Daily Fantasy Sports
Sharp App
Season Long Fantasy
Free Month
Wall of Champions
Discord

Why Ownership Matters in GPP – A Quantitative Analysis

(January 18th, 2018)

Disclaimer: I am not staff at the Army. I’m just some guy who likes analyzing data and playing DFS.

Disclaimer 2: The takeaways for this article are GPP specific (an entire different set of principles applies to cash play). In order to analyze this question, I will use the actual data from the 10-game slate last night (1-17-18) on DraftKings. Although the data is from Draftkings, the same principles still apply to Fanduel.

 

I. Understanding ownership indicators

DFs ownership generally spreads out over a 10-game slate, so I’m going to use 15% as the cut off to indicate high owned players here.  Below are the most owned players from the $30K And-One contest which has a total field size of 35671 entries (20 Max entry limit).

If you were followed last night’s slate, you’d be able to explain a lot of things you see in this table:

  • We knew earlier in the day that Barton would step into an increased role with Jamal Murray out (concussion), so the field was expecting another big night for him following a 42-min game the previous night (Barton ended up finishing at 35-mins and made great value on this slate).
  • Jordan Bell got the start for the Warriors and supposedly smashed against CHI in the last matchup, plus GSW vs CHI was one of the highest totals on the slate (albeit, Bell smashing in the previous CHI matchup was without Durant and many Army coaches were pointing out that they thought this to be a good spot for GPP fade).
  • Westbrook in a great spot vs. fast-paced Lakers in a high game total garnered a lot of ownership (but with a sizable lead in the second half of the game OKC studs weren’t needed to do as much).
  • Gasol and Wright ruled out for MEM about an hour before lock which meant Deyonta Davis would draw the start at min salary.
  • AD posted a MONSTER game the night before in OT against the Celts, so this ownership garnered some interest perhaps in chasing his massive performance (although again, Army coaches warned that AD would likely be very tired, even though he always has high upside)

Draw from those scenarios (and others on the list) as you’d like, but generally you can retroactively understand why these players garnered high ownership.  To me, early news on a starter is one of the biggest ownership indicators, followed by min salary starters (either early or anytime leading up to 15 mins before lock since this gives enough time for the field to adjust their lineups to the news).

Some nights this top 10 list will absolutely crush value (in which case, cash lines will be very high) and other nights it won’t (lower cash lines).  If you felt last night’s cash line was low, you might be able to understand why now since the chalk generally did not hit value.

 

II. Top actual plays in two tables

The first table: top 10 value plays (by fantasy pts per $1000).  This table will almost always be comprised of super low owned players since extreme value is usually the result of variance.

The second table is the top 10 players just by raw DK fantasy pts.  This table, in contrast, will almost always have a couple of busts since DK sets pretty aggressive pricing of the predictable top studs such that, even though they might finish in the top 10 of raw fantasy pts for the night, they’re not necessarily the best plays by value (Boogie is an example from last night).

Make your own observations here as you’d like.  A kudos to the Army coaches in that many nights you’ll see a lot of the players from both top 10 tables talked about in the slack notes, along with good reasoning behind why a coach thinks they’re a good/fadeable GPP play.

 

III. Defining the field’s LU

Hedge fund managers are constantly trying to get creative with how they invest in their portfolios and they have to “beat the market” to truly be successful in order to attract more money.  A GPP contest is no different: the “market” is dictated by the overall ownership percentage in the field.  You will only make money if you can “beat the field”.1

[1 – Fun fact: Download the  .csv file from any particular Draftkings contest.  Sum up the field’s ownership percentages.  You should always get 800% (or very close to 800% since there are some dead entries).  This is because there are 8 total DK roster positions, each which contribute 100% to this total. (Think: if one player were 100% owned across the field, he’d take up 1 roster spot on EVERYONE’s LU)]

“Beating the field” in the DFS context means that the field has some hypothetical lineup where…

  • Its PG roster position is a hybrid player that is 22% Westbrook, 12% Curry, 11% Tyreke, 55% mix of other players…
  • Its SG roster position is 17% Bazemore, 16% Hardaway, etc…

And so forth, as I could keep going down the roster positions until all the field ownership %’s were used up (again, such that the percentages add up to 800%).  Side note: how awesome would that be to have Westbrook-Curry-Evans hybrid player… auto-lock!

If you cannot beat the field’s hypothetical lineup (and then some), then you will not make money in a GPP contest.2

[2 – Other fun fact… if you multiply each player’s (ownership % * fpts) for the entire $30K And-One contest last night and then sum them all up, you’d get a score of 231.4.  This is the median score of the contest (which is essentially what the hypothetical field’s LU would score).  This median score should fall right around the 50% mark of all LUs in that contest.  When I say you must “beat the field’s hypothetical lineup (and then some)” it’s because just beating 50% of the LUs in a GPP is not enough to cash.  The $30K And-One doesn’t start paying until you’re in the top 25% of the field, so therefore you must do a lot better than just the field’s median score.]

 

IV. Where did the winners invest?

This is the most interesting part of the analysis that hopefully answers the original question:  Why does ownership matter in GPPs?

To quintuple up your money in the $30K And-One GPP, you need to score in the top 2% of lineups (ie, place top 750 out of 35671).  If I isolate only the lineups that cashed in the top 2% of the contest, then we can compare how the ownership of the top 2% tier compares relative to the rest of the field.  The table below shows the most owned players that were in the top 2% of winning LUs from the $30K And-One.

You can see that Barton was the highest owned in the top 750 lineups at 71%, whereas the field had him at a chalky, yet still lower 36% owned.  You would say the top tier from this tournament was heavy on Barton relative to the field and due to his great performance, he was good chalk.  JaMychal Green was owned in 61% of the top 750 lineups and only 13% owned by the field.  The third column of the table is showing how ownership in the top 750 lineups compared relative to the field.  A player like Tim Hardaway Jr. was -4.44% from the top tier, meaning the top 750 lineups owned 4.44% less THJ than the overall field.

Again, draw whatever conclusions you’d like from this table, but things get more interesting as we move up the pay scale and ask how ownership spreads across for LU’s that earned 10x profit ($1 in, $10 out) in this contest.  To hit a 10x multiplier in the $30K And-One, you must finish in the top… (wait for it)… 100 out of 35671 lineups.  In other words, you must finish in the top 0.28% of the field!

 

V.  The top 0.28% (10x payout)

Here’s a table of the top 100 LUs (top 0.28% tier).

You will notice that at this higher tier (the LUs that made 10x on their entry), JaMychal Green becomes the more common thread.  Montrezl Harrell also becomes a more important piece to hitting this elite payout level.  Also, the top 100 lineups are much heavier on these players relative to the field.  So even though Barton is a massively chalky play at 36% of the field, the top LUs were much heavier on him relative to the field.

 

VI.  Winning the GPPs for all the monies (the top 3 lineups)

As we move further up the pay scale, we can look at what it took to be in the top 3 lineups.  The top 3 lineups of the $30K And-One earned the following scores and payouts in the table below:

What was the DFS ownership in these top 3 LU’s?

Some observations:

  • Barton is in all 3 LUs. Mitchell, Mills, Porzingis in 2 out of 3.
  • Each LU has 1 player under 2% owned.
  • Each LU has at least 3 plays under 10% owned
  • LU3 has a ton of pivots off high ownership with only 1 chalky play (Barton). LU1 and LU2 each have at least 4 players that were actually more than 10% owned in the field.
  • No expensive studs in any of these LU, with exception of Giannis ($10.3k) which was a “pay up to be contrarian” position with all the other studs on the board last night.

 

VII. Takeaways

  • Ownership matters in GPP (again, different story for cash games since cash games are the honey badger of ownership – cash game don’t care ‘bout no ownership!)
  • Payout in GPP is skewed such that you have to beat a LOT of people just to even double, triple, quint up. Thus it’s not a viable strategy to think you can be profitable in the long run ONLY playing GPPs.
  • You need to be willing to play at least a FEW low owned plays if you want to differentiate yourself in a large field GPP…
  • …HOWEVAA (@adotwalsh style), you don’t need to get too “cute” and ONLY take low owned players. It’s okay to play some chalk (see LU1 and LU2 on previous page), but just make sure you differentiate elsewhere.
  • Unless you think a player has massive upside (see GPP value sliding scale chart), then that’s when you can consider fading higher owned players.
  • When you see an Army coach say “I want the opposite of whatever the field does” in GPPs, now you should understand that if a player is going to be highly owned, you want to fade that player. If they’re going to be overlooked, you want to be heavy on that player.  In both cases, you’re employing game theory trying to “beat the field”.
  • Lastly, being “low owned” itself is not a cause for celebration in GPPs. You only have half of the magic sauce – the other part is that they actually need to exceed their value.  So don’t start celebrating at the start of a game just because you have someone that is 0.01% owned since maybe there’s a good reason he’s not on anyone’s radar.  Example:  OMG, I’ve got Kawhi at 0.01%… (1 hour later) – wait, why hasn’t Kawhi scored yet?  What the heck is quad tendinopathy?!?  (proceeds to spam slack on tilt)

– RJC