What is clutch? It’s a sports question as old as the duality of winners and losers, a status that exists outside of the more measurable box score tangibles like passing yards and interceptions. To paraphrase Potter Stewart, most people know clutch when they see it, even if they’re not equipped to properly define it in polite discourse (or in matters concerning the Supreme Court). This gets concerning pretty quickly, particularly when it comes to matters of ranking, the natural mode for most sports-inclined folks.

Luckily for us — and for anyone who’s interested in these sorts of things — the fine people over at FiveThirtyEight have come up with a way to examine just how clutch a player can be in the playoffs, when the clutch is really present. How did they do it? Essentially, they took each of the nominal starters since the NFL really became the NFL in 1970, then measured the QB’s performance against a replacement level signal caller’s expected performance. “Replacement level,” in this case, essentially meaning an average NFL quarterback for that particular season. The results are a mixture of the expected and the totally befuddling, but we’ll get to that momentarily.

5. John Elway

Boasting a lifetime playoff record of 14 wins to seven losses, Elway’s replacement level simulation would have graded out to roughly nine wins and 12 losses (the hard numbers are 8.9 wins and 12.1 losses, because that’s the nature of the evaluation). Furthermore, there’s roughly a 1% chance that the average quarterback would have replicated Elway’s feats. If a working definition of “clutch” is to describe players who do things that their lesser contemporaries can’t, this is looking like a pretty good system so far.

4. Joe Flacco

Weirdly, Flacco places higher on this list than Tom Brady (who finished sixth), despite the fact that the Patriots dispatched the Ravens in the most recent postseason. How does this work? Prior to this postseason, Brady went 18 wins to 8 losses against the average of about 13 and 13 (hard numbers: 12.6 wins to 13.4 losses). Flacco’s 10 and five record is coming against an expectation of nearly the opposite — five wins and 10 losses. In other words, it’d be less likely for a faceless, anonymous quarterback to do what Flacco did than to do what Tom Brady did in the postseason (and here we remember Matt Cassell in 2008), so Flacco’s arguably more clutch, especially when you consider that the Pats were the favorites in most of their games, hurting Brady’s overall score. Make sense? Don’t get deflated over it.

3. Kurt Warner

A replacement level quarterback would have lead Kurt Warner’s playoff teams to a four win, nine loss record. That is exactly the opposite of what the real Kurt Warner did over the course of his NFL career, when he lead his teams to nine wins and four losses, something that, running these simulations over and over again, has a 0.3% chance of happening. Yes, that is three-tenths of one percent. Kurt Warner: Pretty clutch.

Again, these are the conventional wisdom guys — Kurt Warner and Joe Montana were clutch quarterbacks, the sun rises in the East, the planets rotate around the Sun, and so on. That doesn’t make them any less impressive. Montana went 16 wins against five losses during his playoff tenure: The average player would’ve gone 10-11.

1. Eli Manning

This is not a typo. You might be thinking “Hey, where’s Peyton? What is this list?! It’s all bullshit!” Well, Peyton grades out at number 28 on the list (seriously, you can read it here). As for the reasoning behind it, we’ll leave this one to Nate Silver, who did the original research for FiveThirtyEight.

“[Manning’s] New York Giants have often been underdogs in the postseason and projected to a record of 4-7 or perhaps 5-6 in his 11 games. Instead, Eli Manning’s teams have gone 8-3.” Silver continues to reevaluate in his article — because there’s no way Eli comes out on top again, right? — and, ultimately, the younger Manning still comes out on top.