Bill Connelly's Georgia Tech Preview

[gtwcf]

Pretty clean write up. I love these reviews from an outside perspective. I particularly love seeing all these media pieces, tweets, and quotes that praise the staff that Collins put together (Fiutak says they are "terrific"). It really is great that we were able to pull together such a great group of like-minded coaches that all share Collins' vision. I've been especially pleasantly surprised by the amount of praise Marco Coleman has been receiving as a coach from the players, considering it's his first year at this level and second year coaching overall.

Missed the news on Brandon Adams.....

 

[CTJacket]

Our history with Head Coach changes is inconclusive, at best:
Curry -> Ross --- 5-5-1 -> 2-9 (I wasn't alive, don't know what happened)
Ross -> Lewis --- 8-5 -> 5-6 (Bill Lewis was a bad coach, and they lost a lot of talent)
Lewis -> O'Leary --- 1-10 -> 6-5 (Bill Lewis was a very bad coach)
O'Leary -> Chan --- 8-5 -> 7-6 (Chan Gailey Equilibrium)
Chan -> CPJ --- 7-6 -> 9-4 (Dwyer + Nesbitt + maybe our best defensive talent since Ross + a better Coach)
CPJ -> Collins --- 7-6 -> ??? (Losing a ton of starters...beyond that, who knows)

There was a tough culture change from Curry to Ross (who loved his players but expected them to work hard). Also I believe there wasn't a lot of talent, not totally due to Coach Curry (great Tech man, I don't want to disparage him).
Lewis inherited a TON of talent. Donnie Davis was the #2 qb in the nation coming out of high school. There were two (TWO!) NFL running backs on the roster PLUS Jimy Lincoln who was no slouch. The line was still good. Ton of talent on the D side (Coleman Rudolph for one).

Anyway, I have real faith in CGC on the defensive side (and recruiting but that doesn't matter much for this year unless one of the transfers makes an impact). We'll most probably slide some on offense but I think we can more than make up for it on the defensive side. Plus, if we're improving as the year goes on that should make us all feel good.

 

[GT14]

I do agree that this is probably as good a system as can be made. I would even say that on average it is probably decent. I only push back because with the way some people post, it appears that they believe that since it is mathematical and done on a computer that the results are accurate.

Completely fair take. I personally really enjoy these analyses, and appreciate how candid Connelly is about the predictions. In the article it seemed like he was constantly mentioning how it's difficult to predict anything for us this year. I think last year he even poked fun at himself for always being wrong about GT (though to be fair, no one predicted 14 to be as good as it was, and no one predicted 15 to be as bad as it was).

 

[YellowJacketFan2018]

Our history with Head Coach changes is inconclusive, at best:
Curry -> Ross --- 5-5-1 -> 2-9 (I wasn't alive, don't know what happened)
Ross -> Lewis --- 8-5 -> 5-6 (Bill Lewis was a bad coach, and they lost a lot of talent)
Lewis -> O'Leary --- 1-10 -> 6-5 (Bill Lewis was a very bad coach)
O'Leary -> Chan --- 8-5 -> 7-6 (Chan Gailey Equilibrium)
Chan -> CPJ --- 7-6 -> 9-4 (Dwyer + Nesbitt + maybe our best defensive talent since Ross + a better Coach)
CPJ -> Collins --- 7-6 -> ??? (Losing a ton of starters...beyond that, who knows)

Bill Curry was offered a lot of money by Alabama and he thought he could accomplish things at Alabama that he wouldn't be able to at Georgia Tech. I think Bobby Dodd encouraged him to take the Alabama head coaching job.

 

[MostDefinitely]

I don't get all the Bill C hate in this thread. Yeah his prediction system isn't perfect but that's half the fun of it. He's trying to predict team performance based on historical data, which is obviously difficult to do -- but I think it's fun to read about how he does it and see how close he can get.

He certainly doesn't have anything personal against GT, he actually likes us quite a bit. His model just doesn't think we're going to do very good this year. I disagree with the model, but I still enjoyed the article.

 

[85Escape]

Thanks a lot for posting this. I am amazed at the number of presumably engineers that don't understand expectation values.

Oh, you misunderstand. I understand the model just fine. I just happen to have concluded that it is a valid, but non-informative, model.

Research is the process of studying phenomena with the express purpose of predicting and/or controlling outcomes. Building a model is the process of taking that research and applying it (typically) toward the former.

Based on history, I am not convinced that this model is significantly better at predicting the specific outcome of "GT Wins-Losses" than the much simpler model of predicting that we will be 7-5 every year. Given that "right" for these types of models can reasonably include +/- 1 win, then that means the model would have to have a good shot at predicting when we are outside of that range of 6 - 8 wins. The only few times we've been outside of that window the model has, in fact, failed to predict it, right?

That makes this model a long run for a very short slide.

 

[GT14]

Oh, you misunderstand. I understand the model just fine. I just happen to have concluded that it is a valid, but non-informative, model.

Research is the process of studying phenomena with the express purpose of predicting and/or controlling outcomes. Building a model is the process of taking that research and applying it (typically) toward the former.

Based on history, I am not convinced that this model is significantly better at predicting the specific outcome of "GT Wins-Losses" than the much simpler model of predicting that we will be 7-5 every year. Given that "right" for these types of models can reasonably include +/- 1 win, then that means the model would have to have a good shot at predicting when we are outside of that range of 6 - 8 wins. The only few times we've been outside of that window the model has, in fact, failed to predict it, right?

That makes this model a long run for a very short slide.

That comment had nothing to do with how valid the model is. Also, I don't think my comment was aimed at you unless you were the one calling Connelly dumb for predicting 6 wins last year and 4 wins this year when we were favored in 8 and 1 game respectively. My statement was disbelief that someone would say a 49% chance of winning all 12 games would result in a 0 win prediction rather than the expected value of 5.9 (12x0.49=5.9).

FWIW I quite like the model and enjoy predictions. I can predict the SP500 returns 8% every year and be right on average, but it's way more fun to predict based on shorter term data (I am not condoning speculative investing, just giving an example).

 

[orientalnc]

12 x .49 will cluster around 6, but 0 and 12 are also possible.

 

[YellowJacketFan2018]

12 x .49 will cluster around 6, but 0 and 12 are also possible.

The Swarm Message Board would destroy itself like Chernobyl if the unthinkable happened and Georgia Tech went 0-12

 

[orientalnc]

The Swarm Message Board would destroy itself like Chernobyl if the unthinkable happened and Georgia Tech went 0-12

You better stock up now on Woodford's Reserve, just in case we go 12-0.

 

[85Escape]

That comment had nothing to do with how valid the model is. Also, I don't think my comment was aimed at you unless you were the one calling Connelly dumb for predicting 6 wins last year and 4 wins this year when we were favored in 8 and 1 game respectively. My statement was disbelief that someone would say a 49% chance of winning all 12 games would result in a 0 win prediction rather than the expected value of 5.9 (12x0.49=5.9).

FWIW I quite like the model and enjoy predictions. I can predict the SP500 returns 8% every year and be right on average, but it's way more fun to predict based on shorter-term data (I am not condoning speculative investing, just giving an example).

Yeah, I know you weren't aimed at me and apologize if it came off as offended (or offensive)! I just love the topic.

I find that the ability of a model to predict is often over-stated if the base assumption is that we have zero 'best simple guess.' If you compare many models to their ability to predict above the 'best guess', then they are often not all that great. I like to look at models in a 'wins above replacement' mentality. My simple point is that I'm not that impressed with his model when it is used to predict GT's Wins vs Losses. The fact that it is good in aggregate isn't that useful to me personally...I know that the average temperature of everything is -454.76F, but I'm really way more interested in exactly how warm my beer has become while playing cornhole.

 

[CTJacket]

I don't get all the Bill C hate in this thread. Yeah his prediction system isn't perfect but that's half the fun of it. He's trying to predict team performance based on historical data, which is obviously difficult to do -- but I think it's fun to read about how he does it and see how close he can get.

He certainly doesn't have anything personal against GT, he actually likes us quite a bit. His model just doesn't think we're going to do very good this year. I disagree with the model, but I still enjoyed the article.

I agree with this. I think people are upset (and I can argue rightfully so) that he even put out an attempt with us. He DID cage it though - transition doesn't work well. Plus, let's all be honest, he would be better off NOT posting/predicting records etc. until at least three weeks in. The rest of his dialogue I thought was spot on. He identified what we need to do, he pointed out the challenges, and he was actually knowledgeable about our team. Better than most FOR SURE.

 

[CTJacket]

You better stock up now on Woodford's Reserve, just in case we go 12-0.

It works either way, W or L.

 

[YellowJacketFan2018]

You better stock up now on Woodford's Reserve, just in case we go 12-0.

Xanax might be good also

 

[GTpdm]

It works either way, W or L.

I’ll be drinking Woodford on Weekends this Fall no matter where we end up between 0-12 and 15-0.

 

[Coloradojacket]

Bill Curry was offered a lot of money by Alabama and he thought he could accomplish things at Alabama that he wouldn't be able to at Georgia Tech. I think Bobby Dodd encouraged him to take the Alabama head coaching job.

Always forgotten or not know. In Fall camp, Jerry Mays tore his ACL. This had a huge impact on the offense in 87 and he was not fully back in 88. In 89 he was and we went 7-4

 

[TheSilasSonRising]

Bill Curry was offered a lot of money by Alabama and he thought he could accomplish things at Alabama that he wouldn't be able to at Georgia Tech. I think Bobby Dodd encouraged him to take the Alabama head coaching job.

Bobby Dodd, a great coach and horrific AD, also advised GT to leave the sec which almost killed us. So I could not care any less what he advised that pompous, blow hard curry to do.

Pepper gave curry the chance to get into coaching and within 2 years he was putting a knife in Peppers back.

When the, at worst, 2nd greatest pure football coach we ever had - Boss Ross - was taking us to a N.C., curry was sitting up at uvag with vinnie and babs dooley taking credit for the team.

Pompous scumbag hypocrite.

 

[gtphd]

In the article it seemed like he was constantly mentioning how it's difficult to predict anything for us this year.

“Psychics” use two tricks when making predictions: (1) be very vague such that it will almost certainly happen: “You’ll meet an interesting man.” And (2) if they get too specific, heavily hedge the prediction so they can C their A when if it doesn’t happen: “It’s a bit hazy, but I think the man will be important to your future. He might be your future husband, but I can’t be entirely certain. The universe is sending me mixed signals.” Often psychics use new, poorly understood “technology” to make their predictions: crystal balls, palm reading, tarot cards, proprietary computer models, etc.

It’s a scam as old as time.

 

[Bogey]

Maybe everyone should run a high school offense. Maybe they already are.

I know for a fact that all high schools run a high school offense.

Sent from my ASUS_Z01RD using Tapatalk

 

[YellowJacketFan2018]

“Psychics” use two tricks when making predictions: (1) be very vague such that it will almost certainly happen: “You’ll meet an interesting man.” And (2) if they get too specific, heavily hedge the prediction so they can C their A when if it doesn’t happen: “It’s a bit hazy, but I think the man will be important to your future. He might be your future husband, but I can’t be entirely certain. The universe is sending me mixed signals.” Often psychics use new, poorly understood “technology” to make their predictions: crystal balls, palm reading, tarot cards, proprietary computer models, etc.

It’s a scam as old as time.

Well,what else is there to do in the offseason?