Flaw in PR / Is this wrong?
Posted By: eXtreme Gammon In Response To: Flaw in PR / Is this wrong? (Rod)
Date: Thursday, 19 December 2013, at 6:37 p.m.
In Response To: Flaw in PR / Is this wrong? (Rod)
Rob:d The real question is "do you believe that my expected result over time is what XG says it is, given my play" against that player? The answer to that is "no", for sure.
The idea behind dropping these cubes is that it reduce the skill influence. If you were right then the PR of the your opponent on the match will be much higher than if you'd taken. In other words the difference of MWC errors should be same if the decision to take was borderline against the perceived value of the opponent.
So, I would say that the "non adjusted" PR difference does reflect your expect gain. If you ignore the cost of your deliberate error, you are double dipping: the difference of skill take into account the extra error your opponent made without counting your MWC error investment/gamble. The difference of skill will >u>not reflect your expected gain.
Now, about the repeated non-double follow by 6 dances... This is a good point, would it had happened twice more your PR would have been even worse, this does not seems normal. Maybe the "number of decisions" used for PR should be "Number of unique decision". The case is rare (and short of dance/dance, extremely rare)
I will not comment on the choice of some to "ignore" their error when they make them on purpose, however I would agree in that case that you should count that error once only and ignore the 2 others.
To finish, I am surprised UBK did not reply yet to that thread. He would have said (and I agree with him) that Variance reduction is the best way to take into account deliberate errors as it does not require any cherry picking of what errors to account from.
Messages In This Thread
BGonline.org Forums is maintained by Stick with WebBBS 5.12.