/ DFS

How to make thousands of dollars in fantasy football

OK. You might be thinking I am some sort of quack with the title of this article. Yes, you do need a good degree of luck to succeed in a game like DraftKings or FanDuel. It is also a form of gambling which I know is not quite for the faint of heart. Like the game of poker, however, daily fantasy football is also a game of skill and if you have a model/plan that uses the available data correctly, you can do very well.

Take for example the model I created and have used for football predictions since 2000 (called the GE model). At first the GE model was strictly used to rank each NFL team's offense and defense and determine the winners of each game. Several years later I proceeded to run some linear regression analysis to predict point spreads and figure out which direction wagers should be placed. Last winter, I took the model a bit further by calculating the differential between offensive and defensive GE numbers between two teams playing each other in the same game. If I found that the the offensive GE number (GE-O) for a particular team was significantly greater than its opponent's defensive GE number (GE-D) and vice versa (>1), then I suspected the game would be a shootout. Of course if you select the key players in such a game, you should expect a very nice fantasy performance.

Case in point: Week 16 of the 2015 NFL season. Jacksonville was at New Orleans and both teams had solid offenses and terrible defenses. The GE model backed this up by showing a GE-O and GE-D differential of greater than 1 for each team (comparing GE-O of one team with GE-D of the other team and vice versa). I selected Blake Bortles, Allen Robinson, Brandin Cooks, and Tim Hightower and all four went off (at least 120 combined fantasy points).

Another interesting aspect of this algorithm was to pick up a RB on a team which had a positive differential when comparing its GE-O to its opponents GE-D but the vice verse situation was negative. The Steelers and Cardinals were such teams since the Steelers were at the Ravens and the Cardinals were hosting the Packers. The rationale is that if a team has a much better offense than its opponent's defense and also that same team's defense is much better than its opponent's offense, then one should expect the stronger team to build an early lead and then just run the ball for most of the game to protect the lead. DeAngelo Williams and David Johnson were the RBs on the Steelers and Cardinals, respectively, and both RBs performed well by combining for about 50 fantasy points.

Going cheap on tight end and defense can be very beneficial since if you can find some diamonds in the rough and use your extra money to pay up for key positions like QB, RB, and WR, you should earn money in your tournament. During Week 16, I took Zach Ertz who was getting a lot of targets on the Eagles and the Houston defense which allowed just 6 points to the Titans and returned a fumble for a TD. The GE model had a very negative differential between Tennessee's offense and Houston's defense. Both Ertz and the Houston defense combined for about 50 fantasy points.

Then you need to get a bit lucky with one of your cheap players and that player was Jermaine Kearse. The Seahawks hosted the Rams and Kearse caught a TD in the final minute of the game. As you can imagine, I did quite well in this tournament: 3rd place out of 500+ participants and a prize of $10,000 (entry fee of $300).

The one advantage of using this algorithm in Week 16 is that you can really trust the data as the sample size is large. Also, Week 16 games still matter with several playoff spots and seeds up for grabs. The algorithm was kind of useless for Week 17 with so many trivial games and the pool of players was too small for the postseason. This algorithm hasn't been too useful for the start of this season with such a small sample size but now we are heading into Week 7. The data is definitely more reliable now.

So who does the model favor this week? Click here to see which offenses and defenses should do well. Any differential of 1 or greater is good for the offense and any differential of -1 or lower is good for the defense.

Expect San Diego at Atlanta to be a shootout. Also, expect Minnesota at Philadelphia to be a slugfest. You can probably use this algorithm for predicting the Over-Under line but I haven't run the correlation analysis yet.

Here is a possible line up I like: Matt Ryan, Mike Gillislee, Terrance West, Julio Jones, AJ Green, Pierre Garcon, Hunter Henry, DeMarco Murray, and Kansas City DST.

You could also swap Ryan with Marcus Mariota and swap Murray with Rob Gronkowski. Then pay up for a DST like Minnesota.

Hopefully you will find some utility with this algorithm and good luck tomorrow!