Moneyball comes out September 23.
If you haven’t read the book by now, you have roughly two weeks, so get cracking. The book, a must read for brand managers and sports fans alike, examines the story of Billy Beane, managers of the chronically underfunded Oakland A’s and his brilliant application of data intelligence that transformed the A’s into contenders and turned baseball’s conventional wisdom on its head.
Beane’s insight (or more accurately, his stat’s whiz Paul DePodesta’s insight), was to focus on the key stats that contributed to success (in his case, on base percentage) and optimize his strategy to acquire players who would provide it. Lucky for him, the rest of the league was still overvaluing other stats (like slugging and batting avg) allowing him to get his players on the cheap.
It’s an inspiring story, especially for someone who preaches the business value of data intelligence to clients every day. So in honor of the release of Moneyball and the beginning of football season, I thought I’d share some of the secrets to a data-driven approach to fantasy football.
Beating The System
The purpose of fantasy football is to draft a team of professional players who statistically contribute to the success of your team. Each league has a scoring rubric that assigns points for certain statistical milestones. The team with the most points wins.
By far, the most important part of a fantasy football season is the draft. Players alternate picks until they have filled their rosters with as much talent as they can.
There are essentially three ways to “win” a Fantasy Football draft.
- Blind Luck. Sometimes people wind up with amazing picks by sheer luck. Someone breaks out, a starter gets injured, you stumble on an overlooked pick by happenstance… it’s a legit way to win, but not very bankable.
- Superior Football Knowledge. Some drafters have superior football knowledge. This lets them identify players on the verge of breaking out. Yahoo, ESPN and other prediction models tend to undervalue these players and thus they can be snatched up at a bargain
- Better Draft Strategy. If you can’t draft luckier, draft smarter.
Having participated in fantasy drafts before, there are a few insights that are helpful when creating a strategy.
- Most players employ a best player / best position strategy. Most people attempt to choose the best available player for whatever their next most valuable position is.
- This strategy is applied inconstantly. Though players often will alter the order they draft positions (in an effort to get the best player in every position), they have no way of determining when is the best time to draft each position.
- A better draft position should mean a better team. Players fortunate enough to draft early should be able to secure more of the players they want. If a player has not gotten better players with earlier picks (based on available information), he has not drafted well.
Moneyball worked because DePodesta realized that teams should pursue wins, not players.
Once he determined that On-Base-Percentage was the biggest contributor to wins, he was able to optimize his draft strategy to secure those numbers. Lucky for me, I already know exactly what contributes to fantasy wins: points. I also know how to calculate and project points — therefore, I should be optimizing my strategy to create a team that will score the most points, not trying to get the best player in each position.
Putting Strategy to Action
I used a stat called Value Before Replacement (VBR). VBR calculates how much better a fantasy player is over his worst possible replacement.
Since people don’t change their strategies often, last year’s draft is a reasonably good predictor of this year’s draft. I went back to last year’s picks and looked at how many players in each position were chosen by the 10th round.
That is to say, how many quarterbacks were drafted before everyone filled their starting roster. In our league, the answer was 15. So if I don’t pick a quarterback until the last moment, I can be assured to get the 15th best quarterback in the league: Jim Bradford. ESPN predicts he will score 260 points in my league this year.
I compared every QB in the draft to Jim Bradford to compute their VBR. Since Aaron Rodgers will score 101 more points than Jim Bradford, he has a VBR of +101. I did this for every position and sorted my list by VBR. That was my draft list.
This way, I could be sure to always pick the player that would contribute the most marginal points to my squad (though not necessarily the best player available at that moment). This allowed me to secure the most additional points for my team with every draft pick I spend.
So how did it work?
The chart above plots my team’s cumulative points against others who drafted in my league. (I’m in red, having drafted 7th). You can see that my strategy was successful. I drafted more points in the first 12 rounds than anyone else in the league.
Will I win the league this year? Trickier question. There are some flaws in the strategy.
- ESPN projections are likely to be wrong. My strategy maximized projected points, so if they’re not accurate, I’m not accurate. Unfortunately, projections are wrong all the time. Still, data intelligence is about making the best decisions with the information you have, and my strategy allowed me to do that.
- There’s more to fantasy football than just the draft. Trade strategy, start strategy, a bit of luck… I don’t have any special advantage in these departments. I can still mismanage my team to a disappointing finish.
- Stuff happens. Last year, Kevin Kolb got hurt in game three and Mike Vick filled in with his best season ever. Football is unpredictable. Sometimes you never know.
Winning At Business
For Billy Beane it was on base percentage. For me it was points. What metrics are driving your business? How are you optimizing your plans to maximize those numbers?
Don’t just look at metrics that benchmark success, look for the numbers that actually drive success.
Don’t play ball, play Moneyball.