 |
View Old Forum Thread
Old Forum Index » Football Beta Testing » Football Beta Test Discussion » 4th down Decision Glitch?
RichNYC1
Not clear as to why we didnt kick FG´s in either of the situations. I went through my Advanced Options and dont see where any rule dictates "go-for-it" in these instances.
1st Quarter
04:54. NYA 4/7 IND 13 4th and 7 for the Trojans, they huddle and line up in a Shotgun; the Fighting Irish line up in a 4-2-5 Nickel.
Don Litkus takes the snap and drops back into the pocket.
Litkus passes to Kris Krenk at the IND 3.
DEBUG: Calculated completion chance: 78%
The ball is dropped at the IND 3 by Kris Krenk.
Pass by Don Litkus intended for Kris Krenk is incomplete.
2nd Quarter
01:30. NYA 4/4 IND 14 4th and 4 for the Trojans, they huddle and line up in a Pro Set; the Fighting Irish line up in a 4-3.
Don Litkus takes the snap, hands off to Carl Glenn.
Carl Glenn runs up the middle for a short gain.
DEBUG: Crediting Doug Ward with the tackle
[url=http://footballbeta.simdynasty.com/boxscore.jsp?boxscoreid=61310&thid=5911&qtr=2]Game Link NYT-IFI[/url]
4th down Decision Glitch?
August 05, 2014 at 02:25PM View BBCode
From Beta P.Not clear as to why we didnt kick FG´s in either of the situations. I went through my Advanced Options and dont see where any rule dictates "go-for-it" in these instances.
1st Quarter
04:54. NYA 4/7 IND 13 4th and 7 for the Trojans, they huddle and line up in a Shotgun; the Fighting Irish line up in a 4-2-5 Nickel.
Don Litkus takes the snap and drops back into the pocket.
Litkus passes to Kris Krenk at the IND 3.
DEBUG: Calculated completion chance: 78%
The ball is dropped at the IND 3 by Kris Krenk.
Pass by Don Litkus intended for Kris Krenk is incomplete.
2nd Quarter
01:30. NYA 4/4 IND 14 4th and 4 for the Trojans, they huddle and line up in a Pro Set; the Fighting Irish line up in a 4-3.
Don Litkus takes the snap, hands off to Carl Glenn.
Carl Glenn runs up the middle for a short gain.
DEBUG: Crediting Doug Ward with the tackle
[url=http://footballbeta.simdynasty.com/boxscore.jsp?boxscoreid=61310&thid=5911&qtr=2]Game Link NYT-IFI[/url]
blakjakshalak
Must have been the %chance of making the FG was too low.
[Edited on 8-5-2014 by blakjakshalak]
August 05, 2014 at 03:01PM View BBCode
It was really windy...23mph! Hurricane Ricardo:oMust have been the %chance of making the FG was too low.
[Edited on 8-5-2014 by blakjakshalak]
RichNYC1
August 05, 2014 at 04:33PM View BBCode
I saw that, Jack. But these were chip shots. I bet there is something I am not seeing in my Advanced Options, but I went through them twice. If it was the wind, which was NE, I think the effect % may be too strong. Those FG´s should still be higher % than going for 4th and 7 or 4th and 4.
blakjakshalak
[Edited on 8-5-2014 by blakjakshalak]
August 05, 2014 at 05:34PM View BBCode
That brings up a good point. If I'm not mistaken, the % chance setting for FGs is simply a threshold. Right now the rule is something like, 'go for the FG if there is at least X% chance of making it.' If the FG isn't attempted another rule says to punt unless it's within X yards of the goal. This does set up a scenario where bad weather has a large affect on the chance of making the FG so the % chance threshold is met even though you are well within the distance you would normally try the FG. But at the same time, the chance of making a 4th and long is even lower for the same reason (bad weather knocks down an already low % of completing a long pass on 4th down). You're almost certainly better off attempting the FG or even punting...in the NFL, you will sometimes see a team take an intentional delay of game penalty so they can back up and give the punter a better angle to kick the ball out of bounds. Maybe there's a way to evaluate the chance of making a FG against the chance of going for it or perhaps a sliding scale for weather effects.[Edited on 8-5-2014 by blakjakshalak]
Admin
Second kick: Estimated chance is 43 percent
Maybe an easy solution is to have the kicker "underestimate" the wind when deciding the percentage? i.e. calculate the chance of success based on half the actual wind speed? This wouldn't affect the actual kick of course.
August 06, 2014 at 05:11PM View BBCode
First kick: Estimated chance is 44 percentSecond kick: Estimated chance is 43 percent
Maybe an easy solution is to have the kicker "underestimate" the wind when deciding the percentage? i.e. calculate the chance of success based on half the actual wind speed? This wouldn't affect the actual kick of course.
Admin
Now, there are two factors here: Distance and accuracy. Distance absolutely should have an effect on success rate as a function of the length of the kick; fortunately, this is a relatively straightforward physics calculation. My reference for that is the excellent "Football Physics: The Science of the Game" by Dr. Timothy Gay. His work informs a number of decisions in the game. This is a "chip shot" so distance isn't an issue.
The wind effect is broken into vectors. For simplicity there are only eight wind directions, so to simplify things diagonal winds of 1.0 units are broken into vectors of .7 units. So a 23 mph northeast wind breaks down to a 16 mph crosswind vector.
The paper I am currently reading is at http://www.sloansportsconference.com/wp-content/uploads/2013/Going%20for%20Three%20Predicting%20the%20Likelihood%20of%20Field%20Goal%20Success%20with%20Logistic%20Regression.pdf . What is interesting about that paper is it takes situational pressure into account, which the sim currently does not. It acknowledges, however, that the more you break down the kick, the more sample size errors creep in. Thus, the paper simplifies wind as "greater than or equal to 10mph" vs "less than 10mph" and doesn't seem to account for crosswinds vs headwinds or tailwinds. This may be because theoretically headwinds and tailwinds should cancel each other out; however, it still doesn't take into account vectors. Of course in real life winds aren't limited to 8 directions, so the crosswind vector will vary between 1.0 and 0.7, or a cutoff of 7 to 10 MPH in this model. For simplicity I will treat the cutoff as 9 mph rather than 10. Also, a huge potential flaw in the study is that winds are based on conditions at kickoff not at the time of the actual kick.
Cold was a factor in this kick as well; the study divides between above or below 50%, and this game was below 50. In addition, the sim takes wind chill into account, and uses the temperature produced by wind chill, although it was a clear day so the wind chill effect is halved.
I'm going to have to read this paper a few times and see if I can draw some conclusions relevant to the sim from it.
Chris
August 06, 2014 at 06:13PM View BBCode
My instinct is to defend my code and say "Wind should make that big of a difference" but I am currently examining a study that disagrees with that conclusion.Now, there are two factors here: Distance and accuracy. Distance absolutely should have an effect on success rate as a function of the length of the kick; fortunately, this is a relatively straightforward physics calculation. My reference for that is the excellent "Football Physics: The Science of the Game" by Dr. Timothy Gay. His work informs a number of decisions in the game. This is a "chip shot" so distance isn't an issue.
The wind effect is broken into vectors. For simplicity there are only eight wind directions, so to simplify things diagonal winds of 1.0 units are broken into vectors of .7 units. So a 23 mph northeast wind breaks down to a 16 mph crosswind vector.
The paper I am currently reading is at http://www.sloansportsconference.com/wp-content/uploads/2013/Going%20for%20Three%20Predicting%20the%20Likelihood%20of%20Field%20Goal%20Success%20with%20Logistic%20Regression.pdf . What is interesting about that paper is it takes situational pressure into account, which the sim currently does not. It acknowledges, however, that the more you break down the kick, the more sample size errors creep in. Thus, the paper simplifies wind as "greater than or equal to 10mph" vs "less than 10mph" and doesn't seem to account for crosswinds vs headwinds or tailwinds. This may be because theoretically headwinds and tailwinds should cancel each other out; however, it still doesn't take into account vectors. Of course in real life winds aren't limited to 8 directions, so the crosswind vector will vary between 1.0 and 0.7, or a cutoff of 7 to 10 MPH in this model. For simplicity I will treat the cutoff as 9 mph rather than 10. Also, a huge potential flaw in the study is that winds are based on conditions at kickoff not at the time of the actual kick.
Cold was a factor in this kick as well; the study divides between above or below 50%, and this game was below 50. In addition, the sim takes wind chill into account, and uses the temperature produced by wind chill, although it was a clear day so the wind chill effect is halved.
I'm going to have to read this paper a few times and see if I can draw some conclusions relevant to the sim from it.
Chris
Pages: 1