# While studying for a Game Theory exam I found an example that contradicts our offense

Discussion in 'Swamp Gas' started by GatorEcon, Oct 16, 2013.

1. Offline

### GatorEconMember

Joined:
Sep 22, 2012
Messages:
60
9
Trophy Points:
8
Location:
Gainesville
+9 / 0 / -0
2. Online

### JonasWell-Known Member

Joined:
Aug 6, 2011
Messages:
3,569
47
Trophy Points:
48
+155 / 6 / -11
Really one of the fundamental aspects of Boise St's offense is calculated aggression that is justified through game theory.
3. Offline

### socalg8rMember

Joined:
Aug 8, 2007
Messages:
610
3
Trophy Points:
18
+4 / 0 / -0
Fantastic post.
4. Offline

### gatorr4lifeWell-Known Member

Joined:
Apr 3, 2007
Messages:
22,443
206
Trophy Points:
63
+619 / 44 / -13
Very interesting. Thank you for sharing.
5. Offline

### gatorr4lifeWell-Known Member

Joined:
Apr 3, 2007
Messages:
22,443
206
Trophy Points:
63
+619 / 44 / -13
After further inspection, I've come to the conclusion you're one smart sob.
6. Offline

### WeWinWeEatMember

Joined:
Apr 11, 2007
Messages:
234
0
Trophy Points:
16
+2 / 0 / -0
Your theory suggests every player is capable of doing his job. Have you adjusted for our OL? Or picked up a football ever? Just ribbing ya. Fun post. Muschamp should play more video games. It might help.
7. Offline

### gatorbogeyActive Member

Joined:
Apr 3, 2007
Messages:
33,135
19
Trophy Points:
38
Location:
s. e. florida
+23 / 0 / -0
I would say that you also need to factor in down and distance. we definitely need to pass more on first down. after that, this theory might actually support our offense - that is, the defense might be guessing pass, but we run instead.
8. Offline

### OaktownGatorWell-Known Member

Joined:
Apr 3, 2007
Messages:
23,645
1,932
Trophy Points:
113
+3,420 / 81 / -24
Assuming you were going to use this game as a basis for mix of play calling, you'd have to adjust the variables to account for our issues pass blocking.

So the pass numbers would be skewed to a higher negative number when the defense guesses pass, than the positive number when the defense guesses run... for their simplistic example, it might be (-10,5) instead of (-5,5).

Which in turn would lead you to pass less often, but hopefully in situations the defense expects you to run.
9. Offline

### G8R92Well-Known Member

Joined:
Feb 5, 2010
Messages:
2,065
30
Trophy Points:
48
+155 / 10 / -2
Did a hair dresser write this crap?
10. Offline

### OaktownGatorWell-Known Member

Joined:
Apr 3, 2007
Messages:
23,645
1,932
Trophy Points:
113
+3,420 / 81 / -24
My hair dresser has a son playing ball, and she's insulted by that remark. :joecool:
11. Offline

### orangeblueorangeblueWell-Known Member

Joined:
Apr 8, 2007
Messages:
54,784
421
Trophy Points:
83
+1,264 / 88 / -23
Game theory heavily weighs risk aversion, which is applicable here in that there are technically only two plays.

The problem with employing game theory on a single action is it ignores down and distance and assumes a *win* on gained versus lost yards. Risk aversion is a variable on both sides dependent on those many ancillary variables (they included *star* versus *regular* running back here as their only real variable).

For instance, if I'm in 3rd and 25 5 times in the game and I gain 50 yards by running it (10 yards each time), this is counted as a game theory *win* versus passing it (given risk aversion and probability), but I still end up punting the ball.
12. Offline

### MaceoPWell-Known Member

Joined:
Apr 3, 2007
Messages:
2,574
33
Trophy Points:
48
+96 / 5 / -1
2nd paragraph, 1st line

"Suppose that my team, which is on offense, has an equally effective running and passing game."

Obviously, we have been having major problems in protection on passing plays, for the last couple of years. IMO, that supposition throws out the rest of the example.

If and that's a big IF, we could pass-block, only then could we really find out what kind of passing game we have.
13. Offline

### trufloridagatorVIP Member

Joined:
Apr 3, 2007
Messages:
6,274
6
Trophy Points:
38
+7 / 0 / -0
It's hard to measure, but we could just take a logical approach to calling plays that gives us the most EV while increasing future EV where possible. Risk is already factored in. You don't have to break it down, but just analyze at the surface. What we are doing now is fine in many spots, but first and second down are a problem. It's like the coaches don't believe we can complete 15-30 yard throws on first down.

Sent from my iPhone using GatorCountry
14. Offline

### maxgatorWell-Known Member

Joined:
Apr 3, 2007
Messages:
3,240
27
Trophy Points:
48
+102 / 6 / -0
That was an absurdly simple analogy to football. I wouldn't go basing a decision on whether to pass more on that. lol.
15. Offline

### orangeblueorangeblueWell-Known Member

Joined:
Apr 8, 2007
Messages:
54,784
421
Trophy Points:
83
+1,264 / 88 / -23
At its core Game Theory is extremely simple. But it can be effectively applied to any sport with two or more competitors. But this basically treats a single play as a "game," which is why it isn't particularly applicable to full game strategy.
16. Offline

### maxgatorWell-Known Member

Joined:
Apr 3, 2007
Messages:
3,240
27
Trophy Points:
48
+102 / 6 / -0
I understand game theory. I'm basically referring to the expected outcomes and their derivation than anything else. Choosing 5 yards and 10 yards as expected outcomes generates the desired result.
17. Offline

### orangeblueorangeblueWell-Known Member

Joined:
Apr 8, 2007
Messages:
54,784
421
Trophy Points:
83
+1,264 / 88 / -23
Yeah, absolutely, there's a ton of arbitrary data in there. But you can only work on arbitrary data or averaged data.
18. Online

### qwghlmgatorWell-Known Member

Joined:
Apr 3, 2007
Messages:
3,244
66
Trophy Points:
48
+104 / 24 / -9
While superficially ture, this is a highly reductive view of game theory. Game theory does not provide pat answers like "run less."

What makes Game theory (notice the word "theory" there) potentially powerful (and practically impossible to fully utilize in this sort of context) is the inherent fact that in any "game" the effective rules change after every "turn" - with the iterative effect of a dynamically changing model. In other words, if the OPs link is correct and a stud RB causes you to run less but throw more successfully, the defense will respond to that combination of tendency and success and perhaps sub a safety for a linebacker. This will cause you to make a different decision based on different inputs. For instance, you would probably run your stud RB more. But this would cause the defense to restrategize based on your increased running. Etc.

While this give-and-take will theoretically establish some equilibrium, it will be a dynamic equilibrium. Hence, it will provide little explanatory or predictive power. That is it might provide you with insight into how people behave in game-like situations, but will do almost nothing to tell you what you "should" do in any game situation. Keep in mind, chess computers like "Deep Blue" do not use game theory - they use a library of possible board arrangements and associated nested probablilities (as well as famous gambits, etc.). :hairy:
19. Offline

### orangeblueorangeblueWell-Known Member

Joined:
Apr 8, 2007
Messages:
54,784
421
Trophy Points:
83
+1,264 / 88 / -23
Well that was sort of my point - each (and any) play is treated here as a vacuous event. That isn't helpful in determining strategy.
20. Offline

Joined:
Aug 10, 2007
Messages:
736