Analytics

[Nevbot]

14 hours ago, Goatface Killah said:

Bingo.

 

It makes you incredibly predictable. 

 

 

Not necessarily.  Analytics can actually spit out some pretty unpredictable results that go against gut instinct on occasion.  However relying on analytics primarily to run your offense in every situation as opposed to weighing the percentages on a situation basis (like on 4th down) is where Reich began to spiral.  If you rely on them too heavily, you end up with a square peg in a round hole type of situation with your personnel. 

For example, Reich's play design/analystics may have required the X receiver to run a drag route. Constantly.  The play design calls for the drag route to release into space.  The play is designed to open up that player on the sheet, but is Reich really utilizing Pittman to HIS abilities?  Likely not.  That drag should be Campbell, Hines, Dulin.  Pittman should be dogging the other teams CB1 all over the field and used for his big frame. 

This is just one example among many of why Reich playcalled his way out of town.  He never always had his teams ready early- even with Luck- but at least the O eventually got into sync and Reich was able to justify his value by falling back on Offensive production.  But we always saw the flaws- The misuse of players, Hines up the gut, going WAY too pass heavy with leads, going for some very questionable 4th downs, and getting overall way too cute with play calling.  "Frank is gonna Frank" became a punchline for a Coach that over time tried to by the smartest guy in every room. He always went away with what was working at the time because "the other team wont expect this wrinkle" and it got really old, really fast.
That is where I think we will get the polar opposite with Jeff Saturday.  He strikes me as the type of guy that if we gain 4 yards on an inside power run he will run that same play again and again until the other team shows they can stop it. That's what Vrabel does, and its literally the oldest adage in the trenches. Beat the man in front of you into submission instead of trying to outsmart him when you're already beating him.  

[DougDew]

12 hours ago, Defjamz26 said:

The whole analytics in football trend comes from baseball which is all about analytics and tendencies. There’s no play calling so to speak in baseball so it works. But in a sport where the result is determined by the play call, the players on both sides of the ball, the weather, the turf, the health of the player(s), audibles, etc… analytics doesn’t really work. Too many variables.

Yep.  The fundamental difference between baseball and football is that football is about 11 players doing their jobs each play.....working together for success...whereas in baseball success is based off of one of 9 players doing their job individually while the others just watch (until the ball comes at them).  A catch, a throw, a tag, a hit, a swing and a miss (either by fastball, curve ball, etc) is an individual performance.   You can collect data to measure how well a player fields, or how certain batters perform against certain pitches...and even other pitchers.  There is no way to make that direct individual performance because success is very much dependent upon 10 other players doing their jobs at the same time the one is doing theirs, and that's not even counting substitutions.  Its  a data collection nightmare.   

 

Basketball is becoming more like baseball.  Its usually about one or two great players making plays on their individual opponent.  Football is about how 11 players do their jobs together...and maybe different jobs...with different players from time to time. 

11 hours ago, Superman said:

What the data would tell him is 'if we go for it and don't convert, the probability that the Chiefs score on their next possession is X. Also, we have X percent probability of converting if we go for it.' 

 

And then it's up to him to weigh out those factors, along with other relevant information - is our QB playing well, is the OL blocking well, are we healthy at critical positions or are we missing a key player, is their defense healthy, is their best LB in the medical tent right now,' etc. And then he makes a decision based on all the information he has, some of it clear from the data, others based on his own present observations. 

Those two paragraphs are two separate worlds though. 

 

The probabilities that were calculated use data that is of unknown circumstances.  They are a compilation of different players and different teams under different circumstances.  Its not like baseball, where the player makes a play on the baseball coming at him....its the same situation done about 200 times in a season (just a guess), and either the player can catch a baseball well or not.  If not, you pull him in the late innings to protect a lead.  That data is good clean data. 200 repetitions of the exact same individual play...catch a ball out of the air.

 

The second paragraph you don't even need analytics to get you there.  If Bobby Wagner is in the injury tent and replaced with a backup, my offense has a better chance of completing a pass with MAC over the middle.  Computing a percentage is not even relevant because the percentage I calculate is not based upon MAC going up against Wagner 200 times...its based upon 31  other TEs going up against Wagner,  or worse. league percentages that are based upon  31 other TEs going up against 31 different MIKES.  Too many variables.

[Superman]

4 minutes ago, DougDew said:

Those two paragraphs are two separate worlds though. 

 

The probabilities that were calculated use data that is of unknown circumstances.  They are a compilation of different players and different teams under different circumstances.  Its not like baseball, where the player makes a play on the baseball coming at him....its the same situation done about 200 times in a season (just a guess), and either the player can catch a baseball well or not.  If not, you pull him in the late innings to protect a lead.  That data is good clean data. 200 repetitions of the exact same individual play...catch a ball out of the air.

 

The second paragraph you don't even need analytics to get you there.  If Bobby Wagner is in the injury tent and replaced with a backup, my offense has a better chance of completing a pass with MAC over the middle.  Computing a percentage is not even relevant because the percentage I calculate get is not based upon MAC going up against Wagner 200 times...its based upon 31  other players going up against Wagner,  or worse. league percentages that are based upon  31 other players going up against 31 different MIKES.  Too many variables.

 

The bolded is only true if you think you need to choose between data and coach's intuition, as if they can't coexist and even complement each other. Or like a coach can't use his own understanding to contextualize the data.

 

You might have 4th and 2 from the 40, and the card says go for it. But you know your best RB is hurt, and you ran a screen on first down for 3 yards, then a run play for zero yards but the defense was offsides, so it's 2nd and 2 instead of 2nd and 7, and then you threw incomplete, then ran for zero yards, and now it's 4th and 2 but you know you were gifted that favorable situation. Now it's up to the coach to factor in all the relevant info and make a reasonable choice. In this case, the present play of the offense outweighs other information, so you punt. 

 

This is the same calculus that allows coaches to go for it late in the game on 4th and 11, when the likelihood they'll convert is slim to none. They're not dismissing the data, they just know they have to try or else the game is over. 

 

To the rest of your comment, no, you're trying to outthink the room. You can receive a favorable probability, and put it into proper context for the situation at hand. Which is why even if my offense is on fire and the opposing defense cannot stop my best receiver, I'm still not going for it on 4th and 2 from my own 10 yard line, unless the game situation says I have to.

 

If Wagner is covering MAC, I'm probably not throwing to MAC on a crucial down. It may be true that TEs are averaging X yardage against LBs, but I know that's not a favorable matchup. This is not hard to reconcile. 

 

If any coach is just allowing a database to make decisions for them, that coach is misusing analytics. Just like if I was trying to use a Civic to tow a camper up a mountain, I'd be misusing the Civic. It doesn't mean the Civic has no value. It means you should take the keys away from me. Same for a coach who can't weigh out these factors, reconcile then situationally, and make solid decisions. You should take the keys from him. 

 

This isn't that hard. Analytics are a valuable resource, but knowing how to use them to gameplan and make decisions is something some coaches have a better grasp on than others. 

[DougDew]

12 minutes ago, Superman said:

 

The bolded is only true if you think you need to choose between data and coach's intuition, as if they can't coexist and even complement each other. Or like a coach can't use his own understanding to contextualize the data.

 

You might have 4th and 2 from the 40, and the card says go for it. But you know your best RB is hurt, and you ran a screen on first down for 3 yards, then a run play for zero yards but the defense was offsides, so it's 2nd and 2 instead of 2nd and 7, and then you threw incomplete, then ran for zero yards, and now it's 4th and 2 but you know you were gifted that favorable situation. Now it's up to the coach to factor in all the relevant info and make a reasonable choice. In this case, the present play of the offense outweighs other information, so you punt. 

 

This is the same calculus that allows coaches to go for it late in the game on 4th and 11, when the likelihood they'll convert is slim to none. They're not dismissing the data, they just know they have to try or else the game is over. 

 

To the rest of your comment, no, you're trying to outthink the room. You can receive a favorable probability, and put it into proper context for the situation at hand. Which is why even if my offense is on fire and the opposing defense cannot stop my best receiver, I'm still not going for it on 4th and 2 from my own 10 yard line, unless the game situation says I have to.

 

If Wagner is covering MAC, I'm probably not throwing to MAC on a crucial down. It may be true that TEs are averaging X yardage against LBs, but I know that's not a favorable matchup. This is not hard to reconcile. 

 

If any coach is just allowing a database to make decisions for them, that coach is misusing analytics. Just like if I was trying to use a Civic to tow a camper up a mountain, I'd be misusing the Civic. It doesn't mean the Civic has no value. It means you should take the keys away from me. Same for a coach who can't weigh out these factors, reconcile then situationally, and make solid decisions. You should take the keys from him. 

 

This isn't that hard. Analytics are a valuable resource, but knowing how to use them to gameplan and make decisions is something some coaches have a better grasp on than others. 

You've described actual scenarios where the intuition is superior to the analytics.  So why even introduce the analytics when you don't know if the circumstances used to calculate the percentages are even in the same ball park as the situation you are immediately faced with?  

 

Now, I agree with you that analytics can be helpful in forming a game plan during the week.  Partly because you have the time to actually look at the data and understand how reliable it is.  But many of us...and I think Irsay was suggesting (maybe not)..... that Reich was using analytics for in-game management....times when under stress.  Play calling, and then sticking to (still relying on) his analytics-based game plan and not adjusting if it isn't working.

[Nickster]

15 hours ago, DougDew said:

Yes, because Frank is using his analytics to gauge success rates (probably from data around the league or from the specific opponent) and fails to understand that there are idiosyncrasies that are part of the data he's looking at that need to be understood and managed.   He's robotic and predictable because he fails to incorporate enough game flow intuition into his decision making, IMO.

 

One of the things about Frank is I don't think he would not believe "what he saw".  I had a thread couple years back something like, Is Frank too Dogamatic?  This is the kind of thing I was talking about.

 

Analytics are great.  But they aren't God.   

[Nickster]

13 hours ago, MB-ColtsFan said:

I have been a casual watcher of football for over 55+ years.  I may be wrong on this but I started hearing/recognizing the phrase "they'll have to make some adjustments at halftime" only *recently*.  Honestly just asking; is that just coach speak or do they really make major adjustments only at halftime?  Is it analytics that make coaches less flexible during games?

Sometimes there are significant adjustments at HT.  Coaches confer and can change things drastically.  They have a chance though very brief to focus on a few things together without a game going on.  It's only a few minutes, but they are all together.

 

The players don't get much time with coaches, but the coaches definitely have intense discussions. 

 

 

[2006Coltsbestever]

Stats are fantastic but not everything. Gut and eye test + leadership are huge too. Mix all of them and you have a good chance to win if you have good talent. If someone just rely's stats for everything, you will fail in the end because of certain matchups you can take advantage of where you can tell who the better player is by eye test.

 

Stat guy = Drew Brees is better than Joe Montana

 

Common sense guy = no he isn't and anyone with half a football brain would not take Brees over Montana. 

[DougDew]

53 minutes ago, Nickster said:

 

One of the things about Frank is I don't think he would not believe "what he saw".  I had a thread couple years back something like, Is Frank too Dogamatic?  This is the kind of thing I was talking about.

 

Analytics are great.  But they aren't God.   

.........Or, when he is under stress or unsure of direction, he refers to the book.  As a process of habit.

 

  

[Goatface Killah]

1 hour ago, Nevbot said:

Not necessarily.  Analytics can actually spit out some pretty unpredictable results that go against gut instinct on occasion.  However relying on analytics primarily to run your offense in every situation as opposed to weighing the percentages on a situation basis (like on 4th down) is where Reich began to spiral.  If you rely on them too heavily, you end up with a square peg in a round hole type of situation with your personnel. 

For example, Reich's play design/analystics may have required the X receiver to run a drag route. Constantly.  The play design calls for the drag route to release into space.  The play is designed to open up that player on the sheet, but is Reich really utilizing Pittman to HIS abilities?  Likely not.  That drag should be Campbell, Hines, Dulin.  Pittman should be dogging the other teams CB1 all over the field and used for his big frame. 

This is just one example among many of why Reich playcalled his way out of town.  He never always had his teams ready early- even with Luck- but at least the O eventually got into sync and Reich was able to justify his value by falling back on Offensive production.  But we always saw the flaws- The misuse of players, Hines up the gut, going WAY too pass heavy with leads, going for some very questionable 4th downs, and getting overall way too cute with play calling.  "Frank is gonna Frank" became a punchline for a Coach that over time tried to by the smartest guy in every room. He always went away with what was working at the time because "the other team wont expect this wrinkle" and it got really old, really fast.
That is where I think we will get the polar opposite with Jeff Saturday.  He strikes me as the type of guy that if we gain 4 yards on an inside power run he will run that same play again and again until the other team shows they can stop it. That's what Vrabel does, and its literally the oldest adage in the trenches. Beat the man in front of you into submission instead of trying to outsmart him when you're already beating him.  

My point was the other team can also use  analytics to predict what you are gonna do in very key situations, which makes you very predictable. And the Colts offense has been pretty predictable, situationally. I dont think thats a coincidence. 

 

Knowing how an opponent makes their decisions is almost as good as knowing what they are gonna do.

 

I think Reich made this fatal mistake.

 

For example.......Teams knew that in 3rd or 4th and inches, he is always gonna run because the analytics suggest the odds heavily favor running vs passing in that situation. In reality, he almost never threw a pass in those situations, and he went for them constantly. And the results for us didnt mirror the analytics, and I think its because teams knew with almost 100% certainty what he was gonna do, situationally.

 

So what are the stats of succeeding running up the middle when the other teams knows youre gonna run up the middle? Probably not very good. 

 

People have given me an anti analytics label but that isnt really true. I just dont feel like most people know what the numbers actually mean and how best to use them, and I think Frank is a prime example of that. I dont mind them as a tool for gameplanning, primarily using them for opponents and to avoid trends. Not making decisions situationally on the field in real time. 

[chad72]

11 minutes ago, Goatface Killah said:

 

People have given me an anti analytics label but that isnt really true. I just dont feel like most people know what the numbers actually mean and how best to use them, and I think Frank is a prime example of that. I dont mind them as a tool for game planning, primarily using them for opponents and to avoid trends. Not making decisions situationally on the field in real time. 

 

Not me. Intuition vs Analytics, the lines are blurred. We would get on Arians when he ran 10 times in a row with no rhyme or reason while he was trailing or the score was tied with Andrew Luck for that 1 year he was with us. Coaches sometimes go with what is purely working if they have established that their guys are winning at the line of scrimmage. That doesn't mean analytics isn't useful but it sure means that while it is useful to game plan for the most part, adhering too much to it religiously, does make you predictable and lose a pulse of the team or the flow of the game. 

[jonjon]

2 hours ago, MB-ColtsFan said:

I have been a casual watcher of football for over 55+ years.  I may be wrong on this but I started hearing/recognizing the phrase "they'll have to make some adjustments at halftime" only *recently*.  Honestly just asking; is that just coach speak or do they really make major adjustments only at halftime?  Is it analytics that make coaches less flexible during games?

Halftime adjustments have been a common thing for a long time. It's not "analytics." How much adjustment of course varies.  

[CoachLite]

Being able to spell 'analytics' does not make you qualified to understand analytics. Most people that use analytics effectively are standard deviants.

[CoachLite]

Maybe the Colts FO should read this: https://mitsloan.mit.edu/ideas-made-to-matter/decisions-not-data-should-drive-analytics-programs

[sb41champs]

Analytics are used by various teams - in various ways.

 

I don't know that Frank Reich used them more - or - less than any other coach or team in the NFL.

 

Jim Irsay may not know either - as he isn't a part of the gameday coaching staff. 

 

He has a long and undistinguished history of "talking out his a#s!!

[crazycolt1]

I don't know how much analytics Frank used but his play calling became way to predictable. That problem didn't just start this season either.  

[Superman]

On 11/9/2022 at 8:05 AM, DougDew said:

You've described actual scenarios where the intuition is superior to the analytics.  So why even introduce the analytics when you don't know if the circumstances used to calculate the percentages are even in the same ball park as the situation you are immediately faced with?  

 

This is really not that difficult. 

 

If you're landscaping your yard, you might receive recommendations for plants that grow well in your region, but the unique characteristics of your yard make it an outlier, so you know those plants won't be right for you. There's nothing wrong with the information, it just doesn't apply to your circumstances. It's up to you to understand and apply the information properly.

 

The data says TEs have a good success rate vs LBs, but you know your TEs aren't strong producers, and your opponent's LBs are great in coverage. The data isn't bad. It's up to you to determine how well it applies to your circumstances. Different subsets of the data may reveal different potential matchups.

 

Gameplanning is fundamentally a determination of what the teams do well/poorly, and how to exploit those factors to your advantage. But that determination is based on performance, which is judged by the numbers -- the data itself. 'Our team is good at X' is only understood because you've analyzed the data. Some conclusions are more obvious than others, maybe so much so that you just call them "intuition," but for the most part those conclusions are based on analysis of data. 

 

Quote

Now, I agree with you that analytics can be helpful in forming a game plan during the week.  Partly because you have the time to actually look at the data and understand how reliable it is.  But many of us...and I think Irsay was suggesting (maybe not)..... that Reich was using analytics for in-game management....times when under stress.  Play calling, and then sticking to (still relying on) his analytics-based game plan and not adjusting if it isn't working.

 

I don't really know what Irsay was suggesting, it was a throwaway line. Sure, he probably thinks coaches rely too much on data and not enough on old school decision making principles, like 'take the points.' 

 

I don't think Reich did a good job of formulating game plans based on data, nor did he adjust well during games. He had a good run of opening drives for a while, and maybe that was helped by the data, but I wonder if it was just luck. He subbed a ton, using different formations and packages, not anything that appeared to be specifically attacking his opponent. I could be completely wrong on this.

 

Where I think Reich was mostly convinced was game situations -- 4th downs and 2 point conversions, specifically. The value of going for it has to be judged on a macro level, because you won't convert them all. Value is gained on a cumulative basis, so the more you go for it, the bigger the payoff should be. And even there, Reich seemed to realize that sometimes it's best to play it safe. This season, he started with his typical preference to go for it, and then realized that the OL was bad, the offense wasn't playing well, JT wasn't able to get loose, etc. By the end, he understood that we weren't a "go for it" offense, which is why he was more conservative than usual at the end of the Washington game. But that understanding isn't based on intuition; it's based on the fact that the offense converts 36% on third down, and 22% on fourth down, drastically lower than any other season since Reich took over, especially on 4th down. The data was there, and he understood it and applied it correctly in that case. Unfortunately, that shift made it seem like he lost his nerve, or lost confidence in his team, which can be a potential downside to being a more aggressive situational decision maker.

[coltsva]

On 11/9/2022 at 10:54 AM, Superman said:

If any coach is just allowing a database to make decisions for them, that coach is misusing analytics. Just like if I was trying to use a Civic to tow a camper up a mountain, I'd be misusing the Civic. It doesn't mean the Civic has no value. It means you should take the keys away from me. Same for a coach who can't weigh out these factors, reconcile then situationally, and make solid decisions. You should take the keys from him. 

 

 

Okay, Superman, give me the keys before you get to the mountain. 

[Dingus McGirt]

So.  If an opponent knows a team relies heavily, nearly always, on Analytics…could they not use those same Analytics to predict with near certainty what an opposing play caller will do?

[Superman]

36 minutes ago, Dingus McGirt said:

So.  If an opponent knows a team relies heavily, nearly always, on Analytics…could they not use those same Analytics to predict with near certainty what an opposing play caller will do?

 

If this held up in practice, no team would ever convert on third and long when a pass play is a near certainty. Tom Brady's QB sneak would never work. Etc. 

 

You can be certain that a team is going to run, pass, blitz, play zone, whatever, but you still have to account for play variance, execution, and other factors. Analytics is not meant to be a magic spell that allows you to succeed at everything you do, it's just a tool that decision makers can potentially use to evaluate the circumstances and maybe gain an advantage. That can be especially valuable when there's a discrepancy between the data and conventional thinking.

[WHAM-Bam ThankYou Sam]

On 9/19/2022 at 5:44 PM, SOMDColtsfan said:

Up for discussion, forgive if already been hashed out. Do the Colts use this data during games more than other teams? It seems to be a reactive mindset/scheme rather than a dictate. This is what we're going to do, and you have to stop it. Reich stated the direction of the game dictated not using Hines as much as they wanted vs the Jags. Pittman and Pierce being out may have factored into this, but still. That's not forcing your will or gameplan on another team. There were times last year certain things were working well, and we saw the play call's go in a totally different direction. Great coaches have a feel for game flow. While pointing fingers..Are analytics to blame?

Lol well if you ACTUALLY followed the analytics then you would see that following the analytics forces one to take in to account of probability and risk taking is analytically feasible and when I mumbled gobbaly goop and big words along side with stats it makes my careless risk taking seem more calculated. Making me look less of a dumb ——

[Mighty Blue]

Introducing the NEW and Exciting Innovation in ANALYTICS!

Now you can have your own Analytic Analyzer in your own home!

*Its light weight!

*Easy to use!

*Be the envy during game day!

*Plug it in…batteries not included.

*Turn it on!
*Just fill out the questionnaire, place it in the…INPUT? That’s right, you’ve got it !

*Presto!

Your questionnaire is answered via the…OUTPUT? That’s right…You are a winner!

If you are believer in Sports Analytics, this Analyzer makes it easier for you.

Here is the best part, it’s FREE and if you are the first 100 geeks to order…you can be a guess star with Joooooe Thomas!

 

Now there is an analyzer that analyzes Analytics analysis

But wait, if you order now,

you will get a cup holder and popcorn popper.

 

 Really?