Greg Harmon

CFB analytics without an agenda (unless its ND). I just run the numbers. By Greg Harmon. ☘️ ND ’17

[Greg Harmon @BarGraphCFB]

The ND discourse has been all process - rankings shows, head-to-head, ACC politics.

All worth debating. But somewhere in the noise, a simpler question got lost: Was Notre Dame actually a top 10 team?

The data says yes. It's not close.

https://t.co/2Q1w8HBM7D

[Greg Harmon @BarGraphCFB]

@RickRunGood @PoolGenius @splashsports Rick - big fan. Are you sure your odds are right? They seem to differ materially from KenPoms site.

https://t.co/0WfbZhfWpS

[Greg Harmon @BarGraphCFB]

@emollick Anthropic has to know that right? I view more as a marketing tactic. “Cowork does private equity” is really just a signal to get people in industry (who generally live under a rock) to understand cowork exists. Their goal is just to be close enough to create the “oh shit” moment

[Greg Harmon @BarGraphCFB]

@tmj6810 @talia_baia 7 of the 12 playoff teams did not beat a playoff team. Beating playoff team, is not a prerequisite to being a playoff team (in fact, could argue losing by 3 on the road and losing by 1 to playoff teams is an argument FOR them being a playoff caliber team)

[Greg Harmon @BarGraphCFB]

@CubWin @jakeryan2112 Wisconsin is the 12th highest rated team by Sagarin rating of the CFP era

[Greg Harmon @BarGraphCFB]

@goehring_seth The AP poll, the same poll, had Notre Dame over Alabama prior to the playoffs.

That is precisely the point

[Greg Harmon @BarGraphCFB]

@KirkHerbstreit More than half of Indiana and Miami’s rosters are transfers vs <30% for Ohio State (28%) and Georgia (10%).

What are we doing here?

[Greg Harmon @BarGraphCFB]

@BudElliott3 Exactly. If anything the Miami story boosts the meaning of the regular season. An August game against Notre Dame is what enabled them to potentially win the national championship

[Greg Harmon @BarGraphCFB]

A lot of panic over ND’s portal performance today. Marsh was an L, but I think there is more talent in that room than people give it credit for. And always thought it was an uphill battle to get the top boundary corner in the portal to come play nickel

If they don’t get at least 2 of their top 3 DT targets, then I’d start panicking

[Greg Harmon @BarGraphCFB]

To be fair, I believe the headline was $75mm of economic impact on the Orlando community. Which given a three night stay in Disney World for a family of 4 is like $5k… who knows?

If someone was throwing around $50mm to ESPN, that is such a ridiculous number it can’t be taken seriously (pretty sure the TNT sub contract deal on the first round games was only like $50mm a game… not sure what margin they underwrite to, but $50mm of potential add revenue loss on poptarts is order of magnitudes off)

[Greg Harmon @BarGraphCFB]

@slmandel These are good ratings, but would be interesting to do the analysis putting it in the context of the regular season.

For example, per SportsMediaWatch, 3.8 million people watched Iowa Nebraska in Week 14 in the same time slot 18.4 million watched Michigan OSU

[Greg Harmon @BarGraphCFB]

@secnumbersguy Percentage of the Top 5 Power Rated Teams:
- SEC 0%
- Big Ten 60%
- Big 12 20% or ACC 20% (depending on Sagarin or FPI)
- Notre Dame 20%

[Greg Harmon @BarGraphCFB]

@DCHard21 @jrs_rankings @GopherInsight OK. If the argument you are trying to have is, is 80.9 a higher number than 79.3, I agree.

If you are asking me to agree that definitively says the SEC is a better conference in any relevant application, I disagree.

Appreciate the conversation.

[Greg Harmon @BarGraphCFB]

The only reason to compare conferences is to determine the playoff field, which sole purpose is to determine the national champion.

Otherwise we’re just arguing about nothing. That was the “national relevance” comment. I frankly don’t care if Mississippi State or Cal is better because any nationally relevant team should wipe the floor with both of them

[Greg Harmon @BarGraphCFB]

The way I just presented it to you is a better way to determine which conference is better than using a simple average

The strength of the top portion of the conference should be weighed more heavily when having any conversation of national relevance. For example, the average national champion in the CFP era has a Sagarin rating of 101 - whether the 10th best team is rated 79 or 74 is somewhat irrelevant as the team with rating of 101 should be winning 95%+ of the time against both teams

Averages are often, arguably typically, not the best way to determine whether one thing is better than the other

[Greg Harmon @BarGraphCFB]

I will concede that - that is correct.

Averages are deceiving and I would argue that is the wrong way to look at it when trying to determine "best" conference particularly in the context of a CFP discussion (for example, if you look at the top 8 teams in each conference, the Big 10 is rated demonstrably higher on avg; even if you look at the top 16, the Big 10 is rated higher on avg).

This year the SEC's top 4 (87.5) was closer to the Big 12 (85.0) than the Big 10 (94.1), and they got 5 teams in.

Below is current Sagarin (ideally would use ratings prior to the post-season but don't have the time to pull it right now), admittedly haven't checked the others:

[Greg Harmon @BarGraphCFB]

@DCHard21 @jrs_rankings @GopherInsight This is just factually untrue. The SEC had none of the top 5 power rated teams this season.

[Greg Harmon @BarGraphCFB]

@ESPN_BillC And with that, Alabama's post-game expected win total finishes the year at 9.3... in 15 tries...

[Greg Harmon @BarGraphCFB]

Why arbitrarily cutoff at the top 25 teams? This is why computer models exist - humans are very bad at simultaneously ranking 136 teams (which is required to rank 12 given minimal schedule overlap)

By not ranking all 136 teams, we underweight bad losses.

Getting blown out by the 44th ranked team is much more indicative of not being playoff caliber than losing by a final minute field goal and a missed extra point to playoff teams and winning your other 10 games with 99%+ PGWE

[Greg Harmon @BarGraphCFB]

Well actually, no.

Colley (first column) - doesn’t account for score at all. Literally just applies the transitive property based on wins and losses only across the whole system.

Massey (second column) - does the same but accounts for margin of victory. However it is probabilistic w/
diminishing returns (so minimal credit for winning 70-7 vs, for example, 45-21)

FPI / Sagarin / SP+ are predictive models. So yeah, they do take into account beating the crap of your opponent (as they should). I’m not sure I’d advocate using them to rank teams, but included in my article (linked below if interested in learning more) because power ratings were the crutch SEC fans hid behind for years until this year (when the paradigm inverted - 0 of the top 5 power rated teams are in the SEC and they still got 5 playoff spots).

ALL of these are inherently opponent adjusted (well except for AP/Coaches polls which are just… people)

https://t.co/XMUiLlrTG4

[Greg Harmon @BarGraphCFB]

In Ari’s defense. I’m not sure he ever said Notre Dame wouldn’t perform well in the playoff. If I’m not mistaken he went out of his way to say they were probably a top 5 team

That said, I think it was a “miss” to spend so much of the airtime on ND vs Miami which was just the wrong conversation to be having - the SEC should have never been a 5 bid league this year with no real elite teams (none of the top 5 in Sagarin or other power rating systems). In hindsight, it was somewhat obvious