CPJ in One-Score Games

[ilovetheoption]

So, one of them is 55% in coin flip games and the other one is 46% in coin flip games?

That sounds like coin flip results to me.

 

[Fatmike91]

again, I don't think this is about Chan vs. Paul (unless I am obtuse and completely unaware of it). This is looking at the data to see what can be done to improve going forward. isn't that what we do at work (or at least many of us)?

Let’s keep in mind that CPJ’s entire strategy is to play limited possession games which should always be “closer”. Even against teams that we shouldn’t beat (like FSU in their Natty year).

If CPJ has lost more “close ones” than Chan then the data should tell you Chan got “blown out” more since the overall records are similar.

What analysis do we choose to do?

/

 

[CuseJacket]

Personally I love threads where people present data, offer their own conclusions, and give others a chance to respond. The OP offered something new, rather than the same data-less rants after a disappointing loss.

I did not interpret the OP as suggesting Chan > CPJ on the whole. Rather, he offered comparison data. I'm sure it took quite a bit of time/effort to put it together. My guess is that time permitting we'd see more teams/coaches for comparison and not just GT (not a request @bke1984, just my 2 cents on why the data ends with CPJ + Chan).

As always data needs to be understood with appropriate context, which I think @bke1984 explained well and then clarified in follow-up posts. Others have made good points and caveats to the contrary too. Good stuff.

 

[WreckinGT]

Too many variables of how and why a game ended within one score to have any meaning.

This is my biggest issue with the data. It equates two games like the 2015 Notre Dame game, in which we were never really even in the game, to this years Tennessee game in which we blew a decent lead. Games can be wildly different and fall into the same category.

 

[OldJacketFan]

This is my biggest issue with the data. It equates two games like the 2015 Notre Dame game, in which we were never really even in the game, to this years Tennessee game in which we blew a decent lead. Games can be wildly different and fall into the same category.

I was thinking along the same lines. There are wildly differing one scores games and simply looking at the win/loss record doesn't reflect the difference between a game that Tech dominates for 95% of the games and yet wins/loses by one score, a game in which Tech is dominated for 95% of the game but manages to win or lose by one score or a game when you have equally matched teams and, win or loss, there is a one score margin.

 

[bke1984]

This is my biggest issue with the data. It equates two games like the 2015 Notre Dame game, in which we were never really even in the game, to this years Tennessee game in which we blew a decent lead. Games can be wildly different and fall into the same category.

Very true. It also counts games like the 2013 uGA game and the 2012 Miami game where had three score leads and lost...which to me is even worse than losing a game like yesterday.

Look, the point of my post was to say that I felt like we were losing more than our fair share of close games and I wanted to see if that was correct. I picked a single score game as a metric for this, since it's really the most reasonable way to determine if a game is close or not based on a simple statistic. This data also excludes games where we were winning by 2-3 until late when a big play happened to make it a two score game...so sure, maybe we're winning more "close games." Honestly though, I just don't feel like we are. After those first two years it's really seemed like most of the time we either win big or we lose...there are definitely exceptions, but I think the data supports what I was feeling at least to some extent.

If you want to get into why we're losing, that's a different story. Yesterday it was a combination of things...but mostly it was players not making plays in key situations. Brad Stewart's drop, the missed block on the Taquon run, Taquon under throwing Ricky, and the coup de gras tipped ball catch at the end (which was eerily similar to the Pitt game last year). Bottom line is that for the last 8 years we've lost 65% of the games that were ultimately decided by one score or less...statistically significant or not, that is a fact, not an opinion.

 

[bke1984]

So, one of them is 55% in coin flip games and the other one is 46% in coin flip games?

That sounds like coin flip results to me.

Yep, just most of our wins under PJ in those situations came in the first to years. I wish we'd have another year or two like that instead of the ones like 2012, 2013, 2015, and this year.

A lot of football left to be played, and based on numbers we'll probably play three more one score games this year...maybe we can get to 3-2....or maybe we can just blow everyone out the rest of the way and somehow still win the division when Miami loses to Syracuse and Va Tech...

 

[bke1984]

What we need is the CPJ death march again.

This was part of the problem yesterday.

 

[alagold]

What we need is the CPJ death march again.

yep
we saw a death march SAT,only it was uM doing the march.

 

[AE 87]

Personally I love threads where people present data, offer their own conclusions, and give others a chance to respond. The OP offered something new, rather than the same data-less rants after a disappointing loss.

I did not interpret the OP as suggesting Chan > CPJ on the whole. Rather, he offered comparison data. I'm sure it took quite a bit of time/effort to put it together. My guess is that time permitting we'd see more teams/coaches for comparison and not just GT (not a request @bke1984, just my 2 cents on why the data ends with CPJ + Chan).

As always data needs to be understood with appropriate context, which I think @bke1984 explained well and then clarified in follow-up posts. Others have made good points and caveats to the contrary too. Good stuff.

I agree that posts which present data and interpretation make interesting threads.

However, some such posts don't have a clear logic.

For example, there has been a lot of discussion about W-L after one team or the other has a bye-week. For it to be a directly meaningful stat, all teams must be basically equal so that the only variable is the bye-week. To be indirectly meaningful, they teams coming off a bye must average-out as basically equal. However, we never see evidence of either of these claims to be true.

The same sort of problem pertains to this post/thread. It assumes that all one-score games are against basically equal teams so that the only meaningful variable is the head coach. It ignores issues of bye weeks, injuries, quality of opposition, and a particular GT team's team-chemistry.

In my opinion, this sort of thread is an insult to those of us who like to use statistics.

Perhaps, a comparison of results to the Vegas spread might make it more interesting. Certainly losing by 1 score to a team favored by 2 scores is better than losing by 1 score to a team you were favored to beat by 2 scores. Treating them as if they are the same is just weird in my opinion.

 

[takethepoints]

Sometimes seasoned judgment trumps data.

Not very often.

 

[LibertyTurns]

Not very often.

I work with quite few people that can’t make a decision unless they’re provided with reams of supporting data. They perform at an adequate level but the majority of them hit their glass ceilings very early in their careers due to an inability to transition from simple yes/no scenarios with highly accurate data to more complex multivariate ones with limited data requiring judgment. If only life was so easy.

There’s really no way of telling whether or not changing out any coach will result in the same, better or worse outcome. Look at USCe- they’ve had Spurrier & Holtz and still sucked. What’s Notre Dame’s excuse? Nebraska? Florida?

Ultimately the man in charge makes the call. For me, I just wish the man in charge would support athletics to the extent we support academics. I see no reason why we can’t excel at both but it appears we have another administration that feels pursuing athletic excellence somehow taints our academic reputation.

 

[takethepoints]

I work with quite few people that can’t make a decision unless they’re provided with reams of supporting data. They perform at an adequate level but the majority of them hit their glass ceilings very early in their careers due to an inability to transition from simple yes/no scenarios with highly accurate data to more complex multivariate ones with limited data requiring judgment. If only life was so easy.

Personally, every time someone tells me that they have used "judgment" to reach a conclusion that data doesn't support, I put my hand over my wallet and back slowly away. Most people will take a leap of faith every now and then (I sure have) and sometimes it works. No doubt that happens. Also, no doubt that it doesn't happen very often and when it does it is pure chance. I know that those limited successes can have a positive influence on a person's career, as you say. I think this is a species of the "Halo Effect"; i.e. the tendency we all have to impute good characteristics of all sorts to people who are successful in one area. (He made lots of money! He must be a genius!) We should avoid that, if we can, imho.

 

[LibertyTurns]

Personally, every time someone tells me that they have used "judgment" to reach a conclusion that data doesn't support, I put my hand over my wallet and back slowly away. Most people will take a leap of faith every now and then (I sure have) and sometimes it works. No doubt that happens. Also, no doubt that it doesn't happen very often and when it does it is pure chance. I know that those limited successes can have a positive influence on a person's career, as you say. I think this is a species of the "Halo Effect"; i.e. the tendency we all have to impute good characteristics of all sorts to people who are successful in one area. (He made lots of money! He must be a genius!) We should avoid that, if we can, imho.

A lot of truth here. I guess I’ve been in too many intel & business decision briefs where the data said “a”, your gut said “b” and the actual answer was none of the above. By the grace of God I’ve been lucky enough to have avoided the dreaded career ending decision.

 

[Animal02]

I work with quite few people that can’t make a decision unless they’re provided with reams of supporting data. They perform at an adequate level but the majority of them hit their glass ceilings very early in their careers due to an inability to transition from simple yes/no scenarios with highly accurate data to more complex multivariate ones with limited data requiring judgment. If only life was so easy.

My wife is a behavioral psychologist. Her stock quote is "Without data, you are just another ******* with an opinion"
I am more of a gut feeling type person.

 

[Jean-Baptiste Rochambeau]

Wake Forest beat Ga Tech 9-6 in the 2006 ACC Championahip game.

Don’t start wishing for Chan back in ONE SCORE GAMES. We had Calvin Johnson on the field. It doesn’t get any worse.

Sorry - not buying into your story.

/

I think his point was, for all his prowess calling games without a play sheet, Paul Johnson is not a very good sideline coach or game manager in one score games that usually come up when we are playing teams at or above our talent level.

 

[Jean-Baptiste Rochambeau]

Johnson was asleep at the wheel with the offense and you wanted him to take control of the defense? What a joke. If we want to win we need our offense to get it's head out of its behind and start putting teams away.

Paul Johnson is our offense. Always will be until he leaves. Our players do what they are directed to do.

 

[CuseJacket]

I agree that posts which present data and interpretation make interesting threads.

However, some such posts don't have a clear logic.

For example, there has been a lot of discussion about W-L after one team or the other has a bye-week. For it to be a directly meaningful stat, all teams must be basically equal so that the only variable is the bye-week. To be indirectly meaningful, they teams coming off a bye must average-out as basically equal. However, we never see evidence of either of these claims to be true.

The same sort of problem pertains to this post/thread. It assumes that all one-score games are against basically equal teams so that the only meaningful variable is the head coach. It ignores issues of bye weeks, injuries, quality of opposition, and a particular GT team's team-chemistry.

In my opinion, this sort of thread is an insult to those of us who like to use statistics.

Perhaps, a comparison of results to the Vegas spread might make it more interesting. Certainly losing by 1 score to a team favored by 2 scores is better than losing by 1 score to a team you were favored to beat by 2 scores. Treating them as if they are the same is just weird in my opinion.

Understand your point to an extent. It sounds like you're arguing for less grey/ambiguity and greater controls, which I'm all for if someone has the time/interest.

Fwiw, I look at this data as if it has similar clarity to calculating a ypp/ppd vs. P5. Not all P5 opponents are anywhere close to the same and the stat does not account for bye weeks, injuries or an opposing team's chemistry. But I think both say something.

 

[JorgeJonas]

Perhaps the difference is that the conference is tougher, and but for Johnson, the one possession losses would have been losses by double digits under Gailey.

Generally, though, these things are random, but that doesn’t mean they even out over time.

 

[AE 87]

Understand your point to an extent. It sounds like you're arguing for less grey/ambiguity, which I'm all for.

Fwiw, I look at this data as if it has similar clarity to calculating a ypp/ppd vs. P5. Not all P5 opponents are anywhere close to the same and the stat does not account for bye weeks, injuries or an opposing team's chemistry. But I think both say something.

Actually, I don't think that they're even close to being the same.

If you go back and look at my threads on ppd vs pwr5, I've both acknowledged the limitations you mention and given reasons for showing that over a whole season they do seem to largely average out. I also try and overcome these limitations to some extent by only including teams that have played at least 3 pwr5 opponents. You may notice that I have not posted any PPDvsPWR5 rankings yet for this year. It's because even though I recognize that strength of schedule can vary wildly even with 3-4 PWR5 opponents, the stat begins to become more reasonable at this point.

The fact that you don't appreciate these differences, well, I don't know what that means.