Over Valuing Analytics in Football

this was literally Jason Garrett's philosophy: just line up against the man in front of you and beat him. That didn't work out too well for us or Garretts career.

Come on guys, are we really going to waste time responding or debating with someone whom is just looking for reasons to discredit something? The one and only thing I will say on this is that we've had a coach in Garrett using philosophies from the 90's that might of worked for a team that was stacked before the salary cap took place, but this approach was laughable.

So now we have a proven winning coach whom has spent a year putting together analytic analysis that can be of potential help against certain formations, down and distances, and we should find fault with this in some way? PLEAZE.....

aikemirv said:

Understand. I just really don't care much for these probability of winning %. Analytics in general to me would carry very little weight in my game decisions. I would be much more cognizant of specific game flow than any analytic tendancy that was produced over the course of many different situations, different game conditions and different teams.
I, as a coach need to have a feel for what my team and the opponent is doing that day.

If I am 7-10 on third down for the day or 2-10 on 3rd down for the day, that is going to have a huge effect on my decision making over some analytic of winning % based on going for it or not. Someone was clamoring for us using analytics in the NE game in horrid weather conditions. Those conditions makes the analytics pretty much useless!

Click to expand...

I called it!

I feelz it!

@ScipioCowboy

eromeopolk said:

To all, my point is that Analytics is not new. Since the Landry days, play calling tendencies have been tracked in the NFL for an edge. Now that you have it on an IPad does not make it new. The coach saying he is big on analytics should have been using statistics before it was called analytics.

How much analytics did it take for Jimmy Johnson in the 1992 NFC Championship game to say "I want to win" to Norv Turner after the 49ers got within 10 points? What analytics was used for Norv to call the most often used go to pass play in the Cowboys playbook to gain a 1st down. Or, Where was the analytics after being down at halftime in the Super Bowl to the Bills did it take for Jimmy and Norv to run the football with the NFL MVP RB they neglected in the 1st half?

Football is more about putting your playmakers in position to make plays and the instincts to call those plays. That is not analytics. That is coaching. Since 1994, Cowboys have not had leadership at Head Coach and a real Head Coach (except when Jerry had to get his stadium and hired Bill Parcells).

Analytics does not give a Head Coach leadership, and game instincts to win.

Click to expand...

This would have been a better OP.

Vtwin said:

I agree completely. "Analytics" has a place but it can water down a specific situation to a one size fits all scenario that ignores the factors specific to that situation.

Coaching/Management got crucified on the Zone for saying they don't pay attention to analytics when referring to that specific fourth down play near the goal line. They absolutely shouldn't have paid attention to a boiled down stew of many many similiar but not equal situations to decide what to do in this moment, with these players, in this stadium, against this defense, in this weather etc etc etc.

Instinct born of experience matters a great deal.

Click to expand...

aikemirv said:

Yep, I don't want analytics to tell my coach what to run on 4th and 1 with the game on the line. I want him to be able to have enough feel for the game to know what is going to work based on his teams performance that day!

Click to expand...

A couple things.

That all sounds good, but the fact is that Garrett wasn't going by "feel" or "this moment, these players, etc." when ignoring the analytics and making his decisions. He always, without fail, made the most conservative choice. This is a man who never, not once in ten years, went for it on 4th-and-longer-than-2 when it wasn't desperation time. (Okay, he did it once in the playoffs). So don't give me this stuff about how he was going by feel of the game. This was true to a lesser extent for almost all coaches: they were not just conservative in a way you could justify by "feel." They were ludicrously conservative in a way that you simply can't defend. It's only started to change recently as more of them pay attention to the analytics.

Why do people think that a pro football team's analytics department would generate "one size fits all scenarios that ignore the factors specific to the situation?" They're not calculating generic averages like we see reported in public. They have tremendous time and resources to take into account all sorts of factors and develop recommendations for all sorts of situations. Sure, the coach should be in a position to say, "I'll take the 45% end of that estimate instead of the 55% because of what's happening on the field at this particular moment," but the information is invaluable. If you're not getting the best information you can and not taking that information into account, you're not doing a good job.

(Of course, there's a lot more to analytics than just 4th down decisions, and the idea stands: you want the best possible information so you can make informed decisions).

Never a big fan of these "thing I made no effort to understand must be dumb" type takes in football, or anything else.

nyc-cowboy said:

The thing about this situation is, if you fail to make the first down are you giving up the ball at your own 33 to a Tom Brady in his prime or to a Mark Sanchez.

Click to expand...

Not that it matters but in this example you do have Rodgers as your QB

aikemirv said:

Would love to see the amount of data they have on that specific situation.

1) It would be very rare for a team to go for it on their own 33 with 4 minutes left
2) ton of data for a team kicking it in that situation

- so the win percentage of kicking it really can't be compared to the win percentage of going for it because the sample size is not large enough on going for it on 4th and 2 to make the comparison.

That's my take on that scenario.

Click to expand...

That's a really weird argument. You don't have to have exact matches of down, distance, field position and clock to get useful data.

There's a ton of data on going for it on 4th and 2, and for that matter on 3rd and 2 in situations where you'll be punting if you don't make it. Clock and field position are irrelevant to that data. (Okay, you may want to discount plays run close to the opponent's end zone because there's less field to defend).

There's a ton of data on what happens when teams have 1st downs at different positions on the field, and about what you can expect from their running and passing matchups (assuming they'll be mostly running in this situation).

In this particular situation, the Packers willingly gave up the ball and never saw it again, because the Seahawks got two first downs. If they hadn't gotten the second first down, the Packers would have gotten the ball back with about 80 seconds left and no time outs.

I stopped reading when you said that Football probabilities are not as numerous as Baseball.

JD_KaPow said:

That's a really weird argument. You don't have to have exact matches of down, distance, field position and clock to get useful data.

There's a ton of data on going for it on 4th and 2, and for that matter on 3rd and 2 in situations where you'll be punting if you don't make it. Clock and field position are irrelevant to that data. (Okay, you may want to discount plays run close to the opponent's end zone because there's less field to defend).

There's a ton of data on what happens when teams have 1st downs at different positions on the field, and about what you can expect from their running and passing matchups (assuming they'll be mostly running in this situation).

In this particular situation, the Packers willingly gave up the ball and never saw it again, because the Seahawks got two first downs. If they hadn't gotten the second first down, the Packers would have gotten the ball back with about 80 seconds left and no time outs.

Click to expand...

Clock and field position are irrevelant to the data? - seriously - that's just stupid. If you are not looking at game situations - then you are just throwing a dart blindfolded!

So, the Seahawks, who had scored 5 times that game already, made 2 first downs. Seattle is running the ball at over a 5 yrd per carry clip. That surprises you - it does not surprise me? The in game performance of GB's defense told me all I needed to know of that situation. My defense is not likely to stop them. They scored a FG and a TD on their last 2 drives! I am on the road and this could be my last possession. I have Aaron Rodgers and his chances of getting a first down are greater than the chances of my defense stopping them..

I don't need outside analytics influencing my decision there!

McCarthy made the wrong decision but it had nothing to do with analytics of situations prior, he probably just did not want to look like a fool going for it because most coaches would not in that situation!

Bachman05 said:

This is example of where analytics come into play , I got this form another forum that I frequent and share to provide some insight on how it can be applied.

Week 10, 2018 GB @ SEA. GB is 4-4-1 and McCarthy is firmly on the hot seat, coaching for his job.

4:20 and 1 TO remaining in the 4th, GB trailed 24-27. Following an incomplete pass on 3rd and 2 he decides to punt from his own 33.

By deciding to punt he is requiring his defense to get a stop to have a chance to win the game.

Had he decided to go for it and failed, a defensive stop would have still given his team a chance to win with a TD on the final possession. The punt cost GB 9.8% win probability (31% vs 21%)

Click to expand...

You can't always just go by that. I don't remember the game, so to me it looks like the right decision. But what if the packers didn't get a 3 and out all game? What if they were being gashed on the ground? Does it take these things into considering?

Flamma said:

You can't always just go by that. I don't remember the game, so to me it looks like the right decision. But what if the packers didn't get a 3 and out all game? What if they were being gashed on the ground? Does it take these things into considering?

Click to expand...

No, it can't take current in game factors in consideration and the Packers had forced 4, 3 and outs in the game. The Seahawks were averaging over 5 yards per carry and had scored on the last 2 possessions and both were long drives, one 13 plays for 97 yards and one 7 plays for 75 yards. GB's defense had to be gassed on the road! That's what I would base my decision on as a coach, not some win % over the course of ever so many years.

Bachman05 said:

Not that it matters but in this example you do have Rodgers as your QB

Click to expand...

Thats the point - who you have on the field - but even if you have Rodgers, Brady scores and now you down by 10, with prob little time left.

Point is there are many variables besides just numbers.

eromeopolk said:

I will post my thoughts on the new Head Coach after his press conference and staff is announced (or majority is announced).

However, I have been hearing the buzz word of Analytics in football and I laugh because it is basically statistics which has been in football since Engineering major Tom Landry was coaching. When you hear a coach talk about getting a analytics group or using analytics like it is some new thing...BEWARE.

First, on offense you can only do one of 2 things on 1st, 2nd, and 3rd down which is run or pass. You have three choices on 4th down which is punt, run or pass. So the probabilities are not numerous as in baseball. Lets get real, this is a sport that uses X, O, arrows and lines for communication in a playbook.

Tom Landry took analytics out of defense with the flex which was read, then react, and flow your linebackers to the football. Jimmy Johnson a psychology major was attack and have the offense react to you. That is pretty much all you can do on defense.

However, because there are timeouts, 15 minute quarters, play clocks, and 2 minute warnings, time management is important. But that is not analytics, that is time management.

So, on offense 4th and 1, on the opponents 30 yard line, with 45 seconds left, down by 4 with 1 time out with Ezekiel Elliott at RB and Dak Prescott at QB what will you do? Run, Pass, or kick a field goal? Jason Garrett probably kicks a field goal. Jimmy Johnson probably goes for it on 4th down. Instincts give the information for statistics/analytics. So if you want to know the answer.... it is below....

Click to expand...

In today’s game they can measure every players speed, arm velocity and jumping ability on every play with cameras. So play design is both tendencies and now utilization of sports science. For example if you know a team is in zone then you know your WR is fast enough or not fast enough to outrun the zone based on how deep they typically play.

aikemirv said:

Clock and field position are irrevelant to the data? - seriously - that's just stupid. If you are not looking at game situations - then you are just throwing a dart blindfolded!

So, the Seahawks, who had scored 5 times that game already, made 2 first downs. Seattle is running the ball at over a 5 yrd per carry clip. That surprises you - it does not surprise me? The in game performance of GB's defense told me all I needed to know of that situation. My defense is not likely to stop them. They scored a FG and a TD on their last 2 drives! I am on the road and this could be my last possession. I have Aaron Rodgers and his chances of getting a first down are greater than the chances of my defense stopping them..

I don't need outside analytics influencing my decision there!

McCarthy made the wrong decision but it had nothing to do with analytics of situations prior, he probably just did not want to look like a fool going for it because most coaches would not in that situation!

Click to expand...

Exactly why analytics help , with this info available to him he might of based his decision on it and not his feelings.

Point being having more info to gain a slight edge to win I'll gladly accept

nyc-cowboy said:

Thats the point - who you have on the field - but even if you have Rodgers, Brady scores and now you down by 10, with prob little time left.

Point is there are many variables besides just numbers.

Click to expand...

The point is , you have to stop them if you punt or don't make the 1st , you're toast either way if they score

Lol. Love it when old people start threads.

Analytics and metrics a very, very important to any business who values quality assurance. McCarthy comes from a QA background. Their matra is "what doesn't get measured, doesn't get changed."

I don't know what the OP is taking about that over-valuing analytics. But we have ignored it for so long. So it's refreshing that we have an old school coaching that's embracing and is implementing it as part of his coaching repertoire.

Bachman05 said:

Exactly why analytics help , with this info available to him he might of based his decision on it and not his feelings.

Point being having more info to gain a slight edge to win I'll gladly accept

Click to expand...

I just don't see it because if the score and teams were reversed and I were Seattle I would have definitely punted because I had pretty much shut them down in the second half and I was at home with a crowd that makes a difference!

Bachman05 said:

The point is , you have to stop them if you punt or don't make the 1st , you're toast either way if they score

Click to expand...

Well in the scenario, when Brady takes over, he's basically already in FG range - so say you do stop em - they still get the FG (up by 6) and now you have to get 7.
Now if you punt and stop em - now they have to punt it back and you still only down by 3.

Again we can go in circles here with a bunch of what ifs - point is do you just go by the numbers or does who you have on the field matter.