Coaches are too conservative (Not Riverboat!)

[Happy Panther]

http://regressing.deadspin.com/charts-just-how-wimpy-are-nfl-coaches-on-fourth-down-c-1475806291

 

 

The Bot uses game data going back to 2000 to tell you whether you should punt, kick a field goal, or go for it on fourth down, based on your position on the field and the yards to first down. For most of the game (the first three quarters or so), its suggestions are based on maximizing expected points, but with around 10 minutes remaining in the fourth, it switches over to maximize winning percentage. The charts themselves are interactive and the article has a great explanation of Burke's "expected points" formula, so you should go check it out.

There's no mention of Riverboat Ron, though, whose Panthers are eight-for-eight on fourth down during their eight game win streak, with six of the seven drives resulting in touchdowns. 

 

 

[dos poptarts]

I remember looking at this in the middle of the season. I think 4th down bot is too agressive sometimes. But overall I agree. Not sure I understand how 4th and 3 and the 8 yard line is a go for it...but 4th and 3 at the 15 yard line is punt. (I guess the oppo starting position after the punt is taken into consideration)

 

 

[d-dave]

Didn't Ron start to be more aggressive than the bot?  This was brought up earlier this season, and I thought that Ron was more aggressive in several instances versus the bot.

 

Either way, I hope we have the ability to be as aggressive for 2014.

[sml1950]

According to a sirius interview before the last game of the season, RR had gone on 4th down 13 times....and the league average was 14. So Riverboat Ron was only a gambler vis a vis his previous decision-making, not overall coaches.

[Happy Panther]

I remember looking at this in the middle of the season. I think 4th down bot is too agressive sometimes. But overall I agree. Not sure I understand how 4th and 3 and the 8 yard line is a go for it...but 4th and 3 at the 15 yard line is punt. (I guess the oppo starting position after the punt is taken into consideration)

 

The data is empirical meaning the bot looks at when teams actually went for it on 4th and 3 from the 8. Which is probably like once since 2000. 

 

Basically there is little or no data in some of those scenarios. Not sure how the bot is getting to some of the results.

[t96]

This is ridiculous. You should absolutely never go for it on 4th down within your own 20, unless you have to at the end of the game (like Smitty's conversion against the Dolphins).

[updownsigma]

The data is empirical meaning the bot looks at when teams actually went for it on 4th and 3 from the 8. Which is probably like once since 2000. 

 

Basically there is little or no data in some of those scenarios. Not sure how the bot is getting to some of the results.

 

It might use combined data from similar situations, like I've seen Barnwell on Grantland say teams overall have roughly a 50% chance of gaining at least 2 yards, regardless of field position. So maybe its the expected point value of converting a first down from the 8 and the probability of getting 3 yards?

 

Not sure exactly how that would work, but it would have to be better look at the few times a team has gone for it from 4th and 3 from the 8, not only because of the small sample size that you pointed out, but also I'm sure whatever team did go for it on 4th and 3 from the 8 did it because they were down with little time left, which also biases things. For example, they're more likely to call a run than a pass, if the conversion fails the team that gets the ball would probably just kneel instead of kicking a FG, meaning the expected points aren't as "bad" as it might be going for it in a normal game.