They say the perfect quarterback stat doesn’t exist. Until now.

Just kidding. They (whoever they are) are right.

Sadly, for an analytics nerd like me, there will never be a single, all-encompassing metric that accurately measures and accounts for every aspect of playing quarterback. Football is a messy game with far too many variables for one statistic to ever fully grasp.

However, just because perfection doesn’t exist doesn’t mean we should stop trying to improve the tools we have for evaluating quarterback play.

Power Rankings: Why the Seahawks aren’t No. 1

Chad Graff

And that’s what we’re chasing with our new stat: Quarterback Index, or QBI.

QBI is a metric we’ve created that strips away the contributions from a quarterback’s 10 offensive teammates and credits only what the quarterback provides on a per-play basis. Again, it’s not perfect; no QB stats are. But the hope is that QBI provides a better picture of quarterback quality than the traditional stats (passing touchdowns, passing yards, completion rate) as well as the more advanced metrics (EPA and success rate).

For those familiar, it’s not dissimilar to ESPN’s QBR, which was created to provide one number to “grade” a QB’s performance. We’re looking to do something similar, but we’re putting our own spin on it.

What is QBI?

QBI is a forward-looking quarterback efficiency metric that uses play-by-play data, adjusts for context in each play and is opponent-aware. Put more simply, QBI takes into account passes, rushes, scrambles, sacks, turnovers and anything else that happens when the football is in the quarterback’s hands.

But it goes a step deeper and adjusts for whether the quarterback was pressured, if that pressure was quick (less than 2.5 seconds), as well as the quality of the offensive line and receivers. QBI factors in fumbles in an attempt to take the luck out of recoveries — this can work for and against a quarterback depending on who recovered. QBI also takes into account drops by receivers, as well as yards after the catch and uses an expected YAC model to assign or remove credit from the quarterback. Of course, EPA and success rate are factors in the model, as well. There is still value in knowing that a 12-yard completion on third-and-20 is not as valuable as a 10-yard completion on third-and-10.

Another area I wanted to incorporate in QBI was the combination of analytics and film. The best way to accurately account for what’s happening on the field is a marriage of the two — not one or the other — and that’s where PFF grades are helpful. There are things that happen in football that analytics can’t account for. While this might surprise you, I am happy to admit that.

The last thing to know before you examine the QBI standings from 2025 is my scale of measurement. I’ve taken a page from baseball’s analytics handbook and am using 100 as “average,” similar to baseball’s weighted runs created plus (wRC+) statistic. This means you can expect a QBI of 100 to be average, 101 to be starter-level, 124 to be Pro Bowl-caliber and 133 to be considered an All-Pro.

It’s worth noting that to create that scale, I went back 20 years to create a standard of quarterback play. But more on that in a bit. For now, let’s look back at 2025:

The Caleb conundrum

OK, so many of you just looked at that list and thought to yourself, “How in the world is X ranked Xth?” You probably used a specific name and ranking, but just know that I know you didn’t look at the list and agree with everything you saw. And that’s OK. Stats and rankings are breeding grounds for disagreement. That’s largely why they’re fun.

But to better understand QBI, I wanted to highlight one name who might have stood out above the rest. In this case, I’m talking about the Madden NFL 27 cover athlete ranked all the way down there at No. 23. Why is Caleb Williams ranked so low? Let me explain QBI’s process.

First and foremost, the Chicago Bears quarterback finished 21st in dropback success rate among qualified quarterbacks last year. When a QB rates below-average at consistently manufacturing successful plays, that’s a tough place to start. Then we move on to the supporting cast: QBI really didn’t like how inefficient Williams was despite boasting pristine protection; my model graded the Bears’ offensive line as the top pass-protecting unit in the NFL in 2025.

QBI did like that Williams avoided negative plays, as he limited both interceptions and sacks, but overall, the down-to-down efficiency just wasn’t there. This doesn’t mean QBI won’t grade Williams much higher this year with improved play. In fact, you can make the argument that QBI thinks Williams still has a lot of room to grow, and that should excite Bears fans about how high his ceiling is. But for now, Williams has work to do before QBI considers him a top-10 quarterback.

How else can we use QBI?

As I mentioned, I used historical data to help craft QBI, so I can compare and contrast the quality of quarterback play from year to year. That will help us provide historical context. Secondly, I wanted QBI capable of being predictive, not simply a backward-looking stat that only grades what has already happened.

To that end, I made QBI capable of projecting quarterback play into the future. Again, that means taking into account historical context, not just using one year of data, because as we all know, quarterback play can fluctuate year to year. There’s more background to my methodology, but one interesting note is that when testing out QBI’s projective capabilities, I found it was beneficial to shrink the impact of a quarterback’s rookie season. Suffice it to say, a QB’s rookie year is not always an accurate portrayal of the player that QB will become. For a young QB, their projections might seem regressed to the mean, and that’s on purpose.

As I said, there’s more to QBI projections than that, but we’ll leave the in-depth methodology for another day.

Where will I see QBI?

To start, you’re going to see the 2026 QBI rankings every week in Jeff Howe’s NFL QB stock report, published every Tuesday morning during the season by The Athletic. A warning: Early on, the rankings might look weird. That’s because the sample size will be small. If Jacoby Brissett has two monster games to start the season, you’re going to see the Arizona Cardinals quarterback near the top of the QBI standings. As the season progresses, with more data, the rankings should look more and more accurate.

Beyond the QB stock report … we’ll see. As I said at the top, I’d love for QBI to become the industry standard for evaluating quarterback play. (I’m dreaming big!) Ultimately, the success of QBI will be up to you all. So if you like it, love it or think it’s the worst QB metric you’ve ever seen, let me know! And if you think there are ways to make QBI better, fire away. I’m always looking for ways to improve.

That said, let the QB indexing begin.