Advanced Stats VS. Eye Ball Test | Page 5 | FinHeaven - Miami Dolphins Forums

Advanced Stats VS. Eye Ball Test

I'd agree as far as that, but I'll give you an example where metrics are useful, yet ignored: going for it on 4th down. Metrics strongly suggest coaches should go for it way more often than they do at present. In this case interpretation and context haver very little to do with it. In fact, I bet most people's gut would agree with this. And yet... there we are. NFL head coaches are extremely conservative in that regard.

Interesting. What is this based on?? If a team gets 4th and 1s 69% of the time, would that be considered a metric that would indicate they should go for it more often?? If so they are way off. Just because it is above 50% does not mean they should take the risk. See the risk is not 50/50. They are gaining a yard 69% of the time but losing 50 yards of field position 31% of the time. In this example you would have to be picking up 1st downs 98% of the time to even the risk. That was just an example, with many other vairables but I was curious if that was what the metrics was saying.

That is the biggest problem with stats. It is not the stats fault that people use the stats for things the stats don't tell. The stats only tell one part of the story, and are only useful if the person using them knows how to use them correctly.
 
Here are some more metrics evaluations from another site, this time on explosiveness ratings for defensive ends and rush linebackers, and also some input on defensive tackles. The guy who does them is a Patriots fan. He puts out his list every year before the draft. Khalil Mack, Kareem Martin, Ben Gardner and Ra'shede Hageman receive the highest grades this year. Last season Jamie Collins had the highest rating.

http://www.footballsfuture.com/phpBB2/viewtopic.php?t=547890&postdays=0&postorder=asc&start=0

Here is the 2013 version. Dion Jordan did not fare very well. I'll paste his summary below the link:

http://www.footballsfuture.com/phpBB2/viewtopic.php?t=521433

Dion Jordan, OLB, MEI 70.23 – The reach. Jordan isn't a bad player, he just shouldn't be in consideration for a top five pick. He's tall but skinny, athletic but inert. His mediocre vertical of 32.5 inches likely portends a critical lack of explosiveness that will limit him as an edge rusher, given his wiry frame. Some guys can get away with less burst to their game by using length, power and leverage, but Jordan only possesses one of those traits. He'll probably have to stand up and work as a linebacker exclusively, as it's hard to see him enduring the wear and tear of working as a down lineman in the trenches regularly. He is still relatively young and has only played on the defensive side of the ball for three seasons, so there's a chance he grows into his body and puts it all together. But given his lack of production and rapid ascent in draft circles, I wouldn't want to be the one to take that chance with a premium pick.

2012:

http://www.footballsfuture.com/phpBB2/viewtopic.php?t=488386

2011:

http://www.footballsfuture.com/phpBB2/viewtopic.php?t=442175
 
Here are some more metrics evaluations from another site, this time on explosiveness ratings for defensive ends and rush linebackers, and also some input on defensive tackles. The guy who does them is a Patriots fan. He puts out his list every year before the draft. Khalil Mack, Kareem Martin, Ben Gardner and Ra'shede Hageman receive the highest grades this year. Last season Jamie Collins had the highest rating.

http://www.footballsfuture.com/phpBB2/viewtopic.php?t=547890&postdays=0&postorder=asc&start=0

Here is the 2013 version. Dion Jordan did not fare very well. I'll paste his summary below the link:

http://www.footballsfuture.com/phpBB2/viewtopic.php?t=521433

Dion Jordan, OLB, MEI 70.23 – The reach. Jordan isn't a bad player, he just shouldn't be in consideration for a top five pick. He's tall but skinny, athletic but inert. His mediocre vertical of 32.5 inches likely portends a critical lack of explosiveness that will limit him as an edge rusher, given his wiry frame. Some guys can get away with less burst to their game by using length, power and leverage, but Jordan only possesses one of those traits. He'll probably have to stand up and work as a linebacker exclusively, as it's hard to see him enduring the wear and tear of working as a down lineman in the trenches regularly. He is still relatively young and has only played on the defensive side of the ball for three seasons, so there's a chance he grows into his body and puts it all together. But given his lack of production and rapid ascent in draft circles, I wouldn't want to be the one to take that chance with a premium pick.

2012:

http://www.footballsfuture.com/phpBB2/viewtopic.php?t=488386

2011:

http://www.footballsfuture.com/phpBB2/viewtopic.php?t=442175

Very cool, Awsi. I have Kareem Martin as a top-15 player and the 2nd best DE in the draft. One area where I disagree with the writer on Martin: I don't believe his effort is inconsistent (82 total tackles is a lot for a DE and speaks to his motor), but his technique can be inconsistent.

I also agree on guys like Jeffcoat and Lawrence. Some people have Lawrence sneaking into the 1st, and that's silly. Pierre-Louis is underrated, but I'd be interested in seeing the formula that ranks him as the better athlete than Shazier.
 
Here are some more metrics evaluations from another site...

Its presented very well, and he is an eloquent writer, but when someone says "no, I won't release the formula I use" it becomes suspect.
 
Its presented very well, and he is an eloquent writer, but when someone says "no, I won't release the formula I use" it becomes suspect.

I don't think that's fair. This whole business of draft evaluation is a game, and if you've developed a technique that gives you an advantage (whether it actually does or not), you're under no obligation to "level the playing field" by sharing your formula.
 
I don't think that's fair. This whole business of draft evaluation is a game, and if you've developed a technique that gives you an advantage (whether it actually does or not), you're under no obligation to "level the playing field" by sharing your formula.

Well thats fine, but lets be real - if you wont show the math, then don't show the answer. That's like me telling you I have mathematical proof that Johnny Football will be a bust, but then I refuse to share that proof with you. We have all kinds of clowns on the internet pulling that one.

And in stats, its not the formula you use, its the data you use. The regression models are no secret, its what goes into them that matter.
 
Well thats fine, but lets be real - if you wont show the math, then don't show the answer. That's like me telling you I have mathematical proof that Johnny Football will be a bust, but then I refuse to share that proof with you. We have all kinds of clowns on the internet pulling that one.

And in stats, its not the formula you use, its the data you use. The regression models are no secret, its what goes into them that matter.

Or you can put it out there, live and die by the results, and have people judge you accordingly. Looking back at his rankings for previous seasons, he's done well above average.

I can appreciate the frustration, and I even mentioned that I'm interested to see how Pierre-Louis ranks as a better athlete than Shazier, but I don't think he's under any obligation. It's kind of like Awsi with his betting formulas. If I remember correctly, he had a radio show in Vegas (I think?), but he never divulged trade secrets. Awsi, please feel free to correct me here.
 
Or you can put it out there, live and die by the results, and have people judge you accordingly. Looking back at his rankings for previous seasons, he's done well above average.

Right, but when these models get vetted, they aren't vetted against themselves. You don't say its accurate because he was right most of the time, because he could be right most of the time without using statistics. You have be able to to say he was more right than a different model. You have to hold this guy's model up to another guy's model, aka an "f-test", to see the accuracy of his formula. Any wannabe Nate Silver I've seen puts their model out there for testing.
 
Or you can put it out there, live and die by the results, and have people judge you accordingly. Looking back at his rankings for previous seasons, he's done well above average.

I can appreciate the frustration, and I even mentioned that I'm interested to see how Pierre-Louis ranks as a better athlete than Shazier, but I don't think he's under any obligation. It's kind of like Awsi with his betting formulas. If I remember correctly, he had a radio show in Vegas (I think?), but he never divulged trade secrets. Awsi, please feel free to correct me here.

In the early going, late '80s, I made the mistake of blurting out too much. You want the audience to think you know what you are doing. I made the mistake of giving away my system of betting against any home underdog that had won its previous game straight up as home underdog, as well as a few other trends. A well known football/basketball writer and handicapper named Dan Gordon from New York was in the audience at the Stardust. He treated me to lunch the next day and warned me not to make the same mistakes he had. Gordon developed the zig-zag theory of betting the NBA playoffs. Basically you take the straight up loser of the previous game. Now there is an adjustment in the betting line for that trend, but Gordon discovered it prior to the adjustment. The public was actually betting it the other way so Gordon had a huge edge until he allowed the secret to spill. Gordon told me that based on the systems I had mentioned on the radio show it was obvious I had done lots of work uncovering them, so it was foolish to give them away on a public forum. This was before I used a computer so it was indeed tons of work by hand alone.

Larry Ness of the Jim Feist sports handicapping service took one of the angles on NFL totals that I had mentioned and used it as a tip on the weekly televised tout program. At least Ness was honest about it. He was a friend of mine from my early months in Las Vegas. Ness pulled me aside and said he used the angle on the show and hoped I didn't mind. He said Feist wanted trends like that but he didn't have any. The funny thing about that angle was that it had a fantastic percentage, well above 70% for many years, then almost immediately it flopped to barely above 50% as soon as I gave it out. I never understood that one. It had something to do with a road game after a string of division games going over the total.

One time in the same time frame I mistakenly gave out an angle and chief oddsmaker Michael Roxborough wrote it down and said, "Thanks. I'll make sure you never make any money on that again." I felt like a moron because it was one of my best systems. At that point I more or less shut up until one mistake many years later. I gave out a trend involving the Jeff Sagarin power ratings. My friend Dave was on the radio panel that night. When we went to commercial he smacked me on the leg and said he would kill me if I ever mentioned Sagarin again on that program. Using the Sagarin ratings was his pet method for not only betting, but figuring out which way the line would move.

So yeah, I can't really blame this Patriots fan for maintaining the formula as his own. It's so early in the metrics process in regard to the NFL draft there is potential for recognition and profit if you get the jump on something. ckparrothead told me he is working on a collaboration with a guy who used to post here and now writes metrics articles for rotoworld.

Anyway, regarding the numbers and conclusions, I've long been intrigued by Kareem Martin since I'm convinced there are hidden gems on any team recruited by Butch Davis. I attended the Miami/North Carolina game on Thursday night in Chapel Hill last October. They flashed athleticism all over the field.

Frankly I had forgotten that Ben Gardner was in this draft. He got hurt late last season. I bet Stanford all the time and I know that Gardner is very popular and well respected on that team. Based on those metrics numbers I'd have to put him near the absolute top of the all-underrated list. He could be available in a late round. I read the Stanford school paper online last night after finding those metrics numbers. The article said Gardner had lost 15 pounds leading to the draft to improve his pass rush potential.

I've been skeptical about Khalil Mack since I seldom like a guy who was rated low out of high school and now is atop the ratings. Based on all the metrics evaluations that place Mack extremely high, I'm probably off base on that one.
 
Last edited by a moderator:
In the early going, late '80s, I made the mistake of blurting out too much. You want the audience to think you know what you are doing. I made the mistake of giving away my system of betting against any home underdog that had won its previous game straight up as home underdog, as well as a few other trends. A well known football/basketball writer and handicapper named Dan Gordon from New York was in the audience at the Stardust. He treated me to lunch the next day and warned me not to make the same mistakes he had. Gordon developed the zig-zag theory of betting the NBA playoffs. Basically you take the straight up loser of the previous game. Now there is an adjustment in the betting line for that trend, but Gordon discovered it prior to the adjustment. The public was actually betting it the other way so Gordon had a huge edge until he allowed the secret to spill. Gordon told me that based on the systems I had mentioned on the radio show it was obvious I had done lots of work uncovering them, so it was foolish to give them away on a public forum. This was before I used a computer so it was indeed tons of work by hand alone.

Larry Ness of the Jim Feist sports handicapping service took one of the angles on NFL totals that I had mentioned and used it as a tip on the weekly televised tout program. At least Ness was honest about it. He was a friend of mine from my early months in Las Vegas. Ness pulled me aside and said he used the angle on the show and hoped I didn't mind. He said Feist wanted trends like that but he didn't have any. The funny thing about that angle was that it had a fantastic percentage, well above 70% for many years, then almost immediately it flopped to barely above 50% as soon as I gave it out. I never understood that one. It had something to do with a road game after a string of division games going over the total.

One time in the same time frame I mistakenly gave out an angle and chief oddsmaker Michael Roxborough wrote it down and said, "Thanks. I'll make sure you never make any money on that again." I felt like a moron because it was one of my best systems. At that point I more or less shut up until one mistake many years later. I gave out a trend involving the Jeff Sagarin power ratings. My friend Dave was on the radio panel that night. When we went to commercial he smacked me on the leg and said he would kill me if I ever mentioned Sagarin again on that program. Using the Sagarin ratings was his pet method for not only betting, but figuring out which way the line would move.

So yeah, I can't really blame this Patriots fan for maintaining the formula as his own. It's so early in the metrics process in regard to the NFL draft there is potential for recognition and profit if you get the jump on something. ckparrothead told me he is working on a collaboration with a guy who used to post here and now writes metrics articles for rotoworld.

Anyway, regarding the numbers and conclusions, I've long been intrigued by Kareem Martin since I'm convinced there are hidden gems on any team recruited by Butch Davis. I attended the Miami/North Carolina game on Thursday night in Chapel Hill last October. They flashed athleticism all over the field.

Frankly I had forgotten that Ben Gardner was in this draft. He got hurt late last season. I bet Stanford all the time and I know that Gardner is very popular and well respected on that team. Based on those metrics numbers I'd have to put him near the absolute top of the all-underrated list. He could be available in a late round. I read the Stanford school paper online last night after finding those metrics numbers. The article said Gardner had lost 15 pounds leading to the draft to improve his pass rush potential.

I've been skeptical about Khalil Mack since I seldom like a guy who was rated low out of high school and now it atop the ratings. But based on all the metrics evaluations that place Mack extremely high, I'm probably off base on that one.

That's probably Greg Peshek, formerly known around here as NU_Gap. His articles are excellent. His "adjusted accuracy" pieces on quarterbacks -- where he tries to create a unified system for completion percentage by adjusting the depth of pass attempts -- are particularly good. A shame he couldn't find a niche here. This site rewards effort threads that align with the received wisdom of the moment but isn't nearly as kind when you go against the grain, no matter how much you line up your ducks in a row. Not a surprise, unfortunately, but I can see how it would grate a numbers guy like Peshek to see his work trashed by people who can barely count.
 
Interesting. What is this based on?? If a team gets 4th and 1s 69% of the time, would that be considered a metric that would indicate they should go for it more often?? If so they are way off. Just because it is above 50% does not mean they should take the risk. See the risk is not 50/50. They are gaining a yard 69% of the time but losing 50 yards of field position 31% of the time. In this example you would have to be picking up 1st downs 98% of the time to even the risk. That was just an example, with many other vairables but I was curious if that was what the metrics was saying.

That is the biggest problem with stats. It is not the stats fault that people use the stats for things the stats don't tell. The stats only tell one part of the story, and are only useful if the person using them knows how to use them correctly.

You are correct, the calculations involve the conversion percent but also expected points per drive by field position for both teams. That is, what are the expected points the other team will score if you do not make it. What it does not involve is the specific offense/defense matchup and game context, but overall it's sound math.
 
This thread has gone way over my freakin head - LOL

I was just trying to keep people busy before the draft - :chuckle:
 
Back
Top Bottom