Making sense of ESPN’s 7-win FSU football FPI projection

Earlier this week, ESPN released its Football Power Index (FPI) Rankings. FPI is a projection, like any projection system. It gathers statistical inputs, weighs the inputs for each team, and spits out projected results.

And the result for the Florida State Seminoles is 7.3 wins.

On The Nolecast (Apple Podcasts (iTunes), Stitcher, or Google Play Music), I have previously said that I think somewhere between 8-9 regular-season wins is a reasonable expectation for Florida State in Year 1 under new coach Willie Taggart. And I was called a Debbie Downer. So this ESPN projection of 7 wins is seriously pessimistic.

Let’s figure out how FPI got there.

What FPI sees

FPI is going to look at a team’s recent history, and as with all projection systems, the most recent season is going to be weighted the most heavily, followed by 2016, then 2015, 2014, etc. This is the best way of projecting. And FSU’s most recent season was a disaster in pretty much all phases, with injury, a coach quitting on his players and leaving to take another job, etc. But 2016 and 2015 were not incredible campaigns, either, though FSU did make a BCS Bowl in both seasons.

It is also going to look at returning starters and production. Offensively, FSU does not have many huge losses other than Auden Tate and Ryan Izzo. But defensively, FSU is losing its best player at every position — S Derwin James, LB Matthew Thomas, DT Derrick Nnadi, DE Josh Sweat, and CB Tarvarus McFadden (you can quibble with McFadden, but FPI certainly will see a ton of career production). Any reasonable projection system (or human) will project a step back on defense.

So FPI sees a team that wasn’t anything special, losing by far its best player at every defensive position.

And yet, because of how FSU recruits, it still thinks FSU will be one of the 20 best teams in the nation.

Note that I didn’t say top-20 ranking. Because FPI also sees something else: FSU has perhaps the toughest schedule in the nation.

Depending on how you calculate it, FSU’s schedule is going to come out as one of the three or four toughest in the nation. Phil Steele has it at No. 1. FPI has it at No. 2. Take a look, are there more than four games on FSU’s schedule you’d consider to be a lock? Samford and Northern Illinois are for sure, and most would include hosting Wake Forest and a road trip to Syracuse.

Compare FSU’s schedule rating to that of Miami (50) and Florida (55th), and you see it’s entirely possible that FSU could beat both rivals and finish with a lesser record.

So FPI sees FSU as one of the best 20 teams in the nation (18th), and sees it is playing a ridiculous schedule, and projects FSU to do what a Top-20ish team will do against a schedule of this difficulty: Lose 5 games.

Remember, there’s a big difference between a Top-25 team (by record/polling) and a Top-25 team by quality.

What FPI doesn’t see

Projection systems have improved a lot thanks to advanced metrics over the last decade, but there are still things they can’t be expected to calculate, like the impact of a coaching change. A coaching change will not change the talent on a team, but it could change the buy-in, culture, etc., and that could have a meaningful impact on a team.

This is also an element that fans tend to wildly overrate.

But don’t expect Vegas to hang a 7

Vegas uses these projection systems heavily in making its numbers. But it also has humans making adjustments. And in FSU’s case, I expect the adjustment to be a full win or 1.5 wins of helium, to account for the new coach impact and the return of Deondre Francois (who the national pundits think is a star). I expect to see FSU’s over/under win total number in Vegas at 8.5.