| |
BGonline.org Forums
Win Rate as a function of PR difference - XG Feature
Posted By: AdamStocks In Response To: Win Rate as a function of PR difference - XG Feature (Maik Stiebler)
Date: Saturday, 22 August 2015, at 12:47 a.m.
I'm not sure which part(s) you are having trouble with, but perhaps my introducing the term 'inflation' was a confusing and unnecessary digression on my part, as my snapshot exercise does not need to include rating inflation as such, because inflation happens with respect to a time interval rather than a point in time (my snapshot).
The essence of my thread was simply this: For simplicity let's assume that our system starts with a fixed pool of players, and no additional players join. With no ramp, the average correct rating of the best and worst players will, over a given time, respectively approach some limits (max and min), giving a rating range of (max-min). With a ramp, max and min are pushed further apart in the same timeframe, giving a wider rating range even though the PR range of the player pool can be assumed to be fixed. That makes each 1PR difference worth more rating points than without a ramp, hence the comparison of 39.2 versus 33.
The article ideas you quoted seam perfectly reasonable, so you're not wrong, but I think you got a little sidetracked (as I do lol) because my snapshot isn't concerned with new players joining or whether they are under/over rated etc. However, to address your question, if a strong new player joins and is ramped, he will on average be the beneficiary of extra points over and above normal, whilst his (un-ramped) opponents will on average lose only un-ramped amounts of points, which causes inflation, not deflation.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.