Analytics is a complete failure & HCs are retarded

[lod001 1,235]

Title change to better show the ignorance of todays NFL.

This weeks loser is Cincy. The Chargers should have joined them but the Browns inept play calling at the end caused them to have to try a 54 yarder instead of an easier one. They had 1:10 to get 15-20 yards and make it less than a 50 yarder but they called a deep out like idiots. Wasted down.

Week 4 , the Falcons were the idiots of the week by going for it on 4th & 3 from the 4 instead of kicking the game tying FG.

The Lions went full on stupid at least once week so far. Their idiot coach made 2 bonehead calls that were fireable calls.

Can't remember the other stupid go for it on 4th down moves so far but it's been rampant and the majority have failed.

Analytics takes nothing into account as far as the rank of the defense you are playing against, weather, etc. With the large number of 4th down failures costing games this year, it should be pointing to 'take the points' and 'punt the damn ball' by now. Otherwise it would be more complex like 'when going for it on 4th down vs a top 10 defense, the analytics says to punt or kick the FG.'

Raiders lose going for 2 when down 1.

Week 6. CHI goes for it at the 1. Fail. -3 points. Forced them to score a TD to win instead of a FG.

[nobody 2,031]

Only 1 thing to say about this.

PREACH!!!

[AxeElf 787]

Assuming that they are using the right stats to analyze, the "analytics" can't fail.  The problem only appears when people expect the analytics to recommend the winning action each and every time, when it can only recommend actions based on probability.

If I say, "Do this and you'll win 90% of the time," and then you do it, and you lose, was I wrong?  Of course not; it was just one of the other 10% of the times.

The percentage advantage gained by going for 2 when down by 8 points is pretty slim over just kicking the PAT, but it is a percentage advantage.  A team may have to try it 100 times both ways before they see the advantage to the former, but it's still there.  Analytics doesn't "fail" when going for 2 doesn't lead to a win--the advantage was gained.  It just wasn't enough.

[Ray_T 587]

analytics are just statistical analysis used to determine chances of being successful in certain situations.

but those analyitics have to be combined with a certain amount of common sense and a coaches instincts based on how his team has played that day.  analytics dont take these things into account.

if most of the year your team has won the battle on the line the analytics will tell you to gamble on 4th and 1 everytime. but if you are losing the battle on the line that day and/or your line isnt as good as it is most days, those percentages will be wrong on that day.

yeah, if you can push back the other line 2 yards every time gambling on 4th and 1 would seem to be a no brainer but the game and the world dont work that way

[easilyscan 552]

Kinda like technical analysis in the stock market. 

[nobody 2,031]

3 hours ago, AxeElf said:

Assuming that they are using the right stats to analyze, the "analytics" can't fail.  The problem only appears when people expect the analytics to recommend the winning action each and every time, when it can only recommend actions based on probability.

If I say, "Do this and you'll win 90% of the time," and then you do it, and you lose, was I wrong?  Of course not; it was just one of the other 10% of the times.

The percentage advantage gained by going for 2 when down by 8 points is pretty slim over just kicking the PAT, but it is a percentage advantage.  A team may have to try it 100 times both ways before they see the advantage to the former, but it's still there.  Analytics doesn't "fail" when going for 2 doesn't lead to a win--the advantage was gained.  It just wasn't enough.

Exactly.  Garbage in; garbage out.

My take is the analytics don't consider the manalytics when figuring and feeding their bullshìt models.  I.e., you can't assume the offense has the same focus as the defense when the defense has their back against the wall and the offense doesn't.  I did a whole thing on this in the bengals/ravens thread, but in the history of NFL these situations had both the offense and defense with their back against the wall, so the history is tainted for use and nerds who have never played a sport don't understand there is a mental element to it.

[AxeElf 787]

33 minutes ago, Ray_T said:

if you can push back the other line 2 yards every time

I still don't know why some NFL team doesn't go to Japan and sign a couple of those 500 lb Sumo wrestlers just for this purpose.

They wouldn't be able to do pass protection, but they've spent their lives learning about leverage and how to drive other 500lb men backwards ten feet.  You can't tell me that a couple of those bad boys at RG and RT couldn't push JJ Watt back two yards into the end zone (or over the line to gain) virtually every single time?

Seems like a no-brainer.

P.S.  200 kgs is 440 lbs.

[Ray_T 587]

probably too slow.  though it may be worth it to try if you are a last place team.  why not give it a go?

[Gepetto 976]

Teams should practice certain plays (4th down and 2, 4th and 5, 4th and 1) repeatedly and successfully over and over so that their so confident with the execution aspect of it and the design of their go to plays that they succeed 90% of the time.  Otherwise, kick, unless their kicker just missed two in a row or the distance is too far, or the wind is extreme.

[polecatt 399]

They've failed any team that is foolish enough to put too much emphasis on analytics.

When it comes to making decisions and play calling, analytics is just one of many tools a coach should use.

Most decisions that have a big discrepancy in the odds are obvious and don't need analytics to decide. Such as, 4th and 1, running off tackle vs a flea flicker. Obvious call.

Analytics should be used to show a relevant difference in the odds on a decision that may not be so obvious. Like, on 3rd down and less than 3, run off LT and we convert 45% vs a run off RT and we convert 15%. That's useful.

However, I know some take it too far. I know on ESPN's analytic tracker. They made a big deal about a call that was something like 2.3% off. Which of course is ridiculous.

[Showboat 142]

4 hours ago, AxeElf said:

If I say, "Do this and you'll win 90% of the time," and then you do it, and you lose, was I wrong?  Of course not; it was just one of the other 10% of the times.

Well, in football (and probably fantasy football, too) the only thing that matters is results.  Nobody cares if you made the "incorrect" decision every time - if you win, you were right.

[AxeElf 787]

43 minutes ago, Ray_T said:

probably too slow.  though it may be worth it to try if you are a last place team.  why not give it a go?

Like I said, I don't think they'd be quick enough for pass protection, where they have to react to a blitzer or stunt, so they couldn't be full time players.  But when you need to move the line of scrimmage 2 yards behind a six foot, 1000 lb moving wall, I don't see how they could fail.

[AxeElf 787]

2 minutes ago, Showboat said:

Well, in football (and probably fantasy football, too) the only thing that matters is results.  Nobody cares if you made the "incorrect" decision every time - if you win, you were right.

And yet people are incorrectly judging "analytics" by a small sample size of results.  Even if analytics gives you a 2% edge over a coin flip, you're still going to lose 48% of the time--even if the analytics are perfectly correct.

So people are judging on "results," but not considering all of them--like dismissing global warming because this summer was cooler than last year.

[nobody 2,031]

13 minutes ago, polecatt said:

They've failed any team that is foolish enough to put too much emphasis on analytics.

When it comes to making decisions and play calling, analytics is just one of many tools a coach should use.

Most decisions that have a big discrepancy in the odds are obvious and don't need analytics to decide. Such as, 4th and 1, running off tackle vs a flea flicker. Obvious call.

Analytics should be used to show a relevant difference in the odds on a decision that may not be so obvious. Like, on 3rd down and less than 3, run off LT and we convert 45% vs a run off RT and we convert 15%. That's useful.

However, I know some take it too far. I know on ESPN's analytic tracker. They made a big deal about a call that was something like 2.3% off. Which of course is ridiculous.

I'd also add that the analytics are always something like, "kicking the FG gives them a 53% chance to win and going for it gives them a 56% chance to win" 

If their models were that accurate they'd get rich gambling because if you had that kind of fidelity, you'd easily find inefficiencies in the betting markets to exploit.

[AxeElf 787]

1 minute ago, nobody said:

I'd also add that the analytics is always something like, "kicking the FG gives them a 53% chance to win and going for it gives them a 56% chance to win" 

If their models were that accurate they'd get rich gambling because if you had that kind of fidelity, you'd easily find inefficiencies in the betting markets to exploit.

Not necessarily.  The vig in sports betting is usually around 10% ("-110"), so even if you had a system accurate enough to distinguish a 3% edge, you'd still be 7% behind.

[nobody 2,031]

Man, I'm playing this exchange forward in my head, and no matter what it's going to be a disaster, but here goes.

First off, you only need to have a win rate of 52.5% to break even on a -110 bet, and your 10% vig citation is assuming you lose every bet.  Typically, when considering vig when betting a side, you assume 50% win rate in which case the vig is 5%. (obviously assumes both sides are -110 - yes I know those shift)

Second, you'd be betting moneylines anyway since the model would predict team X has a 60% chance of winning at halftime or something, and they'd just have to have a moneyline of something better than -150 to be profitable... again assuming the model is accurate (which it isn't).

But all that aside, it doesn't matter.  You would always find lines that are off by more than 5% if you had a model that was accurate to a few percent.  So many variables go into setting lines... the handle on one side or the moneyline, the public's perception, the public's betting, etc.  

 

[nobody 2,031]

In case someone is curious about the 52.5% here is the math...

.525 * 100 - .475*110 = 0.25

In other words, you'd make a quarter on average every time you bet $100

Here's your vig calc

.5*100 - .5*110 = -5

In other words, on average, you lose $5 every time you make a $100 bet -> 5% vig.

 

[Sean Mooney 1,184]

Problem also is that analytics does take into account the things like defenses you are playing and all of that....

Also- the Chargers yesterday I think Staley just trusted his offense could get 2 yards on 4th down. Herbert made a terrible read and terrible pass. 

[jrokh 567]

I don't think 'Analytics' are quite as simple as many are suggesting.

https://www.samford.edu/sports-analytics/fans/2020/The-Future-of-NFL-Data-Analytics

[nobody 2,031]

50 minutes ago, nobody said:

Man, I'm playing this exchange forward in my head, and no matter what it's going to be a disaster, but here goes.

First off, you only need to have a win rate of 52.5% to break even on a -110 bet, and your 10% vig citation is assuming you lose every bet.  Typically, when considering vig when betting a side, you assume 50% win rate in which case the vig is 5%. (obviously assumes both sides are -110 - yes I know those shift)

Second, you'd be betting moneylines anyway since the model would predict team X has a 60% chance of winning at halftime or something, and they'd just have to have a moneyline of something better than -150 to be profitable... again assuming the model is accurate (which it isn't).

But all that aside, it doesn't matter.  You would always find lines that are off by more than 5% if you had a model that was accurate to a few percent.  So many variables go into setting lines... the handle on one side or the moneyline, the public's perception, the public's betting, etc.  

 

Case in point, ESPN right now with Raiders up 14-0 about to kick off with 13:58 left in the 2nd, they have the Raiders win % at 69.6%.

My site has the live betting at Raiders -125.  That is a massively profitable bet if the Raiders are really 69.6% to win this game.  Of course, they're not really 69.6% because the ESPN prediction model is trash.

[AxeElf 787]

1 hour ago, nobody said:

Case in point, ESPN right now with Raiders up 14-0 about to kick off with 13:58 left in the 2nd, they have the Raiders win % at 69.6%.

My site has the live betting at Raiders -125.  That is a massively profitable bet if the Raiders are really 69.6% to win this game.  Of course, they're not really 69.6% because the ESPN prediction model is trash.

Given their 10 point lead on the road at the half, the Raiders probably ARE 69.6% likely to win the game, in the sense that 69.6% of all teams that have ever been in this situation win.

However, "Analytics" fails to raise their formula to the power of Mahomes.

[nobody 2,031]

5 minutes ago, AxeElf said:

Given their 10 point lead on the road at the half, the Raiders probably ARE 69.6% likely to win the game, in the sense that 69.6% of all teams that have ever been in this situation win.

However, "Analytics" fails to raise their formula to the power of Mahomes.

So it's trash.  Cool.

[AxeElf 787]

41 minutes ago, nobody said:

So it's trash.  Cool.

No, it's pinpoint accurate, in what it predicts.

But it can't predict Mahomes.

[nobody 2,031]

okay, "what it predicts" was supposed to be the chances of the Raiders winning the game.  So yeah, it's trash.

[AxeElf 787]

1 minute ago, nobody said:

okay, "what it predicts" was supposed to be the chances of the Raiders winning the game.  So yeah, it's trash.

No, it was exactly accurate.  This was just one of the other 30%--one of those times that they went in against Mahomes.

Which, as we all know, is right up there with going in against a Sicilian when death is on the line, or getting involved in a land war in Asia.

[nobody 2,031]

Sounds good.

[Ray_T 587]

1 hour ago, AxeElf said:

No, it was exactly accurate.  This was just one of the other 30%--one of those times that they went in against Mahomes.

Which, as we all know, is right up there with going in against a Sicilian when death is on the line, or getting involved in a land war in Asia.

you are correct.  Mahomes has a lot of 4th quarter comebacks.   a few of which nobody would have believed to be possible if they didnt see them actually happen.

I saw the prediction but I never thought their chances of winning were 69%.  I figured it was closer to 55% in reality.

[lod001 1,235]

Another loss last night. Stupid McDaniels uses analytics, goes for 2 down 1. Fails of course. Lombardi would have punched him in the face. They lose by 1.

[Sean Mooney 1,184]

27 minutes ago, lod001 said:

Another loss last night. Stupid McDaniels uses analytics, goes for 2 down 1. Fails of course. Lombardi would have punched him in the face. They lose by 1.

He was reading the flow of the game thinking his defense wasn't going to stop the Chiefs. It was the opposite of analytics for the most part. 

[lod001 1,235]

23 minutes ago, Sean Mooney said:

He was reading the flow of the game thinking his defense wasn't going to stop the Chiefs. It was the opposite of analytics for the most part. 

It was all analytics. You're making that BS up. If McDaniels said that BS, he's trying to deflect from his stupid use of analytics. Once again, Lombardi would have punched him in the face and kicked the XP.

Answer this question: DID IT WORK?

[Sean Mooney 1,184]

27 minutes ago, lod001 said:

It was all analytics. You're making that BS up. If McDaniels said that BS, he's trying to deflect from his stupid use of analytics. Once again, Lombardi would have punched him in the face and kicked the XP.

Answer this question: DID IT WORK?

"Did it work" has nothing to do with analytics.

Again- the Chiefs at that point had scored on 5 straight drives. The Raiders were scuffling. McDaniels went for the win. 

Who cares what an old hat coach would've done. 

[weepaws 2,450]

1 hour ago, lod001 said:

Another loss last night. Stupid McDaniels uses analytics, goes for 2 down 1. Fails of course. Lombardi would have punched him in the face. They lose by 1.

Agreed, it was a bad coaching decision. 

[lod001 1,235]

24 minutes ago, Sean Mooney said:

"Did it work" has nothing to do with analytics.

Again- the Chiefs at that point had scored on 5 straight drives. The Raiders were scuffling. McDaniels went for the win. 

Who cares what an old hat coach would've done. 

...and lost. Just like you in this debate.

[Sean Mooney 1,184]

9 minutes ago, lod001 said:

...and lost. Just like you in this debate.

You are the one who is lost.

Analytics does not guarantee a win. I wouldn't have necessarily done what McDaniels did but you argued in this thread "You need to be watching the game.... blah blah blah." Then the guy makes a move based on the flow of the game and you criticize as well. Seems like you just want to complain. 

[jrokh 567]

1 hour ago, lod001 said:

It was all analytics. You're making that BS up. If McDaniels said that BS, he's trying to deflect from his stupid use of analytics. Once again, Lombardi would have punched him in the face and kicked the XP.

Answer this question: DID IT WORK?

You either don’t know, or don’t care what analytics is and how they work…

[lod001 1,235]

18 minutes ago, jrokh said:

You either don’t know, or don’t care what analytics is and how they work…

Well, idiot. When they talk analytics on a current play, such as going for it on 4th and 1, I guess they don't either, huh?

[lod001 1,235]

45 minutes ago, Sean Mooney said:

You are the one who is lost.

Analytics does not guarantee a win. I wouldn't have necessarily done what McDaniels did but you argued in this thread "You need to be watching the game.... blah blah blah." Then the guy makes a move based on the flow of the game and you criticize as well. Seems like you just want to complain. 

He made a stupid move based on analytics and lost. Before the ignorance of analytics, a coach (easily 95 if not 99% of the time) would kick the XP and tie the game.

[jrokh 567]

13 minutes ago, lod001 said:

Well, idiot. When they talk analytics on a current play, such as going for it on 4th and 1, I guess they don't either, huh?

If 'they' or you talk Analytics on a singular play, and determine that it is effective or not based on the result of said play, then yes you are correct you both don't get it...

[Ray_T 587]

3 hours ago, lod001 said:

Another loss last night. Stupid McDaniels uses analytics, goes for 2 down 1. Fails of course. Lombardi would have punched him in the face. They lose by 1.

 

2 hours ago, Sean Mooney said:

He was reading the flow of the game thinking his defense wasn't going to stop the Chiefs. It was the opposite of analytics for the most part. 

 

that being said, they almost won despite the decision.   If Adams gets that second foot down, they may have been in good shape to get down for a game winning FG.

I know, coulda woulda shoulda. 

I'm not saying the decision was the right one.  

but they did nearly pull it out despite the stupid call

[lod001 1,235]

8 minutes ago, Ray_T said:

 

 

that being said, they almost won despite the decision.   If Adams gets that second foot down, they may have been in good shape to get down for a game winning FG.

I know, coulda woulda shoulda. 

I'm not saying the decision was the right one.  

but they did nearly pull it out despite the stupid call

I'll admit the calls were not the worst but they are losing games when they coach makes stupid calls. This is happening weekly. You know its bad when you can sit on your couch and scream at the TV, take the points on 4th & goal with the score 6-0 in the 1st Q. Hell, even Keenan Allen called out his own coach from his couch asking "WTF are we doing going for it on 4th down at our own 30". Coaches are idiots who are hooked on analytics. I'm not sure if its that it gives them someone else to blame for their ignorant calls but that's what it seems like. 'oh I was just being told to go for it by my analytics crew'. Well then why re we paying you big $ to do nothing?. An excuse of 'well the chiefs are scoring on us' is especially super lame when you scored the same number of points.