A while back, we posted a link to an article from SmartFootball discussing a previous article by Malcolm Gladwell on the distinctive categories of strategies used by teams of differing prestige. I strongly encourage you to read each of these articles before continuing with this one, as they are both excellent reads and will provide the necessary context for this discussion.
Gladwell reaches the conclusion that underdogs should rely on "David" strategies in order to have any chance of upsetting powerhouse teams. "David" strategies are designated as "unconventional" approaches that have very high risks, but equally high rewards. Examples cited include the press in basketball, high volume passing games in football, and trick/gimmick plays.
Obviously a David very rarely has a chance to beat a Goliath using normal tactics. Each article accurately notes that if a David tries something risky, and it fails, then it doesn't matter much because David would have probably lost anyway. However, if David team tries something risky, and it works, then they have given themselves a chance to pull the upset.
Gladwell seems to support the idea that all teams, even Goliaths, would do well to incorporate these David strategies into their approaches. SmartFootball, in contrast, argues that if Goliaths were to adopt David strategies, then the large variation of these approaches would lead to the Goliath losing more games that they shouldn't. The example cited (in a related article at SmartFootball, given here) is of Florida in the 1990s.
"I can't believe I'm inclined to say this, but maybe Spurrier should have been more conservative? He might not have won as many games by sixty or seventy, but maybe they would have gone undefeated and won more than one title?"
First, I think they underestimate the role Florida State and Tennessee played in keeping Florida from dominating during those years. Secondly, I think we have learned in college football that virtually every team will lose games that it shouldn't. USC uses a generally Goliath approach, yet they too have lost several games this decade that they shouldn't have. Does this mean USC should be even more conservative in their style, or are there other causes that make these losses inevitable?
After the jump, we will introduce the tenants Statistical Quality Control and see how easily they apply to the football framework. In Part 2 we will discuss how a riskier approach can still yield an increase in expected points. In Part 3, we will look at Goliath's potential advantages of a David strategy from a probabilistic perspective. Finally, Part 4 will take a look at a few current Goliaths in David's clothing within the college football universe.
Many business types will already be familiar with the basic concepts of quality control, as Six Sigma and other strategies have long been an integral part of many management programs. However, when simply taken as a statistical or analytical tool, quality control techniques can be used to improve any process-- including football. The goal of QC is to reduce the variation of a process by isolating the particular inputs that cause this variation. Additionally, the combination of quality control and experimental design allows for the selection of these inputs that will not only enhance quality by reducing process variation but by optimizing each individual unit. I'll explain by a few examples.
Consider a factory production line and two foremen in charge of the line. Each foreman is charged with making sure the line generates the best possible product for their company. As a result, they both decide to analyze their products, remove defective items, and keep the best ones. Foreman A hasn't had any experience with process improvement. He simply waits until all the products come off the assembly line and checks them. Those that are defective are discarded, and the good ones are kept. Foreman B took a Quality Control 101 course in college and understands that he should look at the entire process, not just the result. Consequently, he tracks a few of the products throughout the assembly line and discovers that a loose bolt on one of a particular piece of equipment is causing every 5th product to be defective. He is able to isolate the problem within the process, thereby reducing the number of defective products and increasing the profitability of his company.
Foreman B's bosses are so excited that they want him to try to find other ways to improve their production line. Once again, Foreman B uses his knowledge of quality control in order to find areas within the production process that can be improved. Turns out, at one station on this particular assembly line, there is an oven that bakes the products in order to harden a plastic shell around the product. For the last few years, the products have been placed into a wooden holder within the oven which is set at 200 degrees. Using these settings, the company has found that 80% of their plastic shells meet their design specifications, The remaining 20% are discarded. Foreman B decides to conduct an experiment to check these settings to see if they can be improved. After experimenting with different combinations of temperatures and holders, the foreman discovers that if the oven is set at 250 degrees and the products are placed in aluminum holders, that over 98% of them meet the desired specifications. Foreman B has analyzed the process to find areas of weakness and has optimized the factors over which he has control in order to improve the overall production process-- he has just made himself the head of Quality Control.
The key to reducing variance and improving quality is knowing that there are some variations you can control and some you cannot. The simple fact the assembly line is a repeated process will ensure some slight random variation in the quality of a product (eg... sometimes there might be 11.99998 ounces of Coke in your can, sometimes there might be 12.000001, it's still sold to you as a 12oz Coke). However, there are special causes of variation that can drastically influence the quality of the product that can, and should be corrected (eg... the loose bolts on the equipment). By finding the sources of this special variation and isolating and optimizing the inputs into this system, you can improve the quality of whatever process you're interested in.
I know that might have been heavy for some people. Feel free to take a quick break and get something to drink. May I recommend an 11.99998 oz Coke?
It really only takes an application of terminology to adapt this concept to football. In this case, let us reduce the process of interest to the production of a single successful passing play. Try to picture all of the inputs that go into creating a positive play: a good play call against the particular defensive set, the right personnel, the right formation, a successful snap and catch from the center to the quarterback, a WR running the correct route, a QB making the correct read, a QB making a good throw, and a WR making the catch and possibily getting extra yards (Note: Yards after the catch may be superfluous here depending on the situation. Most completions that aren't bubble screens would be considered a success even without yards after the catch).
Obviously there are natural variations to this process that the players and coaches cannot control: a freak wind could blow the ball off course, the offense could do everything perfectly but a defender still makes a great or lucky play, etc. Most coaches would be satisfied if their player did absolutely everything they were supposed to but the end result of the play was negative. No play will work 100% of the time (except throwing a delayed pass to the tight end against Florida State). If the offense calls the right play and executes the way it should and the defense still makes a play, then you tip your hat and get ready for the next series because you can repeat what you did and the defense isn't likely to. It is the special causes of variation that coaches and players are tasked with eliminating. If the QB is consistently staring down one WR and subsequently every pass is getting intercepted, then this is a problem with one of the inputs that is leading to a defective product. The practice field and the film rooms are the places of experimentation. It is here that players and coaches attempt to optimize their inputs.
Coaches watch film of the other teams to understand their defensive tendencies with the goal of optimizing their play calls based on defensive formation, down, distance, etc. Coaches and players watch film of their own team in order to identify any particular flaws that result in negative plays. Weekly practices are conducted to identify any other flaws and to correct them. It is through coaching and repetition that teams improve their inputs, reduce the variation of the plays by limiting the likelihood of a negative play, and increase their chance at a successful completion.
So we've conceptualized the approach that we will be using in order to reduce the variance of an offensive philosophy.
Once we have established David's plays can result in more expected points per play (Coming in Part 2), we can build an entire playbook of reduced-risk/high-reward plays for David to use. Part 3 will investigate the specific point where Goliaths can gain a statistical advantage from using David's playbook.
To conclude the series, we will take a look at a few examples within college football where Goliaths are thinking like David and thriving.