I wanted to share a quick look at some ownership data for top 10 GPP finishing teams. My original question was, “what does ownership look like for GPP winning teams and how does that vary by buy-in level?” I was wondering this because many lineup builders are good a spitting out as many lineups as you want based on points, or ratings, or ceilings, but they aren’t very good at designing specific rosters based on specific ownership levels. This requires a bit more manual work, which is fine, but if I’m building rosters based on ownership, I definitely wanted to spend some time looking at what ownership looks like for GPP winning lineups. The common wisdom for GPP is that you need a contrarian lineup – and this is definitely true – but what does this look like on a team by team basis and how do you construct a roster based on that info?
A quick note on the data: I used the contest dashboard on fantasylabs to gather this data. I’ve been a pro subscriber since 2015, and I love the product. One annoying thing is that you can’t export or copy/paste data from anywhere on the site. While this is understandable from a proprietary data point of view, it’s makes larger analysis by “outside” people like me difficult. Because of this I only looked at NBA data on slates where they ran large tournaments from 11/1/17 to today (18 slates with at least 5 games, 55 separate contests). For each slate, I looked at the top 10 winning GPP lineups for the low stakes tourney (mostly $8), Mid stakes (mostly $33) and high stakes (mostly $555) and manually recorded the average ownership, player with the highest ownership and player with the lowest ownership for each of these top 10 teams in each of the contests (about 1,650 different data points). Also, I rounded ownership (someone 16.15% owned became 16%)…sorry. Average ownership looks at the 8 guys on the roster and just takes an average of their ownership level for that contest, high ownership is the chalkiest guy on the roster, and low ownership is the least owned person.
Here are a few highlights of the results:
-You don’t really need to fade chalk in GPPs. The average “high ownership” guy is 47.3% owned across all contests – that’s huge. I would feel comfortable playing chalk, I’m not really going to worry about it.
-More important is the low owed guy. Across all contests, the average low owned guy was 4.0% owned. This was an extremely consistent data point. You basically need at least one very low owned guy if you want a top 10 GPP winning lineup. The mode and median here was 2% and 3% – If you look at your lineup and you don’t have a sub 5-7% owned guy, it going to be tough to win.
-Average ownership is 20.4% – this value by itself is not very important. Targeting 20% ownership across your 8 guys is a recipe for failure, instead you need to think about roster construction from the point of view of hitting each end of the spectrum and balancing your chalky plays with a few more off the board plays.
Ok, so how does this break out by buy in level and tournament field size? It’s generally how you would expect. The averages at low stakes large field tourneys are depressed (15.3% average, 35.6% high, 3.3% low), similar to the overall at the mid stakes (21.0% avg, 48.6% high, 4.1% low) and higher at the high stakes (24.6% avg, 57.1% high and 4.6% low). Smaller stakes means you need to be a bit more adventurous with your low owned players.
I fully realize that this isn’t any groundbreaking stuff. So how can we make this actionable from a roster construction standpoint? I think in two ways. First is with multi-lineup generators. Setting exposure vs. the field is great, but you need to take it one step further. I would love to see a toggle built into my lineup generators that would force one (or two or three guys) into each roster who are projected to be low owned. With fantasylabs, I can create a model that includes ownership as a factor, and this is great, but I can’t construct specific lineups the way I would like to. Right now, I have to go through and manually adjust my rosters. As a guy with a family and a job, this is tough to do with 150 lineups on a Monday night slate. To me, tools like the lineup generator and ownership projections at labs and rotogrinders are essential, but they probably have room for improvement.
Secondly, if I put together a single GPP lineup and my low owned guy is projected to be 10 or 12% owned (or even 6-9%), I’m going to go back a rework that lineup to make sure that I have a guy or two at a very low ownership level based on the field size of the tournament that I am entering.
Hopefully that’s a little helpful to people – it was to me. At least I now when I look at my lineups pre-lock, I’ll have a better idea at determining if I even have a chance to win before all the snowflakes start showing up.