thinks the best pick is every round, and why.

I will still go off gut and favorite players sometimes. But at least I will know when I am doing it.

I can already smell the trophies. That's a weird thing to say...

The board's front page, with the 2026 numbers: finkledraft.dustincoledata.com

If you are not in my leagues, I will give you the password.

If you play fantasy football, good luck this year!

If you don't play fantasy football, I'm sorry you made it this far.... But you should start.

Last issue, the prequel: Ray Finkle's AI Fantasy Football War Room

🔗 Newsletter: dustincoledata.beehiiv.com

🌐 Site: dustincoledata.com

📩 Send me a message to see the draft board.

- Dustin