Is a Left-Handed Quarterback Worth Anything Over a Right-Handed One?

Short answer: there's no measurable value either way, and the data can't settle it. Handedness produces one large, specific, measurable difference — lefties throw to the left side of the field much more — but that difference doesn't turn into efficiency, and receivers don't drop their passes more.

Regular season 2006–2025, nflverse play-by-play (392,554 dropbacks); FTN charting 2022–2025 for drops. Reproduce: python3 backtests/qb_handedness/study.py.

The structural problem

Handedness isn't in any nflverse feed, so the left-handed set is curated: Michael Vick, Mark Brunell, Jake Plummer, Tim Tebow, Matt Leinart, Kellen Moore, Tua Tagovailoa, Michael Penix Jr. — ids in study.py, all eight found in the data. Everyone else is treated as right-handed. A missed low-volume lefty would sit among ~170 comparison QBs and change nothing, and nobody is in doubt about the lefties with 300+ attempts in this span.

That's also the problem. "Lefties vs righties" at the player level is really "these seven players vs everyone else", so any headline number is mostly who they are. Tua is at the 92nd percentile and Tebow at the 18th; neither fact is about his arm. The honest design separates that from mechanical questions where player quality cancels out.

1. Is a lefty simply better? — no detectable difference

179 QBs with ≥300 dropbacks, 7 left-handed. The unit is the player.

In points (EPA is already in points): over ~600 dropbacks a season, the per-player interval runs from about −25 to +50 points a season. For scale, ~0.1 EPA/dropback is roughly the gap between a median starter and a top-five one, and this interval is wider than that. The data can't tell "handedness is worth nothing" from "it's worth a lot" in either direction. Seven players can't answer this question, and no amount of re-cutting them will.

That spans the whole distribution, which is what you'd expect if handedness isn't the thing that matters.

2. Where they throw — a large, specific, measurable difference

Share of attempts to the left minus share to the right (the side is from the offense's view, as scored in play-by-play):

Righties lean slightly right; lefties lean further left than righties lean right. By left-leaning, of 175 QBs: Leinart 1st, Vick 2nd, Tebow 4th, Tua 7th, Penix 10th, Plummer 39th.

It's the arm, not the scheme. Lefties sit on few teams, so the obvious worry is that this is Miami's or Atlanta's playbook. Comparing each lefty with a right-handed teammate in the same team-season — same receivers, same coordinator — the lefty threw further left in 23 of 26 pairs (weighted gap +0.153, CI +0.106 to +0.197). Collapsed to one comparison per team-season, so a lefty facing three teammates counts once: 19 of 21, sign test p = 0.0002. Examples: Vick +0.114 vs Kolb −0.230 (2010 PHI); Tua +0.119 vs Brissett −0.089 (2021 MIA).

In plain terms, a left-handed QB moves about 6–7% of his throws from the right side of the field to the left. It's the clearest real effect of handedness here.

3. Arm side — no, and the side effect that does exist is field-wide

The intuition: a QB is better throwing to his own arm side. Tested within each QB (his left-side EPA per attempt minus his right-side), so good and bad QBs are each compared with themselves.

If arm side mattered, lefties' gap would have the opposite sign to righties'. It doesn't: both groups are better throwing left. So the side effect is field-wide, not about handedness. Measured the arm-side way, QBs are actually slightly worse to their arm side (−0.033), only because nearly all of them are right-handed and do better on their off side. The effect doesn't appear in completion over expected (CPOE +0.3, CI −0.2 to +0.9), so it isn't accuracy. Corrected by §6: pooled over every attempt, CPOE is +0.7 (CI +0.3 to +1.0) — the per-QB test was nullnull resultA test that found nothing. "Null" is the starting assumption that there is no real effect; a "null result" means the data gave us no reason to abandon that assumption. It does not mean the data was missing or the test failed to run. means "we could not tell", which is very different from "there is nothing there".">underpoweredunderpoweredNot enough data to detect an effect even if it is really there. An underpowered null means "we could not tell", which is very different from "there is nothing there"..

Putting 2 and 3 together gives the only concrete "worth" figure: leaning ~6–7% of throws toward the slightly better side is worth about 0.065 × 0.037 × 550 ≈ one point a season. Nothing.

4. Drops — no penalty detected

The ball spins the opposite way from what receivers catch all week, so a lefty might get more drops. FTN charting, catchable balls, 2022–25, compared within each receiver who caught at least 15 catchable balls from each hand (19 receivers: 1,377 targets from lefties, 1,641 from righties — mostly the Miami and Atlanta receivers with Tua or Penix and a right-handed backup or Cousins).

No detectable drop penalty. With a 5% base rate, the upper end of the interval still allows up to ~35% more drops, so a modest effect can't be ruled out — this sample can't see one. Receivers often got their lefty and righty targets in different games and contexts; drops are mostly the receiver, which helps, but it isn't a controlled experiment.

What this doesn't measure

• Blind side. A lefty's blind side is his right, so teams move their best pass-protecting tackle over. Play-by-play doesn't record the direction a pressure or sack came from, so the cost or benefit of that can't be tested here.
• Defensive adjustment. If defences shade toward a lefty's favoured side, that would show up as the side effect in section 3. It doesn't, but a tiny lefty sample can't separate adjustment from noise.
• Price. There are too few lefties to estimate a contract premium or discount.

Bottom line

Handedness changes where a quarterback throws — reliably, by about six or seven percent of attempts, and it holds up against same-team teammates. It doesn't measurably change how well he throws, whether he does better to his arm side, or how often his receivers drop the ball. As a factor in valuing a quarterback, it's noise next to the quarterback himself.

6. Why the left side is better (left_side.py, left_side_receivers.py)

Every located pass 2006–2025, regular season, no sacks or two-point tries (269,101 thrown left or right). CIs resample whole games.

• It's real and steady. Left minus right is +0.041 EPA per attempt (CI +0.030 to +0.053), better on the left in 19 of 20 seasons.
• It isn't play mix. Holding depth, down, distance, field position, shotgun, score, quarter and season fixed leaves +0.042 (CI +0.030 to +0.053). It appears at every depth except 20+ air yards, where it is −0.02 (CI −0.07 to +0.02).
• It's in the throw and after the catch, about evenly. Air EPA is +0.023 and YAC EPA is +0.019. Completion is +1.5 points and CPOE is +0.7, both with CIs above zero. Left throws also travel 0.2 yards further.
• About 40% of it is who is targeted. Tight ends are 22% of right-side targets and 18% of left-side ones. That fits the default formation, with the tight end on the right, and TE targets are worth less. Within each position the gap shrinks to WR +0.020 (CI −0.002 to +0.043), TE +0.045 and RB +0.026. Those average about +0.027 against +0.044 overall (2016–25).
• It isn't the #1 receiver. Each team's top target does sit slightly more on the left (23.5% vs 21.6%), but excluding him the gap is larger, not smaller (+0.047).
• It isn't defensive shading. Across 177 QBs, how hard a passer leans right has no relation to his left advantage (r = −0.003, CI −0.14 to +0.14). Across 640 defense-seasons, facing more right-side throws has none either (r = −0.01).
• It isn't a scorer artifact. The gap barely varies by stadium beyond noise: the SD is 0.047 against about 0.046 expected from noise alone, and the left is ahead in 25 of 32 stadiums.

Verdict. About two-fifths of the gap is formation (tight ends on the right). The rest is a small, persistent advantage for left-side throws that none of this explains. Defenses may be slightly weaker on the offense's left, which is the defense's right, but these data can't show it. No betting or fantasy use: it is league-wide, so it is already in every baseline.