PST S28 Draft Review
If you crunch the numbers for all the signups this season you get an MMR average of 4057 and a standard deviation of 1313. In the effort of getting someone other than math goblins to read my blog this will be (almost) the last time I use numbers and will instead use pretty picture. This is the actual distribution of players this season, sorted into ‘buckets’ of a fairly wide MMR on the X axis and then the number of players in each bucket on the Y axis.
This is more or less the same shape as the theoretical graph generated from the same numbers.
The neat part about this graph is that it is much easier to do math on. For instance, if you were to randomly select groups of 5 from the pool and then make a distribution of the averages of those groups it would look like this:
Now here is the same graph but with the location of the actual highest and lowest player marked by vertical blue lines and team marked by vertical green lines.
Zoomed in on the top of the green curve, representing the distribution of our teams:
For any distribution like this we would expect the higher the point on the curve, the more often we will find teams at that MMR. We would also expect this distribution to be pretty tightly clustered around the center, though this graph is representative of random teams, which RD2L teams are not, and I would expect the actual distribution to be even tighter than this. This is still a good jumping off point, so I’ve shown a selection of the graph along the actual MMR range of the teams this season. This range between the highest and lowest MMR team is a bit low, historically speaking, at 490. It also seems that this is at least somewhat accurate model of how the teams actual shook out, since the actual distribution of the teams this season looks like this:
Which is the shape you would expect it to be given the range on the above curve. My point with going through this is to say that this distribution is not a bad thing, and we should expect, and always will have, teams that are above and below the average MMR of the league. No one is being unfairly advantaged or disadvantaged just because a team happens to be above or below the average MMR at the end of draft.
In the more anecdotal sense, despite all of hobos complaining last season about being poor at the start of draft, his team was significantly above the average MMR (this cost him in other ways, but were he to have drafted with less greed for MMR he could have made a team instead of a pipe bomb). This season the high MMR yfu and gronaldo chose to go for pretty balanced team compositions, though on reflection of the draft both of them could have chose to invest more money into a higher MMR player at the expensive of a lower one (or in gronaldos case no expense at all since he didn't spend all of his money) and had a team more in line with the old EU draft. Additonally while I was writing this vuvu and buck both published blogs where vuvu said 2ks never auction draft good teams and buck said that gronaldo was hamstrung and that seems like fairness to me :)
