NFL Daily Fantasy: Using Correlations To Win A Big Tournament
Draftkings and Fanduel are both holding one million dollar guaranteed prize pool (Also known as GPP) tournaments next week for NFL Week One. Max wrote a great article on how to go about making a lineup. This article will go into another factor you should use when making a lineup specifically for a GPP: Maximizing your lineups potential and variance.
Understanding Your Goals
For those who are unfamiliar with Draftkings, Fanduel, or Daily Fantasy, a GPP is essentially the following: A tournament with 1000s of people, where 10% to 20% of the players entered will win money. The tournament normally will pay out around a tenth of the entire prize pool to 1st place, with most of the rest of the money going to the top 10 finishers. This is very similar to the way a poker tournament is paid out.
To win, you need to have a great lineup, one capable of beating 1000s of other players. This is not easy to do, even if you feel like you’ve made all the best picks at every position. Because of the payout structure, all you care about is having a lineup that does extremely well.
QB and WR Combos
Let’s examine a toy situation. The Eagles vs Jaguars game has ended. We know nothing about the results of the game. We look at one player’s statistics without looking at anyone else, Riley Cooper. Cooper has 100 yards recieving and 2 TDs, and incredible day by any receivers standards. Now before looking at the rest of the box score, what do we also know? We know that in all likelihood Nick Foles, the Eagles QB, also had a good day. Why? Because Cooper and Foles have fantasy performances that are positively correlated. If Cooper has 100 yards and 2 TDs, that means 99% of the time Foles will have thrown at least 2 passing touchdowns, if not more.
Now lets start over. Let’s say we look at Foles’ statistics first. Not a good game, he has 200 yards passing, 1 TD, and 2 INTs. What can we guess about Cooper’s performance now? He could have had a good game, but it’s not likely. He would have had to catch the only passing TD, and even if he did, he may not have very many catches or yards.
QBs and WRs have extremely strong positive correlations when it comes to their fantasy output. If one does well, the other is likely to do well too. And if one does badly, the other probably did badly too. Because we don’t care whether we have a good lineup or a terrible lineup, only that we have a great lineup, using QB/WR combos and other positively correlated player combos is essential to winning a GPP.
QB and RB, WR and WR, etc.
There are also players who have strong negative correlations to each other. One being two WR’s on the same team.
Think of a player having a good or bad fantasy day as having a big or small piece of a pie. Andre Johnson has 200 yards receiving and 2 TDs, that’s a big piece of pie, a HUGE piece of pie really on a lowly Texans team not expected to be an offensive juggernaut. A QB will share that pie with him since he threw those TDs, but the rest of the team will now just have crumbs of the pie left to eat. So when Johnson has a big day, it’s extremely likely the Texans RB as well as the other WRs on the team won’t have a great day. In fact, in Johnson’s monster performance against the Colts last year, that’s exactly what we see.
In general, it’s not a good idea to use a QB and RB on the same team in the same lineup for a GPP (although this is perfectly acceptable in a head to head or 50/50 game). Same goes for two WRs or TEs on the same team, a WR and a RB, or a defense going against offensive players on your team.
Sample Correlation Maximizing Lineup
This is a sample lineup for the Fanduel Sunday Million on September 7th, opening Sunday for the NFL.
- RBLeSean McCoyP$9,400JAC@PHI
- WREmmanuel Sanders$6,400IND@DEN
- WRT.Y. Hilton$6,400IND@DEN
- WRCecil ShortsP$5,900JAC@PHI
- TEVernon Davis$6,300SF@DAL
- KPhil Dawson$5,300SF@DAL
- DSan Francisco 49ers$5,600SF@DAL
The main correlation pick here is Kaepernick and Davis at TE. Kaepernick has a great matchup going against an extremely poor defense last year in Dallas who lost key defensive players such as Jason Hatcher to free agency. It’s safe to say they should be one of the worst defenses in the league this year, so its going to one to exploit week after week. Since they lost LB Sean Lee, their best linebacker by far, I think Davis is going to have a field day.
For smaller correlations I go for the 49ers defense. The 49ers defense has a small positive correlation with the offense. If the offense is scoring a lot, it means that the 49ers are likely to be ahead. A defense will create more turnovers when the opposing team is trailing and needs to take risks to catch up. I also use the 49ers kicker Phil Dawson. This will leverage an extremely good game from the 49ers, but I think generally because a field goal means an offense did not score a TD, you can go either way with stacking a kicker on top of a QB/WR combo.
For an even smaller leverage I use two players each from the Eagles and Broncos game. This will be advantageous if either game goes into overtime, increasing the amount of plays and therefore fantasy output from both teams. This is not necessary by any means, but I like the matchup and value for McCoy, Sanders, Shorts, and Hilton, so it just happened to work out that way.
GPP format encourages you to maximize the variance of your lineup, which you can do by choosing players whose fantasy output will be positively correlated. You also should minimize choosing two people from the same team, aside from a QB/WR combo, because all other players on the same team have negatively correlated fantasy output. There are also small correlations you can exploit such as choosing players from the same game, and choosing the team defense of your QB/WR combo.