No Frills DFS Data - Lemonade and Iced Tea Player Core Belated Edition (Longread)

If you’re new to reading this, an introduction to me and the model itself can be found here.

I unfortunately had some things come up so wasn’t able to get anything up before lock. I ran everything over a full day in advance to make sure I could do it but sometimes life just gets in the way. The time I did have available was spent frantically rechecking my model – and because everything was done early this time – this also meant accounting for late withdrawals such as Gooch. Pro-tip – if you ever need to remove a player from consideration, just manipulate their salary by adding a 0 to the end. You never know what may inadvertently break whenever you make any changes, so the best change is the most minor. Changing the data field for a player’s salary won’t have nearly as many unforeseen consequences as removing the player. I’ll discuss this a bit more at the end.

Original, untainted, and unmanipulated player pool

Lucas Glover
Rory McIlroy
Justin Rose
Corey Conners
Hideki Matsuyama
Daniel Berger
Michael Thompson
Luke List
Jason Kokrak
Rickie Fowler
Ian Poulter
Steve Stricker
Henrik Stenson
Chesson Hadley
Joaquin Niemann
Cameron Champ
Bubba Watson
Sungjae Im
Brian Gay
Ryan Moore
Keith Mitchell
Bryson DeChambeau
Jason Day
Keegan Bradley
Stewart Cink
Patrick Reed
Danny Willett
Brooks Koepka
Byeong Hun An
Francesco Molinari

The manually trimmed down final player pool

Lucas Glover
Rory McIlroy
Justin Rose
Michael Thompson
Jason Kokrak
Rickie Fowler
Cameron Champ
Hideki Matsuyama
Ian Poulter
Steve Stricker
Henrik Stenson
Chesson Hadley
Corey Conners
Luke List
Joaquin Niemann
Daniel Berger
Sungjae Im
Brian Gay
Ryan Moore
Bryson DeChambeau
Jason Day
Keegan Bradley
Stewart Cink
Brooks Koepka
Francesco Molinari

Thoughts on the player pool

I feel it’s going to be crazy chalk as most of these names are the names that keep popping up. It’s also very top heavy, my fate is highly dependent upon Glover/McIlroy/Rose/Thompson/Kokrak as there 2 or more of them in all but the exposure control lineup. This was after heavily favoring the lineups that had less saturation of them as well – if I let my model have it’s way it would only play those guys. I think this is one of the downsides of investing so much time in a model – even though I know it’s really risky and really dumb to go such high exposure on just those few guys – at the same time, I feel like I’ve earned it because “I have a process” that validates and encourages this plan. The truth is all I know is that this system provides winning results just frequently enough to provide a positive ROI in the long run – meaning it’s often wrong and thus clearly shouldn’t be given carte blanche.

While I think with only 10x lineups out there, it’s not the end of the world having pretty high exposure to few guys, however, given that they are all chalky and not exactly all top 10 favorites, it’s more risk than I would care to take. I’ll happily go very heavy on a certain player in other sports but then again, you have a much better idea of the possible outcomes for other sports. If Lebron has a bad game it means he only scored 22 points with 6 rebounds and 5 assists. He doesn’t frequently not only not deliver very much in terms of fantasy points but then also sits out the second half – which is the equivalent of missing the cut in golf.

Honestly, the worst case scenario for me is to have an average week with this approach as I clearly don’t have a good understanding of the optimal manner to approach 10x gpp lineups in golf and and average result of slightly positive or slightly negative may not be enough of an incentive to really study up on that and at least try to find an answer so I don’t go through this exposure agony ever again. Either smashing it or getting cleaned out will definitely incentivise me to put this study session higher up the priority list. Having said all of that, this is why I love DFS golf, I’ve never experienced these thoughts or feelings from any other DFS sport.

Thoughts on the exposure

The thing about golf is if I do well this week or not, that’s still not indicative of the strategy in a vacuum. That’s the thing about golf, all analysis is results based rather than process based so there’s a whole lot of post hoc ergo propter hoc at it’s finest. In no other sport am I constantly filled so many pangs of doubt and regret – and this is from a winning player. I have a slight suspicion that’s the main allure of it all.

Now, there were “outs” so so to speak. I had enough options out there where I could have faded those 5 guys entirely instead of forming an infinity gauntlet. So I obviously could have done a more conservative exposure to those players but chose not to do it. The fact that this is a shorter field than usual and a good chunk are non competitive was definitely a big influence in my decision to Ryazan this with a go big or go home approach. Again, I really hope to figure this out going forward. This is actually why I write the blog, before doing so my thoughts were “man this week was tough, ok moving onto next slate.” I’ve already written everything adding this now, that in the process of writing it down and decompressing it all, it really got me thinking more in specifics of what I should work on going forward. I now have a clearer plan of action and am off to the races.

I’d love for people to comment on their typical exposures for a smaller set of lineups in golf, especially if you include the reasoning why.

Manscaping

Trimming this down to a manageable 10 lineups was a greater struggle than I’ve ever experienced before. This slate presented some very unique challenges. The thinking was that I am definitely going to play those 5 heavy exposure guys but came in without any intention of forming the gauntlet in basically an all or nothing approach. In fact, I tried very hard to find any excuse to eliminate lineups they were in as contenders. The main problem I faced here is that none of the horses I could swap in for Rory or Rose were people my model really liked very much.

The last thing I wanted was a single serving of one of the favorites unless it built up upon a lineup that already had one of my main horses in Rose or Rory. The thinking being if Bubba Watson wins this and I have him in just one lineup then I’m still likely to lose money overall on the slate unless Rory or Rose do very well too. Furthermore, the likelihood of a single Bubba dart also having the correct complementary pieces around him to hit yahtzee was very unlikely. However, I do believe that it’s ok to have that lone helping of Bubba so long as he’s paired with a Rory or Rose. Bubba winning or losing won’t impact my slate unless Rose or Rory win too so may as well go for some correlation. So I go through my lineups and eliminate the lone wolfs from contention. Bubba only showed up in two lineups, and both times as a lone wolf, so it was an easy decision to cut those from financial consideration. For what it’s worth, I actually feel the opposite about the cheaper guys. I’d rather have 3 different punts used once than relying on one punt 3x unless there is a very compelling argument he’s dangerously mispriced – which often happens in golf.

Ryazan 101 – be confident in your lineups

I then took a look at the other lineups and got some real oddball numbers out there. An entire series of lineups were running at only about 90-95% salary cap. This had come up before in some of the more niche datasets, but usually as a one off or due to an error (ie I’m solving for lowest amount and N/A was mistakenly converted to 0 rather than 999 because I just copied the code I just used on another set where more was better and wasn’t thinking through). When it was a one off, I’d usually proceed with caution and look for reasons not to play it but wouldn’t take issue with it due to salary alone. While it can feel scary leaving money on the table, golf in particular is an inefficient market and rarely does the perfect lineup use up all the cap. Furthermore, it’s a bit silly, if some of those guys just cost a few more dollars (ie overpriced) then I wouldn’t be raising red flags and jamming in the lineup. However, this time it was different, it was a whole slew of correlated lineups. After looking it over, I could see why these guys were so cheap that I could do a “perfect build” under the algorithm because they really only performed well in the specific metrics these algorithms were chasing. Normally you pick out some random data and you’ll get back a pool of the more expensive players with few surprises mixed in.

While normally I’m ok leaving some money on the table (my cash lineup left 2%) this was a horse of a different color. In fact, the one thing I’ve found is that regardless about how absurd the metric will be, the model almost runs out of money trying to find it and thus the cap heavily influences the final lineup. Missing by a small amount means the cap could still be in play as the next best respective players all cost considerably more and there were weren’t feasible “downgrade to upgrade” options available. That’s the one thing that gave me hope that building this project could lead somewhere and gave me some hope maybe I could find some winning correlations.

The lineups in questions were several in a series that all left serious amounts on the table. I took a look at the players and sure enough, the guys who excelled in these areas were pretty horrible elsewhere. So I then decided one way to qualify the players as worthy would be to ensure the model used at least 95% of the cap and ran it again. Sure enough, now they were using 95% exactly, like I had twisted their arms and instead of giving in they insulted my mother. I tried 100% and it still found players but again it was simply swapping out a single cheap player for an expensive one to meet the financial limitations. I’d put so many artificial constraints in that I was no longer even measuring what these algorithms were intended to measure. I took a look at the originals, the 95%ers the 100%ers again and with each variation I couldn’t see anything that made sense to me. With each one I kept visualizing an upgrade, thinking but this guy is so much better and is easily affordable.

In essence, these were not confidence inducing and again, channeling the ghost of Ryazan, I tossed them to the side to track and analyze later but not put money at play with – of course these lineups were really my last refuge that was free of infinity gauntlet pieces. The remaining two dozen or so were all heavily those players. The end result was what felt like a lifetime agonizing over what to do. Most other options seemed even worse as the lineups that didn’t have these guys looked like a worse path than this gauntlet. In the end, I called upon my exposure algorithm and added a zero to the salary of those guys to ensure they wouldn’t pop up. I ran one of my more go to algorithms and it gave a result that only spent about 85% of the cap. I swallowed the puke forming in my throat and loaded up a new algorithm, this time it maxed out salary and passed the sniff test so onward we would go. Next I felt like using a homer line where I added a zero to a key positive metric on Niemann and Champ to ensure they’d be selected. The reason being I like playing them and DFS is supposed to be fun. At this point when the model suggested the ever so predictable supporting cast, this time instead of swallowing my puke, I just swallowed my pride, reminded myself this is supposed to be fun. I loaded in the rosters, poured myself a drink and called it a day.

About the Author

Comments

  • X Unread Thread
  • X Thread with New Replies*
  • *Jumps to your first unread reply

New to DFS?

Be sure to click through our links and use our exclusive promo codes to receive the industry's best sign-up bonuses, including free access to our premium content.

RotoGrinders.com is the home of the daily fantasy sports community. Our content, rankings, member blogs, promotions and forum discussion all cater to the players that like to create a new fantasy team every day of the week. Our goal is to help all of our members make more money playing daily fantasy sports!

Disclosures: All RotoGrinders content contributors are active DFS players. Contributor screen names can be found on their respective RotoGrinders profile pages. Contributors reserve the right to use players or strategies not discussed in their content on RotoGrinders.