First off, Happy New Year! 2020 is almost over and we can all look forward to a better year, hopefully.
Secondly, we have a nice 4 game slate on FanDuel (FD) tonight. I will be avoiding DraftKings (DK) since I don't care for the earlier game but you can use my sample FD lineups to build something similar for DK.
So wht lineups were produced by our models? I use a multiple linear regression formula with minutes, usage, pace, offensive efficiency, defensive efficiency, implied point totals, and a logarthmic transformation of DK salary as my inputs. Then I use an optimizer I found on github to produce the top 20 lineups. Sometimes I roll with the top lineup and sometimes I will do a combination of players who show up the most often in the top 20 lineups. For tonight, I will do the top lineup and it's displayed below.
As some of you know, I work for Advanced Sports Analytics (ASA) and this year I will be contributing to its NBA data flow. The predictive model ASA uses for it's NBA DFS projections was built by Stewart Gibson who is a brilliant data scientist. It is a more sophisticated model than mine but uses similar inputs. His outputs tend to predict lower scores than mine so it will be interesting to see how both of our top lineups compare this evening. The top NBA lineup for FD from the ASA site is shown below and you can purchase a membership at the ASA site using the provided link.
You will see that both of the lineups have 5 players in common: Mike Conley, Buddy Hield, RJ Barrett, Pascal Siakam, and P.J. Tucker. It is weird to put an obscure player like Barrett on the cover photo of this article but 2020 has been a weird year so not end year in a weid way!
A key player leftout from both models is James Harden. I believe NBA DFS this year will be very different than past years. The new NBA schedule, I believe, will be a key reason for that difference. The NBA is asking these players to play at a higher frequency. Notice how some teams will see each other twice, 2 nights in a row in the same venue, like the Bucks and Heat this week. I think the fatigue level will be high so we could see more load management of the star players. That would make them awful value in DFS. A typical NBA season has 12 - 13 games per month but now it will be 16 - 17 games per month. There will definitely more back-to-back nights. Personally I think some teams will have already decided to throw in the towel before tip off. Two nights ago, Miami was crushed by Milwaukee. Two of the starters played 2 and 4 minutes. Dragic came off the bench and played just 10 minutes. I can see a typical winning lineup in DFS consisting of more 6 - 8 K players without any players over 10K. That is why I won't roster James Harden this evening. During the past two NBA seasons, I would have my model provide a boost for high-priced players but I won't do that tonight.
Let's see if that decision pays off. Either we will end the year with a bang or limp to the finish line. Let's hope it's the former as we say goodbye to 2020 and hello to 2021! Again, Happy New Year and be safe tonight.