DawgTalkers.net Forums DawgTalk Pure Football Forum What Football Analytics is

He had a trade with STL that he passed on. But yes, overall not a good draft.

If this model works and we are indeed a clearly improved team next year, talent wise, roster wise and etc - I wonder how many, if any, other teams that are struggling would consider or adopt this model?

All I know is I don't really have much faith in it, but I likely wouldn't have faith in anything factoring we've not hit a home run draft since like... ever it feels like. Fingers crossed Haslam has made a decision that will help and bring continuity as opposed to bringing another regime change, crap season and blow up.

jc...

I think many people ... the media included are missing one key point or wave that is coming with football too. Player tracking on the field was implemented and tested last year. There will be millions of data points during a game to be analyzed. To me this is the Browns being smart and getting ahead of the curve. Much like the Indians have and did with Pitch and Statcast data. To read more about it http://www.wired.com/2015/08/nfl-players-getting-rfid-chips-season/ Imagine what you could do with that!? Its exciting to a data geek like me. How wide open are receivers running.. Whats the alignment 3 sec before snap... at snap? Who are the top 5 fastest players on the field with pads on, on Sunday? And the list goes on.

Analytics is a tool... its not the end all be all. We use it and I provide it where I work. But its just one of the opinions in the room.

The counterpoint to the Mingo comment was that analytics would've (supposedly) got us Bridgewater instead of Manziel, and probably at the same pick.

And we were at least using performance tracking in practice at at least one point. We used it to keep Miles Austin's hamstrings from popping when he was with us.

Andrew Hawkins talks about performance tracking

Bridgewater?

Since the thread featured Dalton as the example of a current metric superstar, how would a team, drafting a QB in the AFC North, use metrics to determine who can win in the playoffs?

You currently have two quarterbacks who are big, tall, can throw the ball deep and vertically with velocity.

They both have won Super Bowls.

You have a third one who apparently is off the charts when it comes to analytical metrics. Would you suggest we would be better off drafting him? The guy who cannot make those extremely tight throws, at playoff intensity, once defenses push, squeeze, and play twice as fast? The guy who has yet to get a playoff win despite a far more talented team than the other two have?

I admit, Dalton is throwing much better this year after working with Tom House in the off-season. And, he has an excuse this year when they get sent home from the playoffs.

Bottom line: Joey Noodlearm will not allow you to get deep into the playoffs regardless of his efficiency.

There are some metrical standards that take a back seat to what your eyes and experience tell you. (That isn't saying there isn't a good deal of value in analytics, I just think its a tool you use before you go back an re-watch the video. Its just a good tool. Its not the master carpenter.)


- - - - - - - - -

Teddy, loved your remarks last year about not being drafted by Cleveland. Gonna love watching you throw the ball, in playoff mode, against Seattle's defense this weekend. Check down on every play and hope for a good efficiency rating.

Yeah, Teddy stinks. They had 5 wins the year before they drafted him. They had 11 this year, but he had absolutely nothing to do w/that.

Analytics better be rolling already on Lynch vs. Goff. We getting one.

The first article was showing the expected dropoff between Dalton and AJ McCarron. Are you saying that McCarron should perform better than Dalton?

Maybe not. I like both, but I like Dak Prescott in the 2nd round.

I've got one of those speculative questions that some of you hate so much.

If someone was to do an estimate, an analysis, of possible/potential metrics of the Cleveland Browns offense next year with Johnny Manziel and Josh Gordon both on the field together what might it look like? (Just take for granted that a miracle happened and we took them under our wing, handled them correctly, and actually were able to rehab them).

Any thoughts? Football thoughts?

What do you think could happen?



----------------

Teddy, playoff football, against a playoff caliber defense.

Boy, I've been waiting for this.

That sounds hateful. You want him to fail so you can rub it in? Really? Or did I misinterpret you?

And would a playoff loss to Seattle, who made Carson Palmer look like a chump last week, nullify Teddy and the Vikings winning 11 games this year?

I got news for you............Teddy and the Vikings are not a finished product. They are on the cusp. They are not a true contender this year, but they played well enough to make the playoffs after finishing only one spot behind the Browns two years ago. They went from winning 5 games to 11 games, while the Browns went from winning 4 games to winning 3 games..........and you are REALLY going to run your mouth about why we should not have taken Teddy over Johnny???

Dec 6 Final
Seahawks 38
Vikings 7

The Vikings might do better this time around, but Seattle is in a different league and peaking at the right time.

You can put all of that on the rookie QB if you want to, but I would think that the rest of the players have a little to do with it. Just getting to the playoffs is about 3 times better than what we accomplished.

I have question about analytics. What if there is a college player, for example, an OL player. Let's say the analytics are weak, but he has an obvious flaw in technique that is to blame for that. What leeway is there for a position coach to say "I can help this guy learn how to overcome this flaw and become an excellent player, so analytics don't matter as much."

Great point about Seattle peaking at the right time. Kinda like Ohio State did last year.

That's a good question.

Maybe thats where the coach and whoever they hire as the football side of it would come together and make that determination. I dont think it will be strictly..hey the computer is telling me this and overlook the human element all together.

My analysis is best buds Johnny and Gordon would be out every night during training camp.. both get into the NFL drug program getting Gordon banned from football for life and Manziel getting the 1st 4 weeks of season under suspension with a second 1 year ban coming during week 2 of his original 4 week suspension because he was caught in Vegas wearing a red wig and a dress while learning to pole dance while drinking a bottle of champagne...

"I despise Gilbert"

"Josh Gordon is a punk"

That is what I consider direct, hateful, meanspirited speech.

They are direct quotes taken from a guy I refer to, off the board, as the Crown Prince of the Thread Hijack. As a citizen of this country I respect his right to say that, but I don't respect him for saying it. I do not despise Teddy Bridgewater. I did despise his classless, unnecessary, kick in the teeth towards the city of Cleveland. And, I appreciate that very effective hijack of the Manziel and Gordon point. My signature touched on pure football.

Can't really speculate on what the analytics department could come up with if they ran simulated computer models of what Manziel and Gordon might do? What might happen to the running attack? How the offensive line might produce? Tight End play underneath when linebackers are chasing the QB? How it might affect the draft so we could pick up some defensive players who can actually tackle?

Guess not.

Good question with a not so easy answer.

First the analytics really only deals with the flaws in the stats, so you aren't going to see something where it's going say a guy with good stats stinks except in one case, that I will get to in a minute.

What it is meant to be is a tool so that you don't blindly follow stats. So if the guy has bad technique and bad stats it should work out the same way it does now. If you have a good football guy that thinks he can fix the flaw, you'd still take that guy.

There is one case where this isn't true and it comes with a story that I read just yesterday and that is with combine stats.

When Jarvis Jones ran a bad 40 time in the combine, the Steelers were celebrating because they knew he would fall to them in the draft. The thing is that his production was awesome in college and they were certain that it would be in the NFL as well because he had NFL size.

What they didn't take into account was all of his other combine numbers stunk too other than the Broad Jump, which was barely average. This is a huge red flag in the NFL and usually means that the player isn't athletic enough to make it in the NFL. You can be low on one or two things, but not on all of them. I forget the fail rate that they posted, but it was very high. Still a football guy could point out that the guy is sick or injured and over-rule even that I would think.

All the leeway in the world. I think too many have jumped to the conclusion that overnight the Browns are going to pick their players by the numbers alone.

Analytics can't see the player play. Can't see the hips of a CB or the route running of a WR. Analytics are but a tool to aid the football guys, not replace them.

Gordon has a body of work and would be easy to predict as long as he played the full season. Personally, I haven't totally written the guy off, but he's pretty close to the last straw with me and if we got anything decent in trade value for him I wouldn't be unhappy.

The last thing with Manziel was deliberate and because of that, I wouldn't give him that additional chance personally (along with Gilbert for quitting during a game) but playing devil's advocate and just looking at his stats and trying to project, I wouldn't think he would project to be very good. His stats weren't that great and although you can look at his footwork and throwing motion and say that is fixable, the fact is that he doesn't want to put in the preparation that is needed. The football guy Pettine was right and I think analytics would agree.

You think that they don't account for these sorts of things?

Bridgewater definitely went to the right team, a team on the rise, with a very strong defense, and a Hall of Fame RB.

I am not saying that he stinks, because he definitely does not, however, he has been a below average starting QB thus far in his career. However, he has played within their scheme, and has done what they need him to do. (be careful, don't turn the ball over, etc.)

He was 22nd in passing yardage. He threw only 14 TD passes this year, good for 26th in the NFL. He was 21st in passing attempts. As a comparison, Josh McCown threw 12 TD pases in 8 games. I think that the only QB in the playoffs with fewer TD passes are Peyton, Osweiler, and Hoyer, all guys who missed significant games this year.

The Vikings were able to be very careful with Bridgewater. They did not have to force him into carrying the team.

The Vikings were 4th in the NFL in rushing, and were 4th in scoring defense. These 2 are larger factors as to why they are in the position that are in the playoffs.

I think that Bridgewater might have fit in the ideal team we wanted, a team that runs the ball and plays defense, and minimizes the QB position, but he is not yet a great QB. Thus far in his career, he has actually been below average when compared to other QB in the NFL. The advantage he has is that the Vikings seem committed to allowing him to develop as he plays. I don't think that he would have got that here.

Exactly. There will always be the human element involved. It's not like we crunch the numbers and then a computer spits out who we are going to pick.

If anything, it helps identify players with high numbers that are inflated for some fluke reason and players who might grade out low who in reality are solid players. Sometimes players get good results from bad decisions and bad results from good decisions.

what a novel approach.. play a player to his strengths while letting him work on and develop his weaknesses

I know very little about how, specifically, analytics are going to be used here. It's something I need to look into more, watch the video that's been posted, etc. However, I do think it is interesting to look at how statistical analysis and such has been used in other sports, games, and fields for comparison's sake.

The most obvious is the baseball example with the Athletics. I happened to play college baseball under an old-school type coach. I understand the nuances of the game and there are a lot of things that are not easy to quantify, and if we could go back in time 10 or more years, I know that the overwhelming majority of coaches would be adamant that you would NOT be able to quantify a player's projection even though we now know that teams have done exactly that with great success.

Yes, baseball is simpler to break down as a player's performance is far less dependent on his teammates than it is in football for example. It can still be done; it just takes better technology and a more advanced method and all that.

Are there any poker players in here? For many years there has been discussion on whether a bot (a completely automated program with no human input) could profitably play poker. It is actually amazing to see how this has developed. Ten years ago most skilled no limit holdem players would have adamantly said that there is no way that a bot would be profitable in the games, because it would be impossible to program all the nuances in the game-- position, differential stack sizes, the effect of blinds, antes, tournament structure, what your opponent is thinking/what his range is, etc etc. I was one of the few that disagreed with this, that it was only a matter of time before programming skill and computing technology reached a sufficiently advanced stage where even highly-skilled human player would be inferior to a good bot (this happened years ago, fwiw). It's kind of like a continuum:

Heads up limit hold'em is relatively simple, and while that is still immensely complicated, it has essentially been solved. The best heads up limit bot plays unbeatable poker.

Heads up no limit and full-ring limit are far more complicated as there are many more variables to factor in. And full ring no limit is the most complicated of them all, but as per above, there are bots out there that are long-term winners even in very tough online games (with online players being, on average, far better mechanically than live players). Believe it.

So I guess the point of all this rambling is for those who are skeptical, keep an open mind. Realize that computing and technology grow at an exponential rate, and we are still in primitive territory. This sounds odd, given all the recent progress that has been made (in this case, Moneyball in baseball and some limited inroads into the NBA) but that is because we can only look back and can't yet see the even more rapid improvements that will happen in the future.

edit: as far as football specifically, there are actually some pretty good analytics available publicly. Footballoutsiders has some good material, as does Profootballfocus and AdvancedNFLStats (which is AdvancedFootballAnalytics now I believe)

One truth I've discovered while working in software development that holds true in mathematics and analytics.....
If you can sufficiently describe something, you can program it to the extent that you can replicate its functionality and processes completely. If you can do that, you can measure it and its efficiencies. You can track progress & trends, you can do functional comparisons.
Once you do that with enough various items, you can pull them together to measure & quantify a larger whole.


A quick example that just popped into my head:

Let's say we're looking for a new backup OLine that can fill it at Center/Guard. What things would you look for?

The old school might say things like height, weight, 3-cone time, bench press, standing high jump, long jump, etc.... all those Combine things (and let's face it - the Combine is the NFL's earliest, most primitive venture into analytics).

The new school might take a look at the NFL as a whole in that position and find out:

• what type of quickness it takes to have a better than average chance of success in getting a seal block back to the left when the whole line is set to move to the right to create a running lane.
• It might measure ratios of player height to pad height & arm length in the NFL to identify a truly optimal build for a given position.
• it could measure those prospects coming out and tell you who is going to play with good leverage right away and who doesn't
• it can tell you what sort of quickness is needed to make specific types of drops/blocks
• it can measure snap-reaction times and what is a solid baseline minimum... a lineman that reacts .05 seconds slower than another is actually at a disadvatage.... that can be the difference in hand placement by as much as 4-6" in the same timespan, or a difference of that much in arm extension.

Now you have data points to look at that describe what a successful OC/OG looks like to go with your classical list of height, weight, etc that you can take with you to the combine and add to your process of interviews and private workouts. You can filter out and indentify players with a higher chance for success before you ever even talk to them.

You absolutely break it down further. You add in as many things as you can. The more measurables you have, the clearer the picture you have. You can even combine measurables in time... i.e. when attempting to look at ability to get hand placement, it will be a combination of reaction times and arm length. Shorter arms will be able to get into place and get extension a little quicker (aside from the fact that they allow the defender to get into you and ruin your leverage). So you don't want to use each measurable in an isolated way, you want to look at them the way they actually matter. The more information you have, the better.


Short arms - classic example is Robert Gallery. Absolutely TERRIBLE as a Tackle, but he did well at Guard.

To try to answer the earlier poster with the "what if" scenario, I would imagine this analytics stuff is similar to how I use statistics at work, just more complicated. After the numbers yield a result, those results need to be interpreted. It's not like the equations spit out a thumbs up or thumbs down. The results you see, be they good or bad,need to be interpreted. You do that by having a football guy, once they understand what the numbers are saying, step back and look at the player and explain why the results are what they are.

They could say something like, "wow, these results are terrible, but that could be because of this technique that is easily coached", or it would be, " wow, these results are terrible, but the guy next to him was awesome and picked up the slack".

Just an add-on thought here:

Once you are able to quantify what you should be looking for in an individual player, and can then get that data on all prospects, everywhere... you can now easily and instantly apply your search to all schools, large & small, regardless of competition level they played against. If a guy has the size & quicks, he has the size & quicks.... it won't matter who he did it against because subjective things like that can be ruled out (at least initially, anyway - you may want to keep the "eyeball test" in as a tiebreaker).

MIT SLOAN Sports Analytics Conference Feb 28 - Mar 1 - 2014

Increasingly, achieving success in the National Football League is driven by an organization's ability to incorporate data and analytics into daily operations. The football analytics panel will delve into this statistical revolution by asking accomplished panelists from the front office and on the field how analytics influences at-the-line calls, adjustments/reads, in-game decision making, personnel evaluation, coaching hires and contract negotiations. From personnel to coaching to administration to players, all views will be represented as we explore the opportunities and challenges of applying analytics to America's most popular sport.

The panel is made up of the following people:
Scott Pioli - Atlanta Falcons
Kevin Kelley - Pulaski Academy
Brian Burke - Advanced NFL Stats.com
Parag Marathe - San Francisco 49ers
Suzy Kolber - ESPN - (Moderator)

Wow, an hour.... I'm going to have to set aside some time to watch this tonight, but I'm really interested in this. Thank you!

That last 10 secs where that talk about the QB spot sums up why the Browns are constant losers.

Also at about 51 mins and 30 secs they talk about signing players with distractions. I wish a few guys on this board would listen to Pioli's thoughts on that one.

I'm glad they said it bluntly. If you don't have a quarterback, don't even bother showing up. It's the truth that some fans still want to fight.

We shall see. I know these guys have no background in watching film or scouting players. So ALL of that will have to be outsourced. If it works? GREAT. I'll believe it when we go 12-4. Until then I have SERIOUS reservation about these hires so far.... we will see very very soon how good this group is. Starting with the HC hire. And the pseudo "GM" followed by the draft. Another Mingo(Banner was a huge fan of analytics) and I will pull my damned hair out. I can't afford that, I'm getting to the age it might not grow back lol

Absolutely, both years of Pettine we played players out of position. Last year they played Phil Taylor at END. Hell, this season at least 80% of the time we ran a 4 man front and we didn't have a single 4-3 DE on the roster. We let the only one we DID have go in FA(Sheard) and he goes to the Pats and has 8 sacks. It was like no one knew what they were trying to do....

that video really raised my opinion of scott pioli