QB Draft Timing, Tested: Early, Mid-Round, or Punt?

Every draft season the QB debate resurfaces: reach for one early and lock down a passing floor, or let the position come to you and spend those picks on running backs and receivers instead? We took three specific answers to that question and ran each one through a batch sim - a tool that drafts an entire league of bots (one seat running the strategy being tested, the rest running normal draft behavior) over and over, thousands of times, across six real NFL seasons, then plays out the fantasy season and records how each team actually finished. Do that a million times per strategy and the noise washes out, leaving a real signal about whether a strategy helps, hurts, or does nothing.

The three builds: qb-early never touches a QB in round 1, then leans hard into one in rounds 2-3 (pulling from the top 3 QBs still on the board), and is guaranteed to have one in hand by round 3 if that lean hasn't gotten there already. mid-round-qb won't touch a QB before round 5, then pushes hard for one through round 8. qb-punt refuses to draft a QB before round 8, full stop. All three are otherwise identical to a plain ADP-driven baseline - same "best player available, 80/20 coin flip between the top two options" logic everywhere else.

Situational. All three builds are situational in their own way, but not for the same reason. qb-early's defining trait is volatility - it isn't a flat "always worse than ADP" build, it's a build whose results swing harder by season than either of the others. 2022 was its best year by a wide margin; 2025 was its worst by an even wider one. Its downside also isn't fixed across scoring formats - switch to half-PPR and qb-early's gap vs. baseline on Top-1 and Top-4 nearly disappears, with Champion narrowing too, likely because half-PPR lowers the per-catch value of every skill position (RB, WR, and TE all earn less per reception), which makes the skill-position pick the strategy gives up for a round-2/3 QB cheaper to sacrifice (more in the settings section below). mid-round-qb is the closest thing to No Meaningful Difference - it tracks nearest to plain ADP in almost every season, rarely drifting far from baseline regardless of year. qb-punt has looked strong lately, especially in 2025 - its late-QB approach has outpaced baseline's Champion rate in three straight recent seasons, 2025 most of all, even though it still trails on making the playoffs in the first place. Which one looks best depends heavily on which season (and which scoring format) you land in - and the year-to-year swing itself turns out to come down to a simple, testable question: was the top QB tier that year actually elite, and did preseason ADP correctly identify who was in it? More on that below.

The pooled numbers

Across all six seasons combined, here's how each build stacks up against the plain-ADP baseline (12-team, PPR):

Pooled across six seasons, qb-early trails baseline on every tier - down about 1.3 points on Top-4, a full point on Champion, and worse on every downside measure too. But a six-season average flattens out exactly the kind of year-to-year swing that turns out to matter most for this build, which the by-year breakdown below makes obvious.

mid-round-qb is basically a wash pooled - nothing on the headline tiers moves by more than about a fifth of a percentage point, and the downside numbers barely budge either.

qb-punt splits down the middle pooled: Top-1 and Top-4 come in a bit behind baseline, but Champion jumps to 9.86% against baseline's 9.01%. That gap is not evenly spread across the six seasons - it's concentrated recently, as the next section shows.

By year, the story splits hard

This is where qb-early's real story lives. 2022 is its best season by a wide margin - Top-4 hits 40.49%, actually ahead of baseline's 38.21% for that year. 2025 is its worst by an even wider margin - Top-4 falls to 26.87%, well below baseline's 33.02%, and its Champion rate is barely half of baseline's for that season. That's the single largest season-to-season range of any build in this comparison, and it lands entirely on qb-early: no other strategy here swings anywhere close to that hard between its best and worst year.

mid-round-qb, by contrast, is the most consistent of the three. In four of six seasons it sits within a couple of points of baseline on both tiers, tracking close to what plain ADP would have produced anyway. 2021 is its one real outlier (Top-4 jumps to 37.91% against baseline's 34.63%, while Champion drops to 7.50% against baseline's 8.71%) - a split we don't have an equivalent explanation for (see caveats).

qb-punt's story is about recent form rather than consistency. Its Champion rate has beaten baseline in the three most recent seasons in the sample (2023-2025), and 2025 is where that edge is largest by far - 13.35% against baseline's 8.64%. In the three older seasons (2020-2022) the picture is more mixed, with qb-punt ahead of baseline on Champion in 2020 but behind in 2021 and 2022.

Why the swings: a two-factor explanation

Two things have to be true for reaching for a QB early to actually pay off: the top QB tier that year has to be meaningfully better than replacement level, and preseason ADP has to correctly identify who's in that tier. 2022 had both. 2025 had neither. That difference explains almost every pattern above. One distinction worth being clear about: this explanation draws on real-world 2022 and 2025 NFL outcomes - points-per-game finishes, preseason ADP ranks, injury reports - that live outside this project's own simulation data. It's our read of why the real seasons behind these sims produced the results they did; it isn't something computed from summary.json the way the pooled and by-year numbers above are.

2022: the elite tier delivered, and everyone else broke. 2022 is the cleanest "pay up" season in the sample. Patrick Mahomes finished as the No. 1 fantasy QB at 28.4 points per game, with Josh Allen right behind at 28.1 and Jalen Hurts at 27.3 - and critically, Allen and Mahomes were also the top two QBs by preseason ADP. qb-early's rule (pull from the top 3 available QBs in rounds 2-3) would have landed one of those two in most sims. Meanwhile, the pool the other builds were fishing in later got wrecked: Kyler Murray, Lamar Jackson, Dak Prescott, Tua Tagovailoa, Trey Lance, Hurts, and Justin Fields all got hurt at various points that season, and Russell Wilson collapsed outright - he was QB20 at midseason, with Jared Goff outscoring him. A round-2 pick on Allen or Mahomes bought a real, double-digit-points-per-week edge over what a punt team was starting most weeks. That's the whole story behind qb-early beating baseline on Top-4 in 2022, and behind qb-punt posting its worst season in the sample (32.66%).

2025: ADP was wrong at the top, and right nowhere else. 2025 inverts every part of that setup. Here's what the two builds were actually drafting that year, by real-world preseason ADP rank versus how those players actually finished:

Seven of the 19 quarterbacks drafted ahead of Trevor Lawrence by ADP that year - Lamar Jackson, Jayden Daniels, Joe Burrow, Kyler Murray, Brock Purdy, Justin Fields, and J.J. McCarthy - played 12 or fewer games. Maye finished QB2 off a QB18 ADP. Stafford finished QB3 off a QB23 ADP. Lawrence finished QB4 off a QB20 ADP. Jackson fell all the way to QB20 production from a QB2 ADP. So qb-early was burning a round-2 pick on a bust roughly two times out of three that year, while qb-punt got top-four QB production essentially for free and kept every bit of its early-round WR/RB capital intact. That's why 2025 is the only season in the sample where qb-punt beats baseline on Top-4 (38.55% vs. 33.02%), not just Champion - it wasn't variance, it was a free lunch that plain ADP (and qb-early, which leans directly on ADP's QB ranking) completely missed.

Is qb-punt's recent edge durable, or a 2025 fluke? Some of both. 2024 was a milder version of the same pattern - eight of the top 10 QBs by points per game finished above their preseason ADP, five of the top 14 were drafted QB15 or later, and four quarterbacks taken in the QB21-32 range finished top-15. The tier gap itself has also been shrinking: in 2025 the QB1 by points per game averaged 22.0 points and the QB12 averaged 18.0 - a 4-point gap - while the first QB came off the board 23rd overall and the QB12 went 102nd. Part of that looks structural: quarterbacks rushed for their fewest points per game in 2025 since 2021, and rushing production is exactly what created the wide, real spread between elite and replacement-level QBs back in 2022. If that trend holds, the case for reaching early gets weaker going forward rather than stronger - but 2025 specifically was also an unusually brutal injury year concentrated on exactly the quarterbacks ADP loved most, which is as much a one-year event as a structural shift.

League settings and strategy context

The primary comparison ran 12-team, PPR leagues across 2020-2025, 1,000,000 simulations per strategy, randomized draft slot each sim. Seed matching is mixed: qb-early and mid-round-qb share a fixed named seed (ppr-12team-1m), so those two are a true matched-pairs comparison against each other. The baseline and qb-punt are each on independent auto-generated seeds, so any comparison involving either of those two is an independent-samples comparison rather than a matched one. At a million sims per strategy the samples are large enough that the gaps reported above are unlikely to be pure noise, but a fully shared seed across all four runs would have isolated each strategy's effect even more cleanly.

• adp-baseline: "Best available by ADP throughout, with a role-fill boost from round 9. Always picks between the top 2 candidates by score, 80/20."
• qb-early: "Never drafts QB in round 1; heavily favors it in rounds 2-3, drawn from the top 3 available QBs, guaranteed by round 3 if not already drafted. Otherwise ADP Baseline."
• mid-round-qb ("Mid-Round QB (test)"): "ADP Baseline, but avoids QB before round 5 and boosts it hard through round 8. Always picks between the top 2 candidates by score, 80/20."
• qb-punt: "ADP Baseline, but doesn't draft a QB before round 8. Always picks between the top 2 candidates by score, 80/20."

This comparison exists because these three strategies are three distinct, common answers to "when should I draft my QB," and running them against a shared baseline in one batch makes it possible to see how they stack up not just against plain ADP but against each other. As with every batch sim run through this tool, the opponent seats are drawn from production bot archetypes - simplified statistical personas, not a guarantee about how real leaguemates would draft or react.

Does this hold up elsewhere?

We also ran all four builds under three settings variants to see whether the pattern holds.

12-team, half-PPR (all four runs share one seed, half-ppr-12team-1m, so this is a true matched comparison): switching to half-PPR shrinks qb-early's pooled deficit noticeably. In PPR it trails baseline by 0.36pp on Top-1, 1.27pp on Top-4, and 1.02pp on Champion; in half-PPR the Top-1 and Top-4 gaps essentially vanish (+0.07pp and +0.10pp) while Champion narrows to -0.64pp. Our best read: the strategy's rounds-2-3 QB reach is displacing whatever a bot would otherwise take in that range - typically a skill-position player, RB, WR, or TE - and half-PPR lowers the per-reception value of all three, not just receivers. That makes the pick qb-early gives up cheaper to sacrifice in half-PPR than in full PPR, which tracks with the shrinking gap. We can't verify the exact position displaced in every sim, so treat that as our best explanation rather than confirmed fact. mid-round-qb stays a wash in both modes. qb-punt goes the other direction: its Top-1/Top-4 deficit gets worse under half-PPR (-1.35pp/-3.22pp, versus -0.26pp/-0.44pp in PPR), though its Champion edge survives in smaller form (+0.24pp vs. +0.85pp). We don't have a clean explanation for why qb-punt's downside widens rather than narrows here, so we're flagging that as an open pattern too.

10-team (PPR and half-PPR): at 10 teams, PPR, qb-early's pooled deficit gets more pronounced (Top-1 -0.76pp, Top-4 -1.90pp, Champion -0.89pp), mid-round-qb stays a mild negative on Top-1/Top-4 (-0.57pp/-1.36pp) but dead-even on Champion, and qb-punt repeats its now-familiar shape (Top-1/Top-4 down, Champion up 1.13pp). Under 10-team half-PPR (fully matched seed, half-ppr-10team-1m, across all four), the same shapes hold: qb-early down across the board (-0.60pp/-0.95pp/-0.33pp), mid-round-qb a mild wash (-0.55pp/-0.76pp/-0.13pp), and qb-punt's biggest swing of any setting we tested - Top-4 down 4.26pp, but Champion still up 0.65pp. Across four different settings and three different seed setups, qb-punt's Champion-tier edge never flips sign, which lines up with the recent-form pattern seen in the primary comparison.

Caveats

This is simulated drafting against bot opponents running fixed archetypes, not real human leagues - real drafters adapt, panic, and reach in ways a bot doesn't. The "80/20 top-two-candidates" logic behind every persona here (including the baseline) is itself a simplification of how ADP-following drafters actually behave. The pooled and by-year numbers throughout this report are read directly from each run's summary.json and are solid. The explanation for why 2022 and 2025 diverge the way they do, on the other hand, draws on real-world 2022 and 2025 NFL outcomes - points-per-game finishes, preseason ADP ranks, and injury reports - that live entirely outside this project's simulation data; it explains why the simulated draft rules produced the patterns they did, but it isn't itself something computed from summary.json. mid-round-qb's 2021 outlier (Top-4 up, Champion down, both away from baseline) remains unexplained - we didn't find an equivalent real-world driver for that one, so treat it as an open question rather than a settled finding.