We Backtested 10 Years of Fantasy Drafts. Almost Nothing Held Up. Almost.

Why bother

Everyone in fantasy football has a theory. Take a running back first. Wait on quarterback. Grab the elite tight end. Fade the position everyone else loves.

Most of those theories survive on one memorable season. So we ran them.

We pulled ten seasons of NFL scoring (2015 to 2024, nflverse), pulled the public 12-team ADP for each of those seasons, rebuilt draft boards for each strategy, and simulated every strategy from all twelve draft slots in every season. That is 1,320 drafts, each one scored on what those players actually did.

One question: if you had drafted this way every year for a decade, would you have made the money more often?

The method, short version

• Scoring: nflverse weekly stats aggregated to season totals, 2015-2024.
• Draft input: Fantasy Football Calculator 12-team ADP by season and format. PPR and standard cover all ten seasons. Half PPR only exists from 2018.
• League: 12 teams, 15-round snake, 1 QB / 2 RB / 2 WR / 1 TE / 1 FLEX. No kickers or defenses: public ADP for those spots carries no usable signal, so they are handled separately in Kickers and Defenses Without Dogma.
• Opponents: eleven ADP bots taking best available with roster caps.
• Every strategy runs from all 12 seats, every season. No draft-slot luck.
• Success metric: top-3 finish on season-long starter points.

The number that matters most: the control. Drafting straight ADP against ADP bots finishes top-3 exactly 25.0% of the time, because three of twelve draft spots make the money. Every strategy below should be read against that 25% line. Those are the expected odds, but as Blueprint people, we’ve been able to double those numbers.

Finding 1: Nothing separated itself at the top, but a clear bottom.

Ten seasons, 1,320 drafts, and the spread between the best and worst archetype is smaller than the swing you get from one bad hamstring. The best result in the entire study, robust RB in standard scoring, beat pure ADP by 4.2 points of top-3 rate. That is real but it is not a system.

If someone sells you a draft archetype as the reason they win, the ten-year record does not back them.

Finding 2: Zero-RB never beat the coin flip. Not once, in any format.

This is as we expected, for years we’ve talking about the need to go heavy on running backs early.

• PPR: 21.7% (control 25.0%)
• Half PPR: 14.3% (control 25.0%)
• Standard: 18.3% (control 25.0%)

Zero-RB is the most-defended strategy in fantasy football, and in a decade of simulated drafts it was the worst archetype in all three formats. It is also wildly streaky. In PPR it hit 50% in 2022 and 42% in 2015 and 2021, then went 0% in 2016, 2020, and 2024.

That pattern is the whole story: Zero-RB is not a strategy, it is a bet on a specific kind of season. When the cheap-RB tier hits, you look like a genius. When it does not, you have four wide receivers and no floor.

Finding 3: Format decides which archetype is smart

Robust RB, meaning running backs with your first two picks, is the best archetype in standard scoring (29.2%) and among the worst in full PPR (21.7%). Same drafting behavior, opposite verdicts, purely because receptions change what a running back is worth. This is, perhaps, the most important thing to consider this year – our matrix cheat sheet already accounts for roster and scoring settings for your league, but don’t force yourself into RB-RB early. The elite WRs get a greater advantage in PPR so are more valuable and there are pass-catching specialists at the running back position that can help you close the gap after not drafting a top RB.

Anyone recommending a draft archetype without asking your scoring settings first is guessing.

Finding 4: Late-round QB is fine, and that is all it is

Across all three formats late QB landed at 23.3%, 21.4%, and 20.0%: at or just under the control line. It does not cost you the league and it does not win it. It is a preference, not an edge. Same conclusion for hero RB, which tracked the control almost exactly. But, see below about QB now being a cliff position.

Finding 5: The elite TE edge is real but it is thin

Elite TE was the best archetype in both PPR formats (26.7% and 27.4%), and it did it by grabbing the largest positional gap on the board. But look at the season-by-season PPR line: 50%, 8%, 33%, 8%, 25%, 42%, 17%, 50%, 17%, 17%.

Two great seasons, two disasters, six coin flips. Paying up at tight end works when your guy delivers the gap you paid for and washes out when he does not. But that’s the nature of fantasy football at any position. In the end, we’re all guessing. Might as well take a swing on a guy who, if does turn out elite, can win you your league. It’s the same reason Gronk was actually worthy of being the number one pick for many years – we were ahead of that trend and many of our subscribers won championships because of it.

What actually moves the needle

The archetypes cluster around chance because they are all answers to the wrong question. They optimize the shape of your roster. The decade of data says the returns live in which player you take at each pick, and in what you do after the draft. We’ve often said you can’t win your league at the draft, but you can lose it. It takes a lot of management during the season, and we’ve got all the tools to help you.

So the follow-up question is: if picking the right player is the whole game, which picks are most likely to be right? Two different things get called “consistency” in fantasy, and they answer different questions. We measured both and kept them separate.

1. Predictability of a draft slot (Findings 6 and 7): when the market takes the third receiver off the board, how often does that player finish as a startable receiver? This is the one that helps you on draft day, because it is about the slot you are drafting, not about any specific name.
2. Same-player persistence (Finding 8): does last year’s RB8 tend to be this year’s RB8? This is a property of players, and it is mostly a warning about drafting off last season’s finishes. Others in your league will draft off of name value or recency bias. You’re Blueprint people; you won’t.

Finding 6: How predictable is each top pick, by ADP slot

This is measure 1: draft-slot predictability. We took every player inside the top 24 at his position in preseason 12-team PPR ADP, 2015-2024, and scored him against what he actually did that same season. No year-over-year comparison is involved. Startable tiers are QB1-12, RB1-24, WR1-24, TE1-12.

• Hit = finished inside the startable tier at his position.
• Elite = finished top-3 at his position.
• Bust = finished outside twice the tier, or did not post a season at all.
• VOA = points above the average startable finisher at that position, the same baseline math the board uses.

The first three off the board at each position (positional ADP 1-3, n = 30 each)

Positional ADP 4-6 (n = 30 each)

Positional ADP 7-12 (n = 60 each)

Three things fall out of those tables.

The first three receivers are the single most reliable block of picks in fantasy football. 83.3% land startable, 63.3% clear the WR baseline, median finish WR6. Nothing else in the study is that dependable. If I have a top 3 pick, I’d be tempted to go WR-RB-RB. And if I’m in a PPR league, I’m definitely going WR first – not because wide receivers are more valuable than running backs, but because the elite wide receivers offer an even greater advantage over the rest of the wide receivers in PPR. That’s the whole idea of our proprietary VOA+ metric.

Elite outcomes stay near the top of the board and disappear fast. Top-3 finishes come from positional ADP 1-3 at a 27-40% clip depending on position, fall to roughly 20% at ADP 4-6, and to 7-17% at ADP 7-12. The chance of drafting a positional winner is roughly cut in half by the time you reach the seventh player at a position.

Tight end and quarterback are cliff positions. TE1-3 hits 76.7% with the highest elite rate in the study (40.0%). TE4-6 hits 53.3% and TE7-12 hits 43.3% with a 26.7% bust rate. QB goes 80.0% to 56.7% to 50.0%. If you are not taking one of the first three, the difference between the fifth and the eleventh is mostly noise. Don’t panic picking between closely ranked players, take the guy you want. With the increased offense the league has seen over the last several years, elite QBs are separating themselves more than ever. It sounds silly to say this but try to get yourself a top guy at each position.

Running back is the interesting exception: RB1-3 has the lowest hit rate of any top-3 block (66.7%, median RB10), but RB4-12 hits startable more often than any other position in that range (73-77%). The very top of the RB board is priced for a ceiling it misses about a third of the time. The next tier is duller and safer than its reputation. Again, lending a little more creedence (not the clearwater revival type) to going wide receiver first if you have a top 3 draft spot.

Finding 7: So is Zero-RB bad because RB is the most predictable position?

False, on both definitions. By draft-slot predictability, the first three RBs are the least reliable top-3 block in the study (66.7%). By same-player persistence, the RB startable tier turns over at 52.8% a year, mid-pack among skill positions. Neither number says “safe.”

Zero-RB underperformed for a different reason, and the data names it directly. Average points by which the top-3 finishers at a position beat that position’s own startable baseline, 2015-2024 PPR:

Elite running backs beat their own baseline by more than anything else. That is the cost of Zero-RB: you are not fading an unstable position, you are volunteering to skip the biggest surplus on the board and hoping the cheap tier closes it. Ten seasons say it usually does not. It also explains why RB1-3 keeps its price despite a 66.7% hit rate: the payoff when it works is bigger than anywhere else. If not PPR, I’d still lean heavy on running backs early because of this massive advantage. The weaker the rest of the running backs, are the most advantage you get with an elite one. We’ve built this idea of the running back cliff into the matrix cheat sheet.

Finding 8: Same-player persistence is weak everywhere

This is measure 2, and it is a separate question from the ADP tables above: how often is the same name startable in back-to-back seasons?

• 47% of every season’s startable tier is a new name. Averaged over 2016 to 2024, only 52.5% of the 72 startable skill-position finishers were startable at the same position the year before. Turnover never dropped below 40% in any season of the sample.
• Repeat top-3 finishers are rare: 18.5% of a season’s top-3 RBs and QBs were top-3 the prior season. TE is highest at 37.0%, and it is the smallest tier. Again, think Gronk and Kelce level of consistency. We don’t really have that anymore as there as teams are starting to use tight ends more and more.
• Year-over-year repeat rate of the startable tier: QB 54.6%, RB 52.8%, WR 51.9%, TE 47.2%. Nothing separates by more than seven points, so no position is meaningfully “stickier” than another at the player level.

Note that these two measures do not have to agree, and they do not. WR is the most predictable slot to draft (83.3% for WR1-3) while WR names repeat at only 51.9%. Draft the slot and the price, not the name that finished there last year.

The honest bottom line

1. Pick the archetype that fits your format, then stop thinking about it.
2. Do not pay a premium to execute a strategy. The premium is the whole cost and the strategy is worth about nothing. It’s the same reason we don’t advocate doing a ton of mock drafts, everyone has a plan until they get punched in the face. Train yourself, with our help, to adapt to what your league is giving you on draft night.
3. Spend your first four or five picks where being right is worth the most, which the data says is RB and WR, not where a strategy label tells you to.
4. Spend the energy you were spending on draft theory on in-season roster churn, because that is where half your starters come from. And, spoiler alert, it’s why you should ignore bye weeks during your draft.

Limits of this study

These are simulated drafts against ADP bots, not real leagues. The draft simulation has no kickers, defenses, weekly start/sit calls, waivers, trades, or injury replacements. Half PPR covers seven seasons, not ten.

The hit-rate tables use public 12-team PPR ADP as the measure of “top pick,” so they describe how the market’s early picks performed, not how any individual ranking set performed. Players with no scoring season are counted as busts. The turnover and top-3 VOA numbers are year-over-year measurements on realized finishes, so they say how stable a position is, not how well ADP predicted it.