While I know the game theory involved with fantasy football and fantasy baseball is different, there are a couple of elements of fantasy football analysis that bug me. The first is how much of football advice is “what have you done for me lately.” Friends, Darren McFadden and Wes Welker are the same players they were a week ago. McFadden is not going to catch 208 passes and Welker is not going to only snag 42.
The second pet peeve I have is how much of fantasy football analysis is based on anecdotal suppositions as opposed to research that demonstrates your decision is based on probability giving you the best chance to be right. Remember, your goal is to be right, but the real objective is to choose the option that gives you the best opportunity to be correct.
I have often wondered just how much a quarterback should be adjusted if he is facing a weak pass defense. Is it enough to jump a lesser one over a better one? What about playing a weaker tight end against a team that allows the most touchdown receptions to a tight end in the entire league? Is there a way to quantify this sort of thing?
So I thought a cute little series of pieces I could write would focus on different matchup. One week I would look at quarterbacks, then running backs, etc. I’d propose some theoretical lineup decisions and ask for your opinions, along with supplying my own based on “the evidence.”
But a funny thing happened on the way to the rough draft. As is often the case when I embark on endeavors of this nature, I discover there are more layers than I originally anticipated. My head was full of questions and it was obvious it was going to take more than a couple of nights of number crunching to handle each position.
The basic premise revolved around taking a player’s average performance and adjusting it based on the opposition. As an example, say a QB averages 230 yards a game and was facing a defense that allows 270 yards a game (with the league average being 245). Should we anticipate splitting the difference and projecting 250? Should we come up with a defense index like a park factor in baseball, in this case 270/245 x 100 = 110 and then adjust the expected yardage to be 230 x 1.10 or 253? And then do the same for touchdowns?
The analysis could be as simple as a correlation coefficient comparing actual performance with expected performance. Whatever algorithm rendered the best correlation could be used to project player performance on a weekly basis and rankings would be facilitated based on matchups. Seems pretty straightforward, no?
Then the questions started to hit me.
Should home field advantage be factored in?
How much does an anomaly such as bad weather, an injury, having a short week to prepare or coming off a bye impact the situation, considering each contest is a fairly significant 6 percent of the season?
How far into the season must we be before the averages, for the both the player and defense stabilize so they can be used in a calculation like that described above?
How reliable are last season’s numbers when it comes to determining averages assuming it takes several weeks for the current season’s numbers to settle?
How many times will the game change direction via an early defensive or special team’s touchdown, forcing a team to alter its desired game plan?
How are averages, indices and expected performance impacted by divisional games considering the familiarity between the two teams?
How does the venue affect matters (beyond home field advantage)? Are some teams better on turf? Outdoors on grass? Better equipped to handle a noisy environment?
How does recent performance against a team alter things?
So many questions, so few answers.
Maybe just sticking with anecdotal analysis is best,
There has to be a better way. I simply refuse to start a lower tier wide receiver because he happened to catch a touchdown against that team last season. I just can’t start a marginal running back because last week, the defense he is facing allowed the opposition runner to go over the century mark and score.
It all comes back to the ultimate objective in the decision making process. There is no way we know definitely what is right. But I do think we can help increase the odds a bit by identifying some global trends. Then we season to taste and make a decision. This, after all, is the fun (though often frustrating if not maddening) part of the hobby; starting a player then watching him go off on Sunday.
So here’s what I propose. I’ll crunch some numbers and from time to time present my findings. In the interim, each week in this space I will pose some theoretical (or perhaps even real) lineup conundrums. I’ll offer my take, most likely using the same anecdotal reasoning that I am suggesting is somewhat fallacious and invite you to do the same. You can chime in down in the comments section or better yet meander on over to the forums where I will have some polls set up where we can vote on a matchup and share our reasons. Heck, if you so desire, please feel free to post your lineup dilemma and I’ll use that as the discussion point in that week’s column.
Here are this week’s theoretical lineup decisions. Let's go with a standard performance scoring format, 6 pts per TD, .1 point per yard rushing or receiving and .05 points per yard passing. I think I’ll post my opinions in the forums.
Peyton Manning at Atlanta OR Joe Flacco at Philadelphia
Stevan Ridley versus Arizona OR Ahmad Bradshaw versus Tampa Bay
Brandon LaFell versus New Orleans OR Santonio Holmes at Pittsburgh
Kyle Rudolph at Indianapolis OR Brandon Pettigrew at San Francisco