Monday, March 1, 2010

Game Dynamics and Football Part II: Scouting and Play-Calling

In the sport of football coaches are always trying to get an edge, this will never change. One of the biggest things for defensive coaches is scouting offenses in the hopes of being able to call the right defenses at the right time come game day. There are varied views on the use of scouting programs and the reliability of tendencies, however, I think all can agree that we get some positive help from the process. In this article I will explore the game dynamics behind the calls coaches make, the statistical analysis behind them, and finally some ideas and principals that can be used (theoretically) to optimize their play calling.

Dynamic III- Game Theory and Optimization

General Football Strategy

No matter what team you are facing, tendencies will be present, period. Why is this the case? The reason is that coaches understand that certain situations call have particular optimal plays to call. For Example, consider this situation and assume that these teams are evenly matched opponents. You run a power-I offense, it is 3rd and inches on the -20. What type of call makes sense? Some type of inside run or QB sneak seems best, and most would agree with that. Will this play call succeed 100% of the time? No. It will succeed an overwhelming majority of the time, but it is in no way a lock. It is a high percentage play. Do all coaches in this situation run this play type 100% of the time? No. Why not if it appears to be clearly the best choice?

The defense is aware of the optimal play for the offense to run as well. Defense's load up the box and try to stuff the short run, if he could, a standard defensive coordinator would load all 11 of his players to stop the run. Few defenses ever commit all 11 players to stopping the run first, even in a 3rd and short situation. The reason they don't do this, is because the offense could easily exploit them. The offense could fake a base run and then throw a long pass to an uncovered receiver or fake the run and have the QB keep the ball on a naked bootleg. These types of plays force the defense to slow down some of their players on run support. Because of this Offensive coaches throw in little wrinkles here and there to keep the defense honest.

So will a play caller run the inside run/ sneak combo 100% of the time, No, but he will most of the time. Does the small % of time the offense runs something different affect the strategy of the defense? It certainly does. This is the game dynamics that are studied under the field of Game Theory (a branch of Economics and Mathematics). Game Theory applies to the sciences of decision making and competition games in general. Using game theory we can discover optimal choices to make, given certain circumstances.

Analyzing Football Tendencies and Statistics

When we scout, generally there are 4 main tendency areas we analyze:

1. Situation (including Down and Distance)
2. Formation
3. Personnel
4. Field Position

Beginning with situational analysis, we try to group different situations together and see if there is any statistical relationships. 1st and 10, 2nd and long, and 3rd and short are all situations we look at. Most coaches go the extent to break down 2 pt plays. With this data we get a snapshot of what the team is trying to do. For the most part, as coaches we discover obvious things like "they pass more on 3rd and long", "The run inside on 3rd and short", and "They do a bunch of different things on 2nd and medium". Every now and then, we stumble into a very clear down and distance tendency that goes beyond general football strategy. But this is rare, for the most part we discover basic things.

When we begin to break down formations, we get another set of tendencies. Consider this data.

Pro-I Rt formation

70% Run 30% Pass
Runs Right 72% Left 28%

Runs

Iso-Rt 30%
Power Rt 13%
Toss Rt 9%
Counter Lt 18%
Belly Rt 10%
Trap Lt 5%
Iso Lt 5%

When you look at this formation data, you find out that they like to run the right(strong side) twice as much as the left. Does this mean the defense should overload to the strong side when they see this information? No! Why not?

The data shows a tendency to the right side of the formation, however there are enough plays (by number and type) to the left that forces the defenses to respect those plays. If you overplay the right-side of this team, they are going to attack the left side quick. This is a major part of putting together a scheme. A classical example of this is the toss pass. If a team is killing you with the toss, sooner or later your corner (assuming he plays pass first deep 1/3 or 1/4) is gonna come off and play the toss aggressively. When a play caller sees that corner make a tackle close to the LOS. You can bet that the toss pass is coming up real soon. Because of the threat of the toss pass, the corner cannot come up on the toss aggressively. A team might never even have to run a toss pass, the threat of it is enough to keep the defense honest.


From Personnel and field position we can gather similar sets of information, but for the most part we will be in the same situation the first two areas, seeing a tendency, but limited in how aggressive it can be attacked.

Why do we scout then?

If you begin to combine the various things together a more accruate and clear picture can become present. You might find that 3rd and 8 Pro-I right formation is 60% pass with 80% of the passes being bootlegs. Given the game situation you might beleive that the opposing play caller will favor the pass. The other passes he has shown in this situation thus far are 5 step drop back passes. Given this play calling range, sending a backer/safety off the edge to the Boot Leg side seems optimal (as long as you get a body on the full back leaking out). Also, if it happens to be a running play, you don't lose much bringing an edge rusher. By combining different areas of statistics we can find more reliable albeit more particular tendencies. With the limited time we have to scout, we can only analyze so many situations. However the situations we do analyze can become very helpful come game-day. What can the offense do to combat this? Simple, balance his Play-call range.

Balancing Range Example-Poker

I don't know how many of you are poker players, but games like No-Limit Texas Hold-em involve similar dynamics to the football situation discussed above. Consider this situation (not realistic, but simplified for the example, you are playing someone heads up (2 players total), and you know this player will raise Pre-flop with any pair, and any combination of aces and face cards, and a few others. Further more, you know that if you re-raise him after he raises, that he will only call with Pocket Aces, Kings, and Queens. Given this situation, you (ideally) should re-raise every time he raises. You would exploit this player because he would allow you to bluff him too much because of his limited calling range.

How can this player keep himself from being exploited? First off he should call with more hands, and begin pushing all-in with some of his hands too, because you will be raising him with weak hands at times. If he adjust this way, you are put into a guessing game. This forces you to play your hands in a more standard way. How did this opponent keep you from exploiting him? He balanced the way he played his hand to keep you from exploiting him.

This applies to football too.


Back to Football.....

Assume that you are an offensive play caller. On 3rd down and long out of a 2x2 Gun formation you have the following tendencies

5 Step pass 80%

Sprint out Pass 10%

Run 10%

Given this distribution, a defensive coordinator is gonna be excited about this situation. The optimal call here is to blitz, because these passes take time to develop. To prevent this the offense coordinator should make calls to balance the situation and deter the defense from stunting. The calls used to balance the range are plays that hurt heavy blitzing. A more balanced range would like this:

5 Step Pass 65%

Sprint out Pass 10%

Screen 15%

Draw 10%

This presence of 25% screens/draws slows the DC's willingness to blitz. He must respect your ability to screen and run draws. If he stunts into one of these plays he could be in bad situation. This distribution changes the optimal call for the play caller. The optimal call is now to play a more basic defense that is focused on defending the pass. By balancing the play calling range the offensive coordinator can protect his QB from seeing constant blitzes. Balancing the call range does not guarantee no blitzes, just that he won't try to blitz every play. If he did happen to blitz every time in this situation, you could call screens and draws liberally to punish his aggressiveness. The best thing for the defensive play-caller to do, is mix up his own calls to include some blitzes and base calls. If I took the time, game theory models could provide the optimal balance.

A Wrench in the Works

" The Intuitive Play Caller"

Some coaches claim to be "feel" guys. They don't need a lot of statistical scouting information or planning for that manner, they believe that they can feel the game out and know what to call. I am not saying that these people don't exist, but how do you deal with these play callers? Let's assume that some of these people are so good that they can know what you are gonna call before you do at times? I am sure some of you reading this have had this experience before, when you just knew what a person was going to call. How do you combat this person, if they in fact are able to know what plays you are going to call? Simple, be unpredictable, to the extent that you have some randomness in the range of your play-calls. Here is an example to illustrate:

Lets say you are playing a person in Rock, Paper, Scissors. This person is astute at knowing what you are going to select, to the point he can beat you 2/3rds of the time. There is nothing you really can do to get an edge on this person, but you can reduce the edge he has and even the game. How do you do this, be unpredictable. If you look at your watch and the seconds hand is on a #1-3 you pick Rock, #4-6 you pick Paper, and #7-9 you pick Scissors. If your opponent has no idea that you are doing this, there is no way he can guess what you are going to throw. The game will return to its basic design, a guessing game based on chance, similar to flipping a coin, or rolling dice.

In football, if you are faced with a guy that knows how you are going to call plays, you could attempt something similar. Perhaps not to this extreme of an extent, but something to keep your opponent guessing. For Example, assume you want your opponent to respect your punt fakes, and you want to fake 10% of the time. Before you punt look at the score board if the game clock ends in 7 you run a fake. If it ends in any other # you punt the ball. If you keep this to yourself, no one will know when you will run a fake, they will be forced into a guessing game, with no idea of how to know when you will do what.


Conclusion

Game Theory plays a role in any game that involves decision making. The decision making process is at the heart of coaching on game-day. People like Bill Walsh were masters of planning and to an extent his ideas were consistent with Game Theory Optimal ideas. Just like in poker, keeping your opponent guessing makes life easier for you and your team.

The examples and ideas in this article were grounded in the "Ideal/Theoretical" domain. However, most of the time decisions on the field transcend this domain. These examples existed in a vacuum and involved small tidbits of information. This article was aimed at showing how optimal strategies and calls do exist given the information that you have. In the next article, I will go to the dynamics beyond game theory, and involve the importance of understanding the thinking of your opponent. That much of optimal scouting goes beyond the situation and formation tendencies, but extends into understanding the overall philosophy and though process of your opponent. In games of decision making, this psychological and logical dynamic is often the one that separates the good from the great.

3 comments:

  1. You are right. Even bookmakers online track coaches to make prediction about games.

    ReplyDelete
  2. Very informative post, i am regular reader of your blog.
    I noticed that your blog is outranked by many other blogs in google’s search results.
    You deserve to be in top ten. I know what can help you, search in google
    for:
    Omond’s tips outsource the work visit my blog fifa 15 generator

    ReplyDelete
  3. I really like the fresh perspective you did on the issue. I will be back soon to check up on new posts! Thank you!
    professional football scouts

    ReplyDelete