Well it was another exhilarating week if you chose to enter all 5,860 lineups that we offered for the DraftKings Millionaire Maker. And by exhilarating I mean that you would have booked about a $60,000 loss if you’d played every lineup. BUT. They still would have outperformed the field by average. So despite the soul-crushing-jump-off-a-balcony-in-Vegas $60k loss, I’m calling it a win. I’m an optimist at heart and your attachment to money and material things won’t get me down.
Let’s look at the distribution of the RotoViz scores versus the field. The red line is the distribution of RotoViz scores. The dashed line is the score of the top projected lineup, which would have beat 75% of the field (not good enough to cash).
The best score of any lineup would have been 209 and change, or roughly about 52nd place.
|3972||209.52||DEMARYIUS THOMAS – LEVEON BELL – ANDRE ELLINGTON – RUSSELL WILSON – EMMANUEL SANDERS – ANDRE JOHNSON – RODDY WHITE – LARRY DONNELL – BUFFALO BILLS|
Here’s what the distribution looks like just for the top 50 projected lineups. Note that the top projected lineup was only about the median lineup of the top 50 when it came to actual results.
Last week I said that we needed more lineups and more diversity of lineups. I came through on the more lineups part by giving you 2,000 more than the week before. But I didn’t do anything to address the diversity issue. In fact I know there isn’t enough variance in these lineups because I ran some simulations. I used the mean and standard deviation of the RotoViz lineups to simulate the results as if we had 72,000 lineups (basically as if we’d just taken every entry in the contest) instead of the 5,860 we published (and assuming that the additional lineups that we generated would have had the same standard deviation and mean, which might be a stretch). But even if you simulate the 72,000 lineups a number times the top score doesn’t get out of the 220s very often and it would have taken 242 points to beat rayofhope this week.
So I think what the lineups are doing well is establishing a high floor. What they’re not doing very well is establishing a high ceiling because they don’t include enough variance. Note that I re-ran the simulations above by limiting the standard deviation and mean to just rosters that included at least one stack (players of same team) and no more than two RBs. That did increase both the standard deviation and mean a little, which is good. But it wasn’t enough to get the top scores of those simulations up into the range that would have been needed to win.
Again, the offer stands for anyone who entered all 5,860 lineups – free RotoViz subscription for you.
EDIT: Just for fun I went and looked at the standard deviation and mean of rayofhope’s entries since he had 64 entries in the Milly Maker. His standard deviation was 28.5 and mean was 143.2. That’s better than the total population of RotoViz lineups on both counts. The mean score of the RotoViz lineups was 133 with a standard deviation of 22. Interestingly our top 50 lineups had a mean of 145, but couldn’t come close to rayofhope’s standard deviation (28 for rayofhope, just 16 for our top 50 lineups).