Jonathan Bales on draft grades

[FuzzyLumpkins]

One possibility I keep coming back to is that Bales' overall theory is essentially correct, but his criteria for player selection and odds calculation are flawed. Maybe he's right about what is essentially risk aversion, but isn't very good at determining what represents actual risk.

Of course if there was some objective probability ranking to go by then you could judge on the basis of reality. That is clearly not the case though. It's all opinion.

 

[visionary]

All I say is that the vast majority of analysts loved....

Todd Marinavich
Lawrence Phillips
Heath Shuler
Jamarcus Russell
Ki-Jana Carter
Akili Smith
Ryan Leaf
Tony Mandarich

Of course they did
Everything has an error built into it, overtime you find out who was better at it than others

There are clearly GMs who have a sound process and others who don't
That translates directly to team success, it may not happen in 1 year but it does over time (and that does not mean 10 years)

By the same token , there are GMs whose process has been shown to be a failure and no amount of lipstick will change that pig into anything but a pig . Using that process you will sometimes luck into a player but this will not happen consistently enough for the team to have real long term success, it may be a flash in the pan for a year when 'you catch lightning in a bottle' but that is all it will be.
There are real reasons why our fans cling to slogans like hope and believe..... In its current structure this organization is lumped with the Commanders, raiders, and browns not patriots, niners, Seahawks and I don't see much reason to change that.
Bales' points are very good ones. They just happen to hit close to home so some feathers are ruffled. You think ravens and niners fans are up in arms about what he said?
Fuzzy has 2 schticks, you tube pretend scout and blame the messenger so I don't take him too seriously

 

[DFWJC]

Of course they did
Everything has an error built into it, overtime you find out who was better at it than others

There are clearly GMs who have a sound process and others who don't
That translates directly to team success, it may not happen in 1 year but it does over time (and that does not mean 10 years)

By the same token , there are GMs whose process has been shown to be a failure and no amount of lipstick will change that pig into anything but a pig . Using that process you will sometimes luck into a player but this will not happen consistently enough for the team to have real long term success, it may be a flash in the pan for a year when 'you catch lightning in a bottle' but that is all it will be.
There are real reasons why our fans cling to slogans like hope and believe..... In its current structure this organization is lumped with the Commanders, raiders, and browns not patriots, niners, Seahawks and I don't see much reason to change that.
Bales' points are very good ones. They just happen to hit close to home so some feathers are ruffled. You think ravens and niners fans are up in arms about what he said?
Fuzzy has 2 schticks, you tube pretend scout and blame the messenger so I don't take him too seriously

Didn't ruffle my feathers, that's for sure.

 

[ABQCOWBOY]

We can go round and round about any of these kinds of discussions till the Cows come home but the ultimate grade is if you win a Championship.

 

[The Cleaner]

I'm not a football professional, so I can't speak to the scouting side of things, but I can speak about statistics with authority. What Bales is saying is misleading. First of all, it can't be proven that individual drafters return to the mean with the quality of their picks, it can only be shown that there is not sufficient evidence that they pick better than chance. Those are two very different things. Even if this could be proven, the picks you make one year are not independent of the picks you make a second year, and so any causal explanation is questionable. It may be harder to pick successfully in the draft if you have been successful in the past for some reason - that is an alternative explanation for returning to the mean.
Secondly, the kinds of statistical analyses that can be done in football are limited by the small sample size of much of the analysis. Statistical findings may be non-significant due to something actual not being significant, or they may be non-significant due to a lack of statistical power. Statistics are generally used in football in situations where there is a large enough sample size - so you might group football players historically according to position and examine objective measures that predict success at that position. You can't use statistics to analyze more specific situations - how well a drafted player will do as a 1 technique under Marinelli - because the sample size will be too small. Also, there are a lot of subjective and arbitrary decisions that need to be made in a statistical analysis - like defining the success of a particular player, or deciding how far back to analyze players.
Bales seems to say that using predictive stats is better than scouting for the draft, but there is absolutely no evidence for this. He can say that his predictive stats better predict player success according to some measure of success historically, but that is a lot different than having to go pick the players yourself. A lot of the specific factors that get averaged out in a large analysis come in to play when you are the GM of a specific team. Or to put in simple terms - the best player for the average NFL team may not be the best player for the Dallas Cowboys. It probably wouldn't even be possible to really test Bale's hypothesis, because if a lot of teams starting picking in a Bales approved way, it would change the nature of the draft and the historical data wouldn't be predictive anymore, and if only a few teams did this, then it would take too long too build up statistical power to test his hypothesis and the game would have changed by then anyhow.

 

[Doomsday101]

I disagree with Bales here.

Just because some analysts thinks a team should or should not have picked a player, does not remotely mean they are right.

Real results in real NFL games is proof one way or the other...or at least a close to proof that we'll ever get.

The rest is theory...period.

I agree there have been many players that scouts have over valued and turned out to be major bust and yet guys picked late being under valued who have turned out to be HOF players.

 

[visionary]

I'm not a football professional, so I can't speak to the scouting side of things, but I can speak about statistics with authority. What Bales is saying is misleading. First of all, it can't be proven that individual drafters return to the mean with the quality of their picks, it can only be shown that there is not sufficient evidence that they pick better than chance. Those are two very different things. Even if this could be proven, the picks you make one year are not independent of the picks you make a second year, and so any causal explanation is questionable. It may be harder to pick successfully in the draft if you have been successful in the past for some reason - that is an alternative explanation for returning to the mean.
Secondly, the kinds of statistical analyses that can be done in football are limited by the small sample size of much of the analysis. Statistical findings may be non-significant due to something actual not being significant, or they may be non-significant due to a lack of statistical power. Statistics are generally used in football in situations where there is a large enough sample size - so you might group football players historically according to position and examine objective measures that predict success at that position. You can't use statistics to analyze more specific situations - how well a drafted player will do as a 1 technique under Marinelli - because the sample size will be too small. Also, there are a lot of subjective and arbitrary decisions that need to be made in a statistical analysis - like defining the success of a particular player, or deciding how far back to analyze players.
Bales seems to say that using predictive stats is better than scouting for the draft, but there is absolutely no evidence for this. He can say that his predictive stats better predict player success according to some measure of success historically, but that is a lot different than having to go pick the players yourself. A lot of the specific factors that get averaged out in a large analysis come in to play when you are the GM of a specific team. Or to put in simple terms - the best player for the average NFL team may not be the best player for the Dallas Cowboys. It probably wouldn't even be possible to really test Bale's hypothesis, because if a lot of teams starting picking in a Bales approved way, it would change the nature of the draft and the historical data wouldn't be predictive anymore, and if only a few teams did this, then it would take too long too build up statistical power to test his hypothesis and the game would have changed by then anyhow.

What you say is very accurate and part of limitation even beyond this us the statistics , by their very nature, are limited to "0" or "1" and cannot be used to evaluate other aspects if the game like attitude, leadership, heart, how one players impacts another's game and on and on. However, there are clearly GMs and organizations that do have a system that works over long term (like steelers, patriots, etc) or at least over several years (like niners, GB, Seahawks) which allows them to remain truly competitive and in the hunt year after year . Contrast thus with organizations that want to catch lightning in a bottle and whose plan is 'anything can happen' or this might be our year so it's all good.

Clearly there is way to do it and a way not to

There clearly is also a role for short term process evualuation because that process has been in place for several years, so you have long term rests from which you can learn about your process

All successful major organizations work this way, it is called process improvement

Some don't want to hear this and would rather bury their heads in sand

 

[visionary]

Another thing is that good football organizations probably use a combination of "scouting" and statistics/metrics

 

[SilverStarCowboy]

This Bale guy is a friggin' geinus main.

 

[casmith07]

We can go round and round about any of these kinds of discussions till the Cows come home but the ultimate grade is if you win a Championship.

No it isn't. You can draft 7 guys who don't start and win a championship. Is that good drafting? Or were you already good? Or did you get a little luck?

Oversimplification is not good.

 

[visionary]

No it isn't. You can draft 7 guys who don't start and win a championship. Is that good drafting? Or were you already good? Or did you get a little luck?

Oversimplification is not good.

Neither is believing that nothing works at it is all random

Some teams have found a method that increases their chances of remaining truly competitive year after year and it does not include throwing darts

 

[waving monkey]

Another thing is that good football organizations probably use a combination of "scouting" and statistics/metrics

red herring

 

[FuzzyLumpkins]

red herring

Is he trying to say that the Cowboys ignore statistics?

I am not sure but I know that they have people at the Sloan Conferences every year.

 

[waving monkey]

Is he trying to say that the Cowboys ignore statistics?

I am not sure but I know that they have people at the Sloan Conferences every year.

I think his saying they use blood hounds for scouts and can't turn on a computer

 

[ABQCOWBOY]

No it isn't. You can draft 7 guys who don't start and win a championship. Is that good drafting? Or were you already good? Or did you get a little luck?

Oversimplification is not good.

Not in my book. It takes 53 guys to win a championship. At some point, you have to have drafted well to have a roster that is good enough to win. All things are measured, in the NFL, by how many championships you win. If you draft 7 crappy guys but you win a championship that year, then it stands to reason that you probably didn't draft 7 crappy guys. You probably drafted very well in years past and the 7 you drafted this year just aren't good enough to make a Championship caliber roster. No shame in that IMO. Just kinda how it works.

At the end of the day, the object is to build your team so well that guys are not making the team real often, through the draft. Then, you know you've done your job as a personnel man. You've built a strong team through the draft and it's winning championships that is the ultimate and undisputed proof of that.

 

[visionary]

Is he trying to say that the Cowboys ignore statistics?

I am not sure but I know that they have people at the Sloan Conferences every year.

"He" never said that
It was part of "his" larger response to the post by the cleaner

 

[xwalker]

I'm not a football professional, so I can't speak to the scouting side of things, but I can speak about statistics with authority. What Bales is saying is misleading. First of all, it can't be proven that individual drafters return to the mean with the quality of their picks, it can only be shown that there is not sufficient evidence that they pick better than chance. Those are two very different things. Even if this could be proven, the picks you make one year are not independent of the picks you make a second year, and so any causal explanation is questionable. It may be harder to pick successfully in the draft if you have been successful in the past for some reason - that is an alternative explanation for returning to the mean.
Secondly, the kinds of statistical analyses that can be done in football are limited by the small sample size of much of the analysis. Statistical findings may be non-significant due to something actual not being significant, or they may be non-significant due to a lack of statistical power. Statistics are generally used in football in situations where there is a large enough sample size - so you might group football players historically according to position and examine objective measures that predict success at that position. You can't use statistics to analyze more specific situations - how well a drafted player will do as a 1 technique under Marinelli - because the sample size will be too small. Also, there are a lot of subjective and arbitrary decisions that need to be made in a statistical analysis - like defining the success of a particular player, or deciding how far back to analyze players.
Bales seems to say that using predictive stats is better than scouting for the draft, but there is absolutely no evidence for this. He can say that his predictive stats better predict player success according to some measure of success historically, but that is a lot different than having to go pick the players yourself. A lot of the specific factors that get averaged out in a large analysis come in to play when you are the GM of a specific team. Or to put in simple terms - the best player for the average NFL team may not be the best player for the Dallas Cowboys. It probably wouldn't even be possible to really test Bale's hypothesis, because if a lot of teams starting picking in a Bales approved way, it would change the nature of the draft and the historical data wouldn't be predictive anymore, and if only a few teams did this, then it would take too long too build up statistical power to test his hypothesis and the game would have changed by then anyhow.

Bales is a wanna-be statistical analyst.

There is a group of people out there with spreadsheets and no formal education or training in math/statistics that have anointed themselves experts in statistical analysis.

They are convinced that they can prove all kinds of things in sports with statistics. Their favorite term is correlation.

 

[Sarge]

You can grade a draft anytime you like. Having said that, your "grade"can vary a tremendous amount from year to year. Frankly, I could care less about grading a draft.