Author: Stephen A. Boak
Date: 19:53:36 08/19/99
Go up one level in this thread
On August 19, 1999 at 11:07:08, KarinsDad wrote:
>On August 19, 1999 at 00:06:02, Stephen A. Boak wrote:
>
>>On August 18, 1999 at 23:55:09, James Robertson wrote:
>>
>>>On August 18, 1999 at 23:51:56, Stephen A. Boak wrote:
>>>
>>>>I calculated the Tournament Performance Rating (TPR) of each FIDE KO
>>>>semi-finalist (also Polgar) to date: polgar 2692, adams 2744, akopian 2904,
>>>>khalifman 2769, nisipeanu 2837
>>>>
>>>>These figures give you an approximate idea of what TPR a computer would have to
>>>>score in a similar KO tournament to get into the final round.
>>>>
>>>> --Steve Boak
>>>
>>>Could you add up the average ELO faced by each player? I am bummed that Kramnik
>>>lost, and I think that the stupid pairings caused his early exit while players
>>>like Akopian surfed in with relatively easy opponents.
>>>
>>>James
>>
>>Clarifications:
>>1. I am using the FIDE posted (KO website info) start ratings
>>2. Only standard time control games are rated in my TPR calculation
>>3. Calculations only through the Round 6 (i.e. second standard games of
>>Khalifman-Nisipeanu & Adams-Akopian):
>>
>> Polgar Adams Akopian Khalif. Nisipeanu
>>Player 2671 2708 2646 2628 2584
>>Ave Opp 2592 2664 2637 2636 2637
>>TPR 2692 2744 2904 2769 2837
>>
>>It is remarkable, isn't it, that Akopian, Khalifman, and Nisipeanu all have ave
>>opponent ratings (full 6 rounds--all three players) within only 1 rating point!
>>
>> --Steve Boak
>
>This is doubly misleading data. Akopian, Khalifman, and Nisipeanu played round 1
>whereas Polgar and Adams did not. This lowers the average rating of their
>opponents (i.e. the ratings should level out a little more once you use round 2+
>data only). Also, the rating posted at the web site are January 1999 ratings.
>The July 1999 rating came out and should they are more up to date.
>
>Could you please redo this showing both with and without round 1 using the July
>ratings?
>
>Thanks,
>
>KarinsDad :)
Dear KarinsDad,
'doubly misleading data?' hmmm. [ no offense taken. :) ] Where did I
mislead, I am interested to know.
1. Please clarify the first misleading aspect. [Polgar ave opp rtg?]
2. Please clarify the second misleading aspect. [Adams ave opp rtg?]
3. Logic check, please!
If I leave out the first round, in which only three players of my above list
played, A) the ave opponent rating for those three will increase (as you
suggest), and B) the gap between their opponents's averages will increase (not
decrease as you imply when you said the figures will level out a bit when using
only Rounds 2 and up) over the opponents's averages of Polgar and Kramnik.
Why? A) Because their Round 1 opponents were significantly lower rated (only
slightly over 2500) than their average opponent as of the time I calculated
those ratings; and, more importantly, B) because if you look closely you will
see that the average opponent rating for each of the group of three players
(using Round 1 opponent ratings) is already *higher* than for either of Polgar
or Adams. Leaving out Round 1 opponents will *increase* (i.e. the reverse of
'level out') the average opponent rating for the group of three vis a vis ave
opp rtgs of Polgar and Adams.
Do you agree?
Were you trying to make a different point?
4. Contrary to popular disbelief, I *do* know how to calculate 'average'. :)
In the case of Polgar I added the ratings of her 4 opponents and divided by 4.
In the case of Adams I added the ratings of his 5 opponents and divided by 5.
In the case of the other players, I added the ratings of 6 opponents and divided
by 6.
An average is one highly popular method of 'normalizing' data for comparative
purposes. Average may be calculated without regard to how many things are
included in the average, and without regard to the variation in relative sizes
of the individual figures used in the calculation.
5. I don't mind checking my figures. I like solving math problems and puzzles
for a hobby, programming a computer for personal challenge, and I enjoy
practical uses of statistics. In these and other endeavors, for some reason, a
right answer satisfies more than a wrong answer.
6. Don't stop commenting on any of my posts--I do make errors and typos from
time to time, both in my number work and in my postings (and in my brain!).
7. Do you still want me to run some figures another way? I am willing if you
please clarify the above for me. Let me know, thanks.
--Steve Boak, the unintentional misleader
(and I always wanted to be a leader!) :)
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.