Similar teams for Stanford and Iowa

Similar teams for Stanford and Iowa - Printable Version

+- The CardBoard (https://thecardboard.org/board)
+-- Forum: C-House! (https://thecardboard.org/board/forum-4.html)
+--- Forum: The CardBoard (https://thecardboard.org/board/forum-5.html)
+--- Thread: Similar teams for Stanford and Iowa (/thread-11044.html)

Similar teams for Stanford and Iowa - stupac2 - 12-11-2015

I started doing this in another thread, and realized it might be useful enough to merit its own post. The idea here is to use FEI to find teams that play similarly to both Iowa and Stanford, so we know which games to watch of theirs and which of ours are the most informative to how we should perform. To figure this out, I used three different FEI ratings:

1) Overall. This looks at FEI's total offense, total defense, special teams, and value from field position to find teams that are overall similar. (Note: for all of these I used the raw values instead of rankings, which I think is more meaningful.)

2) Offense. This looks at FEI's values for first down rate, percentage of available yards gained, explosive drive rate, methodical drive rate, and value drive rate.

3) Defense is the same as offense, but given up instead of gained.

I then took every team in the FBS and calculated a sum-of-squares of the differences in these values. You can think of this as finding what teams are nearest to Stanford/Iowa in the N-dimensional phase space of FEI's ratings. It's obviously not a perfect metric, but I think it's useful enough to be worth doing. Results are below, with the similarity ranking as one column and the team as the other. Lower ranking means closer in phase space. I don't give the raw scores because I'm not sure they're very useful, but if someone wants to see the data I can provide it. Also note that despite both teams playing 13 games there's only 12 teams listed, that's because Stanford played USC twice and Iowa played an FCS team that I don't have ratings for.

Teams similar to Iowa that Stanford has faced

Overall

Rank	Team
4	UCLA
17	USC
22	WSU
58	ND
62	California
65	Oregon
69	Arizona
75	Northwestern
81	Washington
85	Colorado
98	OSU
126	UCF

Offense

Rank	Team
7	UCLA
15	Arizona
21	USC
56	Washington
64	California
68	Colorado
78	WSU
96	Oregon
98	ND
103	OSU
107	Northwestern
122	UCF

Defense

Rank	Team
5	Washington
30	ND
31	Northwestern
40	UCLA
64	USC
83	Oregon
89	Arizona
93	Colorado
104	WSU
112	California
117	UCF
125	OSU

Teams similar to Stanford that Iowa has faced

Overall

Rank	Team
22	MSU
28	Nebraska
29	Pittsburgh
31	Indiana
54	Minnesota
58	Illinois
77	Wisconsin
78	Purdue
79	Iowa State
82	Maryland
107	Northwestern
123	North Texas

Offense

Rank	Team
17	MSU
22	Indiana
47	Pittsburgh
56	Nebraska
60	Wisconsin
67	Iowa State
94	Illinois
104	Minnesota
108	Purdue
112	Northwestern
113	Maryland
114	North Texas

Defense

Rank	Team
10	MSU
12	Pittsburgh
28	Nebraska
36	Minnesota
38	Indiana
40	Iowa State
46	Illinois
60	Maryland
102	Purdue
108	Northwestern
111	North Texas
123	Wisconsin

What do we make of this?

I dunno! Maybe this whole exercise was pointless, but I think it should be helpful. A few things that I noticed immediately:

1) The offenses most similar to Iowa are ones we had success stopping. I expected this. The offenses that gave us trouble were Oregon and ND, and Iowa is nothing like them. But the ones they are like we managed to hold to 31 points at most (discounting some of UCLA's garbage time scores against 3rd stringers).

2) Similarly, the defenses most similar to Iowa are ones we had success against, except Northwestern. Besides that debacle the least we scored was 31 against Washington. And note, these two points aren't because these teams are similar to Iowa but worse, for the most part they're similar or better.

3) On the other hand, the teams most like us gave Iowa a problems. MSU is consistently the best comparison and they're the only team to beat Iowa, and they did it with a hurt QB. Also, keep in mind, while these are the offenses closest to us in phase space, that doesn't make them good. The two best offenses they've faced, MSU and Indiana, are 23 and 25 overall on offense, we're #4. Many of these teams put up points against Iowa, Indiana had 27, Pittsburgh had 24, Nebraska 20, despite being mediocre-to-poor offenses overall.

4) Then again, the closest defensive comparisons to us besides MSU all lost, and many of them gave up a ton of points. But for the most part they're also worse than us, with only Minnesota (55 to our 60) grading out as a better overall defense (I'm obviously ignoring MSU who has a legitimately good defense).

So what's my ultimate conclusion? That we should be favored in this game, and we should win. It could easily get out of hand in the second half. However, if we get sloppy or come out flat Iowa has the ability to win, but I think that's what it'll take for them to get the upset. To me it feels like a similar game to USC last week.

Re: Similar teams for Stanford and Iowa - BrewHawk - 12-11-2015

(12-11-2015, 01:15 PM)stupac2 link Wrote:I started doing this in another thread, and realized it might be useful enough to merit its own post. The idea here is to use FEI to find teams that play similarly to both Iowa and Stanford, so we know which games to watch of theirs and which of ours are the most informative to how we should perform. To figure this out, I used three different FEI ratings:

Thanks for the breakdown.Â The only thing I would caution about statistical analysis is that it totally ignores stylistic match-ups which will play a much bigger role than any ppg/ypg can convey.

I'm hoping this will be a slug it out game very similar to our recent BTCCG vs. Michigan St.Â I'm afraid it will be more one-sided favoring Stanford due to the fact that Cook was hindered in that game and see Hogan as a comparable QB who can pick apart our defense.

We have a tendency to give opponents too much credit (especially the lesser teams on our schedule) and play so very close to the vest until we absolutely have to go guns-a-blazin'.Â Not sure why we didn't go downfield more early against the Spartans to loosen up the defense and maybe open up the running game a bit more.

Re: Similar teams for Stanford and Iowa - stupac2 - 12-11-2015

(12-11-2015, 01:43 PM)BrewHawk link Wrote:Thanks for the breakdown.Â The only thing I would caution about statistical analysis is that it totally ignores stylistic match-ups which will play a much bigger role than any ppg/ypg can convey.

Certainly, this analysis is only as useful as the input data. I will note that the FEI metrics aren't quite as simple as ppg/ypg, and I think the offensive/defensive breakdowns are somewhat meaningful (in that two teams that both run "bend don't break" defenses will end up near each other, two offenses that are methodical but not explosive will be near each other, etc), and something like this is probably the best you can do without, you know, watching all the games and really understanding what's happening. Ain't nobody got time for dat.

Quote:I'm hoping this will be a slug it out game very similar to our recent BTCCG vs. Michigan St.Â I'm afraid it will be more one-sided favoring Stanford due to the fact that Cook was hindered in that game and see Hogan as a comparable QB who can pick apart our defense.

That's how I see it. Either it's a very back-and-forth affair where yardage is hard to come by and the game turns on one or two big plays, or we win easily. I just don't see Iowa blowing us out, though if you guys get a couple quick turnovers and scores then maybe the game situation would dictate us changing plans in a way that's doesn't play to our strengths. Who knows?

Quote:We have a tendency to give opponents too much credit (especially the lesser teams on our schedule) and play so very close to the vest until we absolutely have to go guns-a-blazin'.Â Not sure why we didn't go downfield more early against the Spartans to loosen up the defense and maybe open up the running game a bit more.

I'm going to try to watch that game tonight, I assume the replay will be available somewhere, I'll report back my findings.

Re: Similar teams for Stanford and Iowa - BrewHawk - 12-11-2015

(12-11-2015, 01:53 PM)stupac2 link Wrote:Certainly, this analysis is only as useful as the input data. I will note that the FEI metrics aren't quite as simple as ppg/ypg, and I think the offensive/defensive breakdowns are somewhat meaningful (in that two teams that both run "bend don't break" defenses will end up near each other, two offenses that are methodical but not explosive will be near each other, etc), and something like this is probably the best you can do without, you know, watching all the games and really understanding what's happening. Ain't nobody got time for dat.

Understood on the metrics.Â I should have indicated that I'm in the "lies, damn lies and statistics" camp.Â Sample sizes, variables and lack of congruent data prevents any/all statistical analysis from meaning much of anything (ducks).

Let's just agree that there's probably going to be a thin margin of error in this one and hope for a wonderful game.

Re: Similar teams for Stanford and Iowa - OutsiderFan - 12-11-2015

(12-11-2015, 02:21 PM)BrewHawk link Wrote:I should have indicated that I'm in the "lies, damn lies and statistics" camp.Â Sample sizes, variables and lack of congruent data prevents any/all statistical analysis from meaning much of anything (ducks).

Let's just agree that there's probably going to be a thin margin of error in this one and hope for a wonderful game.

LOL, if you don't like numbers and deep statistical analysis, you won't like this board.Â

Re: Similar teams for Stanford and Iowa - Rally - 12-11-2015

(12-11-2015, 03:00 PM)OutsiderFan link Wrote:[quote author=BrewHawk link=topic=14024.msg147543#msg147543 date=1449868894]
I should have indicated that I'm in the "lies, damn lies and statistics" camp.Â Sample sizes, variables and lack of congruent data prevents any/all statistical analysis from meaning much of anything (ducks).

Let's just agree that there's probably going to be a thin margin of error in this one and hope for a wonderful game.

LOL, if you don't like numbers and deep statistical analysis, you won't like this board.
[/quote]

Brewhawk, there is plenty to like for those, like me, who struggled through Statistics 101.Â Try these:

http://thecardboard.org/board/index.php?topic=9866.msg88213#msg88213

http://thecardboard.org/board/index.php?topic=5949.msg43931;topicseen#msg43931

http://thecardboard.org/board/index.php?topic=9890.msg88456#msg88456

Re: Similar teams for Stanford and Iowa - stupac2 - 12-11-2015

(12-11-2015, 02:21 PM)BrewHawk link Wrote:Understood on the metrics.Â I should have indicated that I'm in the "lies, damn lies and statistics" camp.Â Sample sizes, variables and lack of congruent data prevents any/all statistical analysis from meaning much of anything (ducks).

Well, yes and no. The sample size in football is definitely a problem, but there's a difference between "statistics don't tell us everything" and "statistics tell us nothing". I'm assuming (and this is probably unfair to some extent) that if Stanford were the one ranked #19 and Iowa #4 in these advanced metrics, you wouldn't have as big of a problem. The thing is, that typically when a team is ranked a lot higher in polls than they are in advanced stats, there's a reason. Look at Utah, they were a top-ten team early in the year, but fell apart down the stretch because their flaws were exposed.

I don't know if Iowa has the same issues, but I do know that teams that go 5-1 in close games are usually outperforming their talent level. (And Stanford fans speak with some amount of authority here, the 2012 Stanford team punched way above its weight class.) So it's not particularly surprising to me that advanced stats don't love the Hawkeyes just from a first-order look.

That said, my goal in this thread wasn't really to analyze the matchup directly, since you can just compare each team's advanced stats for that (and that's where the ~6.5 spread comes from). Rather, it's to get an idea of which games in each 13-game sample are most informative for how the Rose Bowl could look, so if someone is interested in (say) how Stanford's offense will look against Iowa's defense, you could watch either MSU-Iowa or Stanford-Washington, and continue the analysis from there. That was my main goal.

Quote:Let's just agree that there's probably going to be a thin margin of error in this one and hope for a wonderful game.

I'm with you there.

Re: Similar teams for Stanford and Iowa - fullmetal - 12-11-2015

Funny that Washington is most like Iowa's defense amongst our former opponents.Â That UW defense was pressed into extra service from their 7 drives that ended in punts (~4.3 plays per drive that ended in punt).Â UW's starting QB returned after the Stanford game--we won't see 7 punts from them next year, I don't think.Â Also, Stanford holding the ball for 40:55 against UW surely wore them down...!Â

Iowa's near-goal line stand against MSU at the end of the B1G ccg was pretty impressive.Â A lot of size on that defense, and that's something that really does slow down our offense (esp when Shawngren calls plays that run sideways first).

Re: Similar teams for Stanford and Iowa - BrewHawk - 12-11-2015

(12-11-2015, 03:31 PM)stupac2 link Wrote:Well, yes and no. The sample size in football is definitely a problem, but there's a difference between "statistics don't tell us everything" and "statistics tell us nothing"...

I don't expect you to be aware of this, but Iowa plays pretty much everyone close (good and bad teams alike). The blowouts we've had this year are a bit of an anomaly and reflect the fact that it is one of the better Iowa teams in Ferentz's tenure.

Typically, we lose some of those close games, typically to someone we should beat.Â It's a product of the approach Ferentz takes.Â Get the lead and win the game, style points be damned.

Re: Similar teams for Stanford and Iowa - fullmetal - 12-11-2015

Holy cow, how familiar does *that* sound?Â ???

We might be in for a long game.Â On both sides.

Re: Similar teams for Stanford and Iowa - indiancardinal - 12-12-2015

Didn't Bob Bowlsby hire both Shaw and Ferentz? I thought i saw that posted somewhere here or maybe on the Twitterverse? If that's right then we' ll have BB to thank for turning the RB into Dust Bowl 2016

Re: Similar teams for Stanford and Iowa - stupac2 - 12-12-2015

(12-12-2015, 05:58 AM)indiancardinal link Wrote:Didn't Bob Bowlsby hire both Shaw and Ferentz? I thought i saw that posted somewhere here or maybe on the Twitterverse? If that's right then we' ll have BB to thank for turning the RB into Dust Bowl 2016

Yes, Bowlsby hired both coaches.