49ers Analytic Driven OT Possession Decision And What It Means For The Dolphins

[rafael]

I think like the 49ers, and Detroit, Miami is too reliant on analytics for some in game decisions, and they do not take into account other variables that alter the analytics.

49ers decision to take the ball first was driven by analytics. Rationale was if both teams have the ball, and the game is still tied, THEN the game goes into sudden death. IF that happens, 49ers then have the ball first, and any score wins the game.

OK – big IF. I do NOT care what the computer says when playing Mahomes, and the defense is entirely worn out, AND you are giving that guy (not the average QB), 4 downs on every possession, if you have scored first. That is the downside of taking the ball first. If you score, you give the other team 4 downs to match. You are giving them an advantage. BAD decision in this case.

This strategy reminds me of people in the knockout pool that instead of trying to just win the game in week 1 with a high probability pick, they try to look 10 games out, and pick a marginal team to win in the first week, so they have saved the really good team in win in week 10. That person NEVER makes it to week 10. It does not work. You do what is the highest probability thing to do for the first two possessions that are more predictable, then take your chances with what happens if the game keeps going. You do NOT bet on a series of events you may never come to realize, which is exactly what happened in this game. The 49ers never got the ball back to go to sudden death, and they didn’t get the ball back for a reason – the guy they were up against and the fact their defense was exhausted.

In a situation like this, and given the opponent you are playing against, and given your defense is worn out, and since IF you go ahead you are giving Mahomes 4 downs, I could give a crap what the analytics department says. THEY DO NOT TAKE INTO ACCOUNT WHO THE OPPONENT IS AND ALL THE OTHER VARIABLES LIKE THE DEFENSE IS WORN OUT. The analytics are based on the “average” situation, and this was decidedly NOT the average situation or the average QB you are going against!

Both teams have the opportunity to get the ball at least once during overtime unless the team kicking off received a safety on the team’s initial possession. If one team has more points than its opponent after the two possessions, it is declared the winner.

If the game is tied after the two possessions or neither team scored on their first possession, then the game goes into sudden death.

Coach Kyle Shanahan suggested he had a strategy for overtime.

"If both teams matched and scored, we wanted to be the ones with the chance to go win it," he said. "We got that field goal, so we were hoping to hold them to at least a field goal. If we did, we felt it was in our hands after."hg

My thought at the end of the game was that the 49er D was tired and that it needed a break to have a chance. Generally, I prefer to have the ball second in OT, but in this case I felt SF was damned either way.

 

[jazz015]

Analytics are a good tool for the decision maker. However, excessive reliance on them can actually expose poor decision making. At the end of the day there will always be a need for good judgement by the play caller. Risk is also dependent on the physical condition of the players which can be impacted by injury and just getting tired. There will be times to be aggressive and there will be times where it is prudent to be conservative and a coach with good judgement understands the difference. An example of a coach who showed excessive reliance on the analytics was Brandon Staley at the Los Angeles Chargers. However, he got it wrong too often and repeatedly blamed the analytics. He is now a former coach.

My issue with analytics is that it does not quantify momentum or the flow of a game. There are just so many big moments that come down to knowing where the teams is and if they can accomplish the goal.

 

[JamesWsenior]

My issue with analytics is that it does not quantify momentum or the flow of a game. There are just so many big moments that come down to knowing where the teams is and if they can accomplish the goal.

Analytics seems cool in theory but as you say it struggles with variables in live play.

Example.... The data doesn't account for losing your star player mid game. It doesn't take into account field condition or weather or "mo"....see Dan Campbell when he should have kicked the FG to go back up 17. I am sure his data says go for it and we win now...it probably didn't say go for it and fail and you lose the momentum and maybe the game.

It does however provide cover... usually following a loss.

 

[Awsi Dooger]

Once you embrace analytics, or other data-driven approaches, it becomes more difficult to go against them for any given decision.

Exactly. You have to keep doing it. The biggest buffoons are the ones who subjectively decide analytics fit that situation but not this one. That is destined to zig zag fail more dependably than if analytics were used every time or subjectivity every time.

In betting I use systems, to remove the subjectivity and the second guessing. It works remarkably well. I often gave out those systems to friends and acquaintances. Invariably one of them would come up to me and smugly say something like, "Too bad about the Bears last week. I know your system had them. I use that system. But not in that game. I went the other way."

It was impossible not to laugh. That guy has no chance. And likewise with the ones who are criticizing Dan Campbell. Let's just say I wasn't surprised at the names. Loud and wrong.

The best point made in this thread was that analytics can vary by team and that the margins are small. We're talking a few percent, not 30%. From early days on this site I emphasized you have to steal 2-3% from the norm.

Analytics are in their infancy. Every time a coach is looking at a play calling sheet, analytics could do it better. But only if applied every time. If a team used analytics in that role the game will often be over already and you don't need to be holding a play chart.

 

[phinsforlife]

Exactly. You have to keep doing it. The biggest buffoons are the ones who subjectively decide analytics fit that situation but not this one. That is destined to zig zag fail more dependably than if analytics were used every time or subjectivity every time.

In betting I use systems, to remove the subjectivity and the second guessing. It works remarkably well. I often gave out those systems to friends and acquaintances. Invariably one of them would come up to me and smugly say something like, "Too bad about the Bears last week. I know your system had them. I use that system. But not in that game. I went the other way."

It was impossible not to laugh. That guy has no chance. And likewise with the ones who are criticizing Dan Campbell. Let's just say I wasn't surprised at the names. Loud and wrong.

The best point made in this thread was that analytics can vary by team and that the margins are small. We're talking a few percent, not 30%. From early days on this site I emphasized you have to steal 2-3% from the norm.

Analytics are in their infancy. Every time a coach is looking at a play calling sheet, analytics could do it better. But only if applied every time. If a team used analytics in that role the game will often be over already and you don't need to be holding a play chart.

the biggest buffoons are the ones who think analytics are right all of the time. they have a long way to go. far from perfect. they don't account for a lot of things. they are a tool in the toolkit, and a useful one, but just a tool.

 

[Fan Since the 70s]

You have to remember that analytics tell you what is PROBABLE, not what is certain. If a strategy is effective 90 of every 100 times, the strategy is the correct choice, despite that the actual outcome may be one of the 10 of every 100 that's highly improbable.

 

[JamesWsenior]

the biggest buffoons are the ones who think analytics are right all of the time. they have a long way to go. far from perfect. they don't account for a lot of things. they are a tool in the toolkit, and a useful one, but just a tool.

Holy **** we agree on something! Well it is Valentine's Day morning here. Got a date for later?

 

[lurking]

I think like the 49ers, and Detroit, Miami is too reliant on analytics for some in game decisions, and they do not take into account other variables that alter the analytics.

49ers decision to take the ball first was driven by analytics. Rationale was if both teams have the ball, and the game is still tied, THEN the game goes into sudden death. IF that happens, 49ers then have the ball first, and any score wins the game.

OK – big IF. I do NOT care what the computer says when playing Mahomes, and the defense is entirely worn out, AND you are giving that guy (not the average QB), 4 downs on every possession, if you have scored first. That is the downside of taking the ball first. If you score, you give the other team 4 downs to match. You are giving them an advantage. BAD decision in this case.

This strategy reminds me of people in the knockout pool that instead of trying to just win the game in week 1 with a high probability pick, they try to look 10 games out, and pick a marginal team to win in the first week, so they have saved the really good team in win in week 10. That person NEVER makes it to week 10. It does not work. You do what is the highest probability thing to do for the first two possessions that are more predictable, then take your chances with what happens if the game keeps going. You do NOT bet on a series of events you may never come to realize, which is exactly what happened in this game. The 49ers never got the ball back to go to sudden death, and they didn’t get the ball back for a reason – the guy they were up against and the fact their defense was exhausted.

In a situation like this, and given the opponent you are playing against, and given your defense is worn out, and since IF you go ahead you are giving Mahomes 4 downs, I could give a crap what the analytics department says. THEY DO NOT TAKE INTO ACCOUNT WHO THE OPPONENT IS AND ALL THE OTHER VARIABLES LIKE THE DEFENSE IS WORN OUT. The analytics are based on the “average” situation, and this was decidedly NOT the average situation or the average QB you are going against!

Both teams have the opportunity to get the ball at least once during overtime unless the team kicking off received a safety on the team’s initial possession. If one team has more points than its opponent after the two possessions, it is declared the winner.

If the game is tied after the two possessions or neither team scored on their first possession, then the game goes into sudden death.

Coach Kyle Shanahan suggested he had a strategy for overtime.

"If both teams matched and scored, we wanted to be the ones with the chance to go win it," he said. "We got that field goal, so we were hoping to hold them to at least a field goal. If we did, we felt it was in our hands after."hg

Ummmm the Chiefs won the coin toss and deferred its choice to the second half..

 

[phinsforlife]

You have to remember that analytics tell you what is PROBABLE, not what is certain. If a strategy is effective 90 of every 100 times, the strategy is the correct choice, despite that the actual outcome may be one of the 10 of every 100 that's highly improbable.

that is correct and incorrect. in theory that is what the analytics tell you, but they are far from perfect and far from right about what is probable. they are not perfect. there are infinite probabilities, especially the earlier in the game. analytics have a long way to go. computer just got to the point where it could reliably beat humans at chess, issue is the amount of probabilities in chess. even harder in the game of go, where the computers still lose to people. football is even more complex and has more probabilities than those games

 

[Fred Bear]

It means nothing for us