| |
BGonline.org Forums
PR, ELO, outplays, etc.....
Posted By: neilkaz In Response To: XG PR should be the new standard (Phil Simborg)
Date: Wednesday, 16 December 2009, at 6:16 p.m.
Xavier's concept of outplays is an interesting one and it is nice to see when you've outplayed 3 ply. Initially, XG only reported how well you'd played the match based on equivilent ELO, which set XG 3 ply do be 2240. For equivilent ELO I have no problem with outplays being on and noting that if you play better than 3 ply (good luck long term !) then having a higher ELO is fine with me.
However, the BG community,both not being familiar with XG Elo and wanting something similar to Snowie ER, convinced Xavier to come up with PR.
I am very happy to see PR, but note that the BG community when using Snowie ER, often does RO's to try to determine the ultimate truth. Obviously, RO's are going to result in more accuracy than 3 ply evals, and this will mean outplays. However, the BG community wishes to compare to the ultimate truth when using PR or ER and it should not be possible to do better than playing at a 0.0. Playing at a 0.0 would be perfect play (well as close to perfect as our RO's could guarantee)
With this in mind, I strongly recommend that, when selecting PR, that outplays (error compared to XG) be turned off. If you want to see outplays, run the initial analysis with outplays turned on, and then turn them off to look at PR.
What I would really like to see is for the switch to select or not select error compared to XG remain as is and then you can see outplays and include them in ELO if you wish or not, but for outplays to automatically be NOT included when PR is selected!
If this means that XG's PR's are now consistanly higher than GNU's reported Snowie ER's, the BG community can adapt, or Xavier can lower the multiplier of 500 a bit.
.. neilkaz ..
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.