You can do some pretty cool things with data analytics. Like you ever seen that movie Moneyball? It’s the story of how a no-budget baseball team outcompeting the deep pockets of the New York Yankees by using the power of data.

Well, you can do the same with football. And this week, I did some pretty cool sports analytics - who is the MOST CLUTCH quarterback of all time? If you know football, you’re thinking it might be Tom Brady or maybe Patrick Mahomes. Well, that might not be the case…

Let me first show you how I did this analysis (and how you can too). I hope you’ll find it as a nice intro to a sports analytics project.

What Even IS Clutch?

I wanted to find the most clutch QB in NFL history, backed by data, not just opinions. So I analyzed over 1.18 million plays from the last 25 years of the NFL.

Before I could analyze anything, I had to define clutch

I decided clutch happens in crunch time: when the game is close (within 7 points) and there's less than 7.5 minutes left in the fourth quarter. Or overtime.

But HOW do you measure clutch? That's where things got interesting.

I ran two completely different analyses.

Analysis #1: "TV Clutch"

For my first approach, I used basic stats you can see on TV: touchdowns, interceptions, completion percentage, and yards gained.

I calculated something called a Z-score for each stat. It just means "how far above or below average is this player?"

A Z-score of 0 means you're average. A score of around 1 means you're above average. A score of around -1 means you're below average.

I combined all the Z-scores into one "TV Clutch" rating.

And when I did that, there’s the top 5 most clutch quarterbacks of all time…

Here's what I got:

5) Kurt Warner
4) Jake Delhomme
3) Andrew Luck
2) Patrick Mahomes
…
1)Tua Tagovailoa??

Yeah.Tua #1. Tom Brady didn't even crack the top 10!

If you know football, you know this list is… questionable, at best. This clutch factor seems to be capturing something a bit odd. So I tried a new clutch factor.

Analysis #2: "Advanced Analytics Clutch"

I used something called Win Probability Added (WPA). WPA measures how much a single play changes your team's chance of winning.

Here's an example:

Scenario 1:

• You're losing. Two seconds left. 17% chance of winning.

• You throw a 70-yard Hail Mary and score. Game over. You win.

• That's a MASSIVE positive WPA because you just flipped the entire game with one play.

Scenario 2:

• You're winning in overtime. 74% chance of winning.

• You throw an interception and lose the game immediately.

• That's a HUGE negative WPA. You just threw away a win.

WPA captures the weight of winning the game. Not just the stats.

When I did that,

Here's what I got:

5) Peyton Manning
4) Patrick Mahomes
3) Ben Roethlisberger
2) Drew Brees
1) Tom Brady

Now that list makes sense.

The cool part is, data analytics helped me create this list. This type of analysis may help teams pick what quarterback they should trade for or pick up in free agency.

And the cool thing is, you can do this type of analysis!

Especially with Sphinx, who was the partner for this issue.

PS: Want to Recreate This Analysis?

I used a tool called Sphinx (a Python copilot that writes code for you in plain English) to pull and analyze all this data. It read API docs, filtered 1.18 million rows, calculated Z-scores, and ranked QBs, all from simple prompts.

Honestly? It saved me hours.

If you want to try this yourself (or build your own analysis with NBA, soccer, or literally anything else), check out Sphinx. It's free to start, and you can grab my Jupyter notebook to play around with.

Maybe you'll find a better definition of "clutch" than I did! 👀