Picking up the Dominos of Seattle's QB Gamble

Breaking down the fantasy and real life implications for Las Vegas, Seattle and Minnesota

I was hoping to write this after JJ McCarthy was officially named the Vikings’ 2025 starting QB so each limb of this tree was solidified. But it seems Aaron Rodgers has every intention of dragging this out, while the Vikings have not refuted their interest. However, we did get some clarity today on Sam Darnold’s contract, with the truly guaranteed money being less than initially reported. That’s reason enough to let it rip on — in my opinion — the most interesting series of moves in this year’s free agency period to discuss.

Just before free agency, Seattle sent quarterback Geno Smith to Las Vegas for a late 3rd-round pick. This move of course brought a new quarterback to the Raiders, but also resulted in the Seahawks signing Sam Darnold which (seemingly) cemented J.J. McCarthy as the Vikings starter in 2025.

Before getting into the specific players at issue here, I think it’s important to discuss the foundation of my beliefs on these players and situations.

I consider EPA (when used as a quarterback stat) to be the democracy of statistics — at least in the Churchillian conception. Essentially, EPA is the worst data-based single measure of quarterback play … except for all the other ones.

As I’ve written — and tweeted — about extensively in the last few years, I do not think EPA does a particularly good job of describing quarterback play, because it is highly influenced by scheme, and the quality of an offense’s skill players.

This should be patently obvious.

EPA per drop back measures the result of each drop back, a play that inevitably depends on the play-call, the protection, the ability of pass-catchers to separate and complete a catch, and — yes — the quarterback to execute the play properly. But to credit that drop back as a quarterback statistic is quite the leap. Essentially, it reduces the performance of 11 players on offense, the 11 players trying to stop them, and each coaching staff, into a referendum on one man: the quarterback.

That doesn’t mean I think EPA is useless. It provides the most meaningfully descriptive valuation of each play when compared to any other statistic we have available. Compared to say, completion percentage, at least EPA directly measures whether a play contributes to winning or not. And I won’t pretend that any other statistic we use to measure quarterback play can claim to being un-affected by the strength of a team’s supporting cast.

My issue with EPA is that because it presents as a “catch all” statistic rather than one which only measures a single aspect of quarterback play, it is highly susceptible to misuse.

I think the appropriate use of the statistic is to view at as what it is — a description of the results of a team’s offense (or passing game) — and a basis from which to examine the ingredients leading to that result. From there, you can examine other statistics and the film to derive reasoning for how we got there and who brings what to the table.

So with that background let’s talk about Geno Smith.

Geno Smith is Much Better Than you Think

I acknowledge that I’m in the 90th percentile of Geno Smith appreciators in the football community. But I believe it is a well- founded position.

Geno Smith was 17th in EPA per drop back in 2024, which presents him as having led an average passing game.

However, a suite of underlying statistics present a more favourable view of his quality of play.

First off, it’s difficult to dispute he was rather unlucky in 2024. He had 15 interceptions, which is rather high, but had a turnover worthy play rate of just 2.6% — which ranked 16th highest among 29 quarterbacks with at least 320 drop backs. Ironically, it was the lowest TWP% of his career, despite having thrown 4 more interceptions than in 2022 of 2023.

Similar to Josh Allen in 2023 or Dak Prescott in 2022, his turnover worthy throws were being converted into actual turnovers at an outlier rate which negatively impacted his EPA, and certainly hurt his perception with the masses.

When you attempt to isolate for drop backs which are less impacted by game script, scheme or surrounding environment, Smith looks much better.

As Benjamin Solak pointed out, in non-play action, non-screen, non-garbage time drop backs, Smith ranked 13th in EPA per drop back, 8th in success rate, 5th in completion percentage over expectation (CPOE), and 1st in explosive pass rate. Smith’s accuracy at all levels of the field has been a constant since he took over the QB1 job in Seattle. Among 35 quarterbacks with 1000 or more drop backs since 2021, Smith ranks 1st in overall CPOE.

Perhaps the most important consideration with evaluating Geno Smith is the effect of pressure. Since becoming a starter in Seattle, Smith faced the highest rate of “quick” pressure (sub-2.5 seconds) of any QB with at least 1000 drop backs.

Quick pressure rate is a better metric to isolate offensive line play than overall pressure rate, because often quarterbacks create their own pressure by holding on to the ball too long, or are subject to late-pressure because their pass catcher cannot uncover fast enough. For his part, Smith’s 2024 time-to-throw of 2.83 seconds was league average. Further, when Smith was pressured, he was 9th in success rate and 8th in PFF passing grade.

It has been a theme in Smith’s career that his EPA on early downs has consistently outpaced his EPA on third-downs, which has an outsized impact on his overall EPA due to the leverage of 3rd down drop backs, but also suggests that he fairs worst in obvious passing situations when it is most difficult to mitigate pass protection flaws schematically.

Essentially, Smith looks like a top-10 QB when you attempt to view his statistical profile as much in a vacuum as possible. He has also performed as a top-10 QB when pressured. But the overall results slide back toward average because he faces pressure at an extremely high rate.

I maintain that if you dropped every QB into a “neutral” situation (such that this is possible), Smith would lead one of the 10 most efficient passing offenses among NFL QBs. In other words, I think Smith is a top-10 “real life” QB as of today. (not factoring in age or expected progression)

If your reflexive comment is to say “well rank them then,” here you go:

(Note: only ranking currently projected starters, who are signed to teams)