ESPN's FPI 2020

[Quincy A. Wagstaff]

I haven't been on here much without sports, so sorry if this has already been discussed:

ESPN's FPI has us 13th in the SEC in 2020, winning 4-5 games. Any idea why we are so low? Why from 2019 to 2020 would we fall below Ole Miss and Arkansas? I don't expect us to compete for the West or anything, but I'm pretty confident in us extending our bowl streak.

 

[patdog]

I put zero stock into FPI, both when it ranks us highly like it did the last couple of years and when it ranks us lowly. Garbage in = Garbage out.

 

[Seinfeld]

Like you, I don’t have grand expectations from this team right out of the gate, but 4-5 wins would be an unmitigated disaster.

Just looking at the schedule, a 4-8 record indicates that we’re going 2-4 against the group of NCSt, Tulane, Ark, Miz, KY, and OM. It’s not impossible per se, but if that’s all we’ve got after adding Costello and Heath while getting Hill back for another year, I’ll have some serious concerns. If have no idea as to why ESPN would come up with that

 

[My Bru]

I put zero stock into FPI, both when it ranks us highly like it did the last couple of years and when it ranks us lowly. Garbage in = Garbage out.

It gives you a good starting point to argue from, as does S&P, or any other 'strength' metric. It was easy to see last, especially in hindsight, that Moorhead pulled us down and had us playing underneath our talent. Plus he was transitioning an offense. It's no surprise that we underachieved, same for 2018.

A lot of those metrics had us higher in 2016 too, and you could see why. We lose a couple games to inferior teams due to dumb issues, then turn around and beat some good teams once we figured it out.

I'm not sure why we'd be that low this year, but I'm not discounting it either. They are probably looking at defensive ratings, and may not be considering QB, who knows. But they aren't fake news or anything, it's based on real data.

 

[Drebin]

I haven't been on here much without sports, so sorry if this has already been discussed:

ESPN's FPI has us 13th in the SEC in 2020, winning 4-5 games. Any idea why we are so low? Why from 2019 to 2020 would we fall below Ole Miss and Arkansas? I don't expect us to compete for the West or anything, but I'm pretty confident in us extending our bowl streak.

Maybe it will be as wrong as it was a couple of years ago when it had us favored in every game of the year except Alabama.

The FPI is garbage.

 

[Quincy A. Wagstaff]

It's not perfect, but it largely hits on its projections when you zoom out. Sure, it get's a lot of individual games wrong, and it misses on a handful of teams each year, but the larger the sample size the more accurately it predicts win totals +/- 1 game.

It just seems odd that it rates us so low relative to other teams we played last year. It has Ole Miss slightly better than last year. I can buy that. But it has Arkansas from a touchdown below FBS average last year (-7.0) to slightly above FBS average in 2020 (+1.3). And it has us from a touchdown better than FBS average last year (+7.1) to slightly below FBS average in 2020 (-0.7). I just don't see how we got 8 points/game worse and Arkansas got 8 points/game better. Maybe it hasn't factored in things like Costello yet. It will be interesting to see what it is opening day.

 

[engie]

It's judging Costello strictly on last season, when he was injured the whole time basically meaning he doesn't factor into the number at all. Change that to his soph input and adjust it for Leach's traditional overperformance and it's a whole different deal.

 

[engie]

going 2-4 against the group of NCSt, Tulane, Ark, Miz, KY, and OM.

We should go at least 5-1 against this stretch. Anything less than 4-2 here would be an unmitigated disaster IMO

 

[johnson86-1]

It's not perfect, but it largely hits on its projections when you zoom out. Sure, it get's a lot of individual games wrong, and it misses on a handful of teams each year, but the larger the sample size the more accurately it predicts win totals +/- 1 game.

It just seems odd that it rates us so low relative to other teams we played last year. It has Ole Miss slightly better than last year. I can buy that. But it has Arkansas from a touchdown below FBS average last year (-7.0) to slightly above FBS average in 2020 (+1.3). And it has us from a touchdown better than FBS average last year (+7.1) to slightly below FBS average in 2020 (-0.7). I just don't see how we got 8 points/game worse and Arkansas got 8 points/game better. Maybe it hasn't factored in things like Costello yet. It will be interesting to see what it is opening day.

Didn't MOrris have a good recruiting class or two while he was there? I may be thinking of the wrong index, but I think they take recruiting rankings into effect, so they may be getting a decent sized bump from people who haven't really produced yet, but should if you believe the recruiting rankings.

 

[patdog]

Well for one thing, we clearly weren't anywhere near as good as FPI thought we were last year. And yeah, we did bring in Costello, but we also lost a lot off of a 6-7 team, have a new coach, will be implementing completely new offensive and defensive schemes with limited spring/summer practice time. It's not a big stretch to predict 4-5 wins for us this fall. I personally think we're a 5-6 win team, but it really is anybody's guess for our team this year.

 

[Go Budaw]

It may be that they have looked at how young we are going to be this year. A successful SEC football team (especially in the West) has probably 80% of its two deep as 3rd and 4th year players unless you are recruiting at Bama’s level. For us, that’s the 2017-2018 signing classes. We SHOULD have a minimum of 20-25 guys we can count on from those two classes. I count maybe 12 likely contributors at a high level for us from those two classes combined that won’t be just warm bodies (25% of the two-deep). The 2017 class was talented, but loaded with JUCO’s and other early NFL guys who are now gone, and the 2018 class was simply not a good one (thanks Dan). We are paying the price right now for Dan selling out for success in 2017 so that he could move on.

On top of all that, you also have a complete staff and culture overhaul taking place that could easily have short term ramifications in the W-L record. KJ coming in will help and we might be able to outscore enough teams to get to 6-6 or 7-5, but I think we are going to give up a **** ton of points too. We’ll see how it plays out,

 

[johnson86-1]

Well for one thing, we clearly weren't anywhere near as good as FPI thought we were last year. And yeah, we did bring in Costello, but we also lost a lot off of a 6-7 team, have a new coach, will be implementing completely new offensive and defensive schemes with limited spring/summer practice time. It's not a big stretch to predict 4-5 wins for us this fall. I personally think we're a 5-6 win team, but it really is anybody's guess for our team this year.

I don't think FPI takes any of that into account. We were overrated for the past two years because they didn't take into account how bad Moorhead was and instead relied on past performance, and now we will be underrated for the same reason. Still not expecting anything more than keeping the bowl streak alive this year because of the coaching transition combined with losing spring, but we weren't as bad as we looked last year, at least on offense, and on defense we will get dinged for losing a lot of people by the FPI that weren't playing for much of the season anyway.

 

[Megiddo]

FPI is a hilariously bad metric for preseason college football.

In 2018 ESPN had picked Washington State to win between 5 and 6 football games. In 2018 Washington State won 11 football games. Outside of the top 10, the analysts are throwing darts at a board in preseason (and picking favorable spots to the larger fanbases of course). If the analysts did even the slightest bit of research, they would have saw that Washington State won 9, 8, and 9 games in 2015, 2016, and 2017 respectively, so their picking between 5 to 6 wins at preseason was just pure idiocy. Pay it no mind.

 

[My Bru]

FPI is a hilariously bad metric for preseason college football.

In 2018 ESPN had picked Washington State to win between 5 and 6 football games. In 2018 Washington State won 11 football games. Outside of the top 10, the analysts are throwing darts at a board in preseason (and picking favorable spots to the larger fanbases of course). If the analysts did even the slightest bit of research, they would have saw that Washington State won 9, 8, and 9 games in 2015, 2016, and 2017 respectively, so their picking between 5 to 6 wins at preseason was just pure idiocy. Pay it no mind.

Congratulations on finding one random exception. FPI will generally be more accurate than any of us....that is fact.

https://thespun.com/college-football/espn-releases-2018-preseason-football-power-index

I mean who won the national championship again?

 

[VegasDawg13]

I don't think FPI takes any of that into account.

It does.

"In the preseason, these components are made up entirely of data from previous seasons, such as returning starters, past performance, recruiting rankings and coaching tenure."

https://www.espn.com/blog/statsinfo/post/_/id/122612/an-inside-look-at-college-fpi

But still, betting markets are more predictive than FPI, so it really serves little purpose. I believe betting markets have us at 6.5 wins, at least they did when they first came out a month or two ago.

 

[Megiddo]

Congratulations on finding one random exception. FPI will generally be more accurate than any of us....that is fact.

https://thespun.com/college-football/espn-releases-2018-preseason-football-power-index

I mean who won the national championship again?

Did you just gloss over where I said "outside of the top 10"?

I'll assume that you did because otherwise your post is baffling to me.

It's not hard to pick which top tier teams will dominate 95% of the other FBS teams. So sure, if they're picking the top 10, it's going to be pretty accurate. The rest of them though? That's a crapshoot.

 

[Quincy A. Wagstaff]

"All models are wrong, but some are useful."

 

[johnson86-1]

It does.

"In the preseason, these components are made up entirely of data from previous seasons, such as returning starters, past performance, recruiting rankings and coaching tenure."

https://www.espn.com/blog/statsinfo/post/_/id/122612/an-inside-look-at-college-fpi

But still, betting markets are more predictive than FPI, so it really serves little purpose. I believe betting markets have us at 6.5 wins, at least they did when they first came out a month or two ago.

So are we actually getting dinged for moving from Moorhead to Leach??? That's hilarious.

Or does FPI track and rate coaches also?

 

[mstateglfr]

So who is this new guy from Huxley College?

 

[RockKickinSandPoundinSOB]

I haven't been on here much without sports, so sorry if this has already been discussed:

ESPN's FPI has us 13th in the SEC in 2020, winning 4-5 games. Any idea why we are so low? Why from 2019 to 2020 would we fall below Ole Miss and Arkansas? I don't expect us to compete for the West or anything, but I'm pretty confident in us extending our bowl streak.

I wish Vegas would take their Season win over/under line based on that so I can make a bunch on the over.