Army’s S&P+ Prediction for 2019 Is Too High AND Too Low – AKA, Just Fine

Advanced metrics are a tool. I think we can agree with that.
Regardless of the sport, you will never hear someone who knows their stuff try to shout someone down about analytics replacing the eye test rather than supplementing.
Know what doesn’t help? The opposite – “eye test” people shouting down “analytics nerds” for no real reason other than their not understanding – and not wanting to understand – the advanced data.

John Feinstein is not an idiot. You don’t write and publish three dozen books and make a career as a sports writer by accident. But it would appear that he has shoved his foot as far into his mouth as humanly possible. Let’s unpack what’s wrong here, and then we’ll come back later to watch as John starts crapping out shoelaces.
So of course, what John is referring to is this:

What John does not realize – mostly because someone shared this with him and he didn’t do much but scroll untill he found the service academies – is that this is not a pure ranking like the AP Poll that he votes in.
Army can finish the season ranked 19th in the AP Poll and also finish the season ranked 80th in S&P+ and both can be correct. Both are flawed for different reasons when it comes to Army specifically, but both are not wildly incorrect.
The AP Poll will always under-rank a team like Army because they’re not nationally televised unless they’re playing someone like Oklahoma and they’re rarely nationally relevant outside of the Army/Navy game.
S&P+ always has a hard time accurately ranking the service academies (and Georgia Tech until Paul Johnson retired) because they run a triple option. Believe it or not, when your offense is heavily imbalanced and is one of only about five programs that does it, an algorithm like S&P+ tends to have a hard time accurately assessing it.
Is #19 an accurate ranking for Army? I personally think it’s a little bit low, but it’s very understandable given what I know about the history of AP voters, Army’s recent resurgence, and other factors.
Is Army being 84th in S&P+ an accurate ranking? It’s probably a bit low, but given the fact that option teams are rarely ranked accurately, and that these rankings are intended to be predictive (not reactive like the AP Poll), it’s not insanely low.
According to John:

  1. SBNation is nuts for putting out “pre-season rankings” in February
  2. Army (80th), Air Force (90th) and Navy (118th) are all too low
  3. Their ranking proves that the creator of the rankings doesn’t watch much football

So about that…
Every site on the internet has been releasing “way too early top 25” lists since practically the Saturday after the national championship, so this is actually one of the last “way too early” lists to come out if you want to consider them as such.
Except it isn’t one of those kinds of lists, because again, it’s a predictive metric. Army being 80th in this ranking means that – again, remembering that S&P+ struggles some with option offense teams – considering Army’s 2018 performance, the amount of production they have returning, and the same information about the teams on Army’s 2019 schedule, S&P+ expects that the Black Knights will be statistically slightly better than they were last season.
Of course, this has nothing to do with how much football Bill Connelly does or does not watch – which is quite a bit, actually. S&P+ is an algorithm that he created, where you put numbers in and other numbers come out. The algorithm can’t see and doesn’t care about what is actually going on in the actual game.
2018 Army was a tank that rolled through most of the teams they played. They struggled in their opener against Duke and came very close to upsetting Oklahoma on the road, but also went 4-1 in one-possession games and struggled way too much against Miami (OH), Navy and Colgate.
Despite its flaws, S&P plus is an accurate representation of how Army probably should have done considering their actual statistical performance each week (a nine-ish win team that got a little lucky to get to 11, and could have had 12 with a little more).
Considering what little we know about their returning talent, their incoming recruits, and the same info about their 2019 opponents, Army is probably a 10-ish win team that could go undefeated with some luck.
And again, this is merely a prediction based on what we know right now. What does John want? Army to be 60th? Or 50th? Army is good, but they’re not that good.
Oh, about that whole shoelace thing…

Enjoy that thread.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s