Nielsen protests ECU performance calculations

4/12/2011 – The 2011 Individual European Championship was also a qualification tournament, with the top 23 finishers getting a seat in the 2011 World Cup. Problem was that behind the first 15 players there were 29 with a tied score. So the performance was used to break the tie. However the system produced some bizarre results, as GM Peter Heine Nielsen explains.

ChessBase 14 Download ChessBase 14 Download

Everyone uses ChessBase, from the World Champion to the amateur next door. Start your personal success story with ChessBase 14 and enjoy your chess even more!


Along with the ChessBase 14 program you can access the Live Database of 8 million games, and receive three months of free ChesssBase Account Premium membership and all of our online apps! Have a look today!

More...

Summary

The final standings at the 2011 Individual European Championship in Aix-Les-Bains, France, produced the following table (full standings here):

Rk Ti. Name Nat Rtg Pts  TB1   TB2   TB3  Perf
Elo +/-
1 GM Potkin Vladimir RUS 2653 8.5 2849 63.5 78.0 2822
24.3
2 GM Wojtaszek Radoslaw POL 2711 8.5 2826 63.0 77.0 2812
14.5
3 GM Polgar Judit HUN 2686 8.5 2799 63.5 77.0 2781
13.4
4 GM Moiseenko Alexander UKR 2673 8.5 2755 62.0 74.5 2790
16.3
5 GM Vallejo Pons Francisco ESP 2707 8.0 2819 66.5 80.0 2764
8.2
6 GM Ragger Markus AUT 2614 8.0 2783 62.5 76.0 2768
22.8
7 GM Feller Sebastien FRA 2657 8.0 2766 58.5 70.5 2763
15.4
8 GM Svidler Peter RUS 2730 8.0 2751 62.5 76.5 2757
4.2
9 GM Mamedov Rauf AZE 2667 8.0 2751 61.0 74.0 2754
12.4
10 GM Vitiugov Nikita RUS 2720 8.0 2741 63.0 76.5 2744
3.9
11 GM Zhigalko Sergei BLR 2680 8.0 2732 59.5 72.0 2731
7.5
12 GM Jakovenko Dmitry RUS 2718 8.0 2719 60.0 72.5 2704
1.3
13 GM Korobov Anton UKR 2647 8.0 2697 61.5 75.0 2740
13.4
14 GM Inarkiev Ernesto RUS 2674 8.0 2695 60.0 72.5 2735
8.5
15 GM Postny Evgeny ISR 2585 8.0 2633 52.0 64.0 2676
14.0
16 GM Azarov Sergei BLR 2615 7.5 2776 62.5 75.0 2723
17.1
17 GM Khairullin Ildar RUS 2634 7.5 2771 61.5 74.5 2720
13.8
18 GM Kobalia Mikhail RUS 2672 7.5 2754 57.0 70.5 2716
7.0
19 GM Guliyev Namig AZE 2522 7.5 2739 59.5 71.0 2652
24.3
20 GM Zherebukh Yaroslav UKR 2560 7.5 2739 59.0 71.5 2712
25.1
21 GM Riazantsev Alexander RUS 2679 7.5 2728 60.0 72.5 2687
2.0
22 GM Iordachescu Viorel MDA 2626 7.5 2725 62.0 76.0 2716
14.2
23 GM Lupulescu Constantin ROU 2626 7.5 2722 58.0 71.0 2677
8.8
24 GM Mcshane Luke J ENG 2683 7.5 2718 59.0 72.5 2684
0.9
25 GM Fridman Daniel GER 2661 7.5 2717 56.5 69.0 2684
3.9
26 GM Motylev Alexander RUS 2677 7.5 2716 59.0 71.0 2710
5.4
27 GM Ivanisevic Ivan SRB 2617 7.5 2712 58.5 71.0 2704
14.1
28 GM Jobava Baadur GEO 2707 7.5 2711 58.0 71.5 2656
-3.3
29 GM Parligras Mircea-Emilian ROU 2598 7.5 2709 65.0 78.5 2735
21.7
30 GM Romanov Evgeny RUS 2624 7.5 2709 55.5 68.5 2668
7.6
31 GM Esen Baris TUR 2528 7.5 2707 61.0 73.0 2669
25.3
32 GM Nielsen Peter Heine DEN 2670 7.5 2703 55.0 67.5 2707
5.9
33 GM Cheparinov Ivan BUL 2664 7.5 2698 62.0 75.0 2693
4.8
34 GM Gustafsson Jan GER 2647 7.5 2687 55.0 67.0 2687
6.7
35 GM Kulaots Kaido EST 2601 7.5 2669 54.5 67.5 2633
6.0
36 GM Smirin Ilia ISR 2658 7.5 2668 56.5 69.0 2675
3.4
37 GM Saric Ivan CRO 2626 7.5 2651 58.5 72.5 2692
10.9
38 GM Pashikian Arman ARM 2642 7.5 2649 55.5 68.0 2640
0.6
39 GM Edouard Romain FRA 2600 7.5 2634 52.5 66.0 2602
2.4
40 GM Bologan Viktor MDA 2671 7.5 2629 56.0 68.5 2673
1.0
41 GM Beliavsky Alexander G SLO 2619 7.5 2627 57.5 70.5 2668
8.6
42 GM Rublevsky Sergei RUS 2678 7.5 2627 55.0 67.0 2672
0.1
43 GM Volkov Sergey RUS 2621 7.5 2625 54.5 67.0 2630
3.0
44 GM Sjugirov Sanan RUS 2643 7.5 2594 56.0 68.0 2584
-5.2

As you can see, behind the first fifteen players there are 29 vying for eight remaining places in the 2011 World Cup. All have 7.5/11 points, and to break the tie the ECU used performance ratings, but calculated them after eliminating the games against opponents with the highest and lowest Elo rating. This method of calculation leads to bizarre results, as Nielsen shows in his protest. For instance in his case he drew in the final round against a player with a 2626 rating. His EC performance rating was calculated to be 2703. Had he instead drawn against a player with a rating of 1200 his EC performance would have been 2717 and he would have qualified for the World Cup.

PH gives an even starker example of two fictive player who have the same rating and have played against the same eleven opponents, both scoring nine points against an average of 2500, having played one opponent rated 2400, nine rated 2500 and one rated 2600. However their EC 2011 Performance Rating can be strikingly different – in fact one could be rated as performing 538 points higher than the other.

Nielsen concludes that the examples are more than sufficient to prove that "the EC 2011 implementation does not match the standard of Performance Rating which is expected from the tournament regulations and does not calculate on the relevant number of games but leads to wrong, unfair and misleading tie-breaking results." He calls for the final standings to be changed, and if they aren't for the ECU to give him and other participants who have been harmed by the EC 2011 implementation of the Performance Rating proper compensation.


To the European Chess Union

As a participant of the 2011 European Individual Championship, I, Peter Heine Nielsen, hereby protest against the final standing.

Claim:

Primary: The Performance Rating for players with the same number of points in the 2011 European Individual Championship must be recalculated in such a way, that the highest and lowest opponent’s Elo rating is excluded when calculating the opponents’ average rating, but points achieved against these two players are included when calculating the percentage score. The final standing must be corrected in accordance to the recalculated tie-break.

Secondary: Proper compensation given to me and other participants of this ECU tournament, who have been harmed by the implementation of the Performance Rating.

The tie-breaking was not calculated from the actual score in the tournament but from only a selected number of games. As a consequence the crucial numbers defining the finishing position for players with equal score in points was in no way the Performance Ratings for the results in the tournament as stated in the regulations.

The scoring and tie-breaking system applied in the European Championship that took place March 21st – April 2nd in Aix-les-Bains is published in ECU regulations and described as follows:

6.1 The score in each game is 1 for a win, ½ for a draw and 0 for a lost game.
6.2 Tie-breaking in individual competitions.

The order of players that finish with the same number of points shall be determined by application of the following tie-breaking procedures in sequence, proceeding from (a) to (b) to (c) to (d) the extent required:

(a) Performance Rating
(b) Median-Buchholz 1, the highest number wins;
(c) Buchholz, the highest number wins;
(d) Number of wins, the highest number wins.

In case of (a) the highest and the lowest rated opponent will be deleted and the maximum rating difference of two players shall be 400 points.

In the case of unplayed games for the calculation of (a), (b) and (c) the current FIDE Tournament Rules shall be applied.

I argue that “In case of (a) the highest and the lowest rated opponent will be deleted” should be interpreted such as the rating of these opponents shall not be included when calculating the average rating of the opponents, but such as the points scored against them shall be included when calculating the procentual score.

The second part of the section “... and the maximum rating difference of two players shall be 400 points” is also open for two interpretations: Either such as a game against an opponent, where the difference exceeds 400 points is to be deleted before calculating both the average rating and the procentual score, or such as the opponent shall be corrected to the fictitious rating diference 400 points before calculating the average rating, while the result in the game counts when calculating the procentual score. It is worth noticing, that the interpretation of the first and the second part of the section goes into opposite directions in the implementation used for the 2011 European Individual Championship.

Implementation of tie-break rule 6.2 (a), by deleting the highest and the lowest rated player before calculating the average rating used for Performance Rating, would still make it a Performance Rating for the full tournament. And this is what I, and probably the ECU General Assembly, the ACP and most of my chess-playing colleagues expected from the text. However by deleting not only the rating of the opponent but also the obtained score against them, the calculated Performance Rating counts only the performance in nine selected games and can not - mathematically, legally or in any reasonable interpretation – be accepted as the tie-breaking relevant Performance Rating in the full tournament.

Using some examples I shall prove that the ECU Performance Rating in the way the regulation 6.2 (a) is interpreted by the tournament administrators is nothing else but FIDE Performance Rating for selected nine games, while counting two of the eleven valid games as not played. These examples will reveal that by deleting the results of the two games, the system of tie-break is changed so drastically, that it cannot be called Performance Rating for the tournament as stated by 6.2 (a).

Example 1:

Mircea-Emilian Parligras scored 7½ points in 11 games. In this example I compare the FIDE Performance Rating system with the system implemented in the 2011 European Individual Championship (below called EC 2011 “Performance” Rating).

 
FIDE
 
EC 2011
Round 1
2236 1
 
2236 1
Round 2
2462 1
 
2462 1
Round 3
2729 1
 
2729 1
Round 4
2672 ½
 
2672 ½
Round 5
2675 1
 
2675 1
Round 6
2480 ½
 
2480 ½
Round 7
2657 ½
 
2657 ½
Round 8
2664 1
 
2664 1
Round 9
2653 ½
 
2653 ½
Round 10
2680 ½
 
2680 ½
Round 11
2718 0
 
2718 0
 
7½/11 = 68.18%
Average rating: 2602
FIDE Performance = 2735
 
5½/9 = 61.11%
Average rating: 2629
EC 2011 “Performance” = 2709

Parligras’ FIDE Performance Rating is calculated as his opponents’ average rating for the whole tournament, plus the added bonus for his percentage score for the whole tournament. Parligras’ Performance for the full 11-round tournament is his opponents’ average rating 2602 plus 133 points added for his 68% score resulting in a FIDE Performance Rating of 2735.

His EC 2011 “Performance” Rating is calculated exactly as a FIDE Performance Rating, just with two results crossed out (in this case two victories!). It gives him the average rating of 2629 for the 9 games plus an 80 point bonus for a 61% score in these 9 games. Therefore his EC 2011 “Performance” Rating is 2709.

Conclusion: In reality EC 2011 “Performance” Rating just shows Parligras’ FIDE Performance Rating for 9 selected games in a 11 round tournament.

Example 2:

For a system to be described as Performance Rating it must have certain attributes:

A) As the most fundamental attribute it is obvious that two players with the same rating, playing the same opponents, making the same number of points, hold the same Performance Rating.

B) Lowering the rating of a specific opponent should never result in improving one’s Performance Rating.

However, as demonstrated below, the implementation of the system in the 2011 European Championship did not fulfil these mandatory criteria.

 
‘Player 1’
opp.rating score
‘Player 2’
opp.rating score
Round 1
2400 0
2500 0
Round 2
2500 1
2400 1
Round 3
2500 1
2500 1
Round 4
2500 1
2500 1
Round 5
2500 1
2500 1
Round 6
2500 1
2500 1
Round 7
2500 1
2500 1
Round 8
2500 1
2500 1
Round 9
2500 1
2500 1
Round 10
2500 1
2600 1
Round 11
2600 0
2500 0

‘Player 1’ and ‘Player 2’ have the same rating and both players have played 11 games against the very same 11 opponents, both scoring 9 points against an average of 2500, having played 1 opponent rated 2400, 9 rated 2500 and 1 rated 2600.

However their EC 2011 “Performance” Rating is strikingly different:

‘Player 1’ scoring 9 points in 9 games: EC 2011 “Performance” Rating = 3300
‘Player 2’ scoring 7 points in 9 games: EC 2011 “Performance” Rating = 2762

Conclusion: A difference of 538 points calculated for two absolutely identical tournament results proves that the EC 2011 interpretation of Rating Performance does not fulfil the most fundamental criteria (A) of a Performance Rating system.

B) Here I shall use the example from my own tournament:

Round 1
2385 1
 
2385 1
Round 2
2539 1
 
2539 1
Round 3
2584 0
 
2584 0
Round 4
2530 1
 
2530 1
Round 5
2593 ½
 
2593 ½
Round 6
2586 ½
 
2586 ½
Round 7
2571 ½
 
2571 ½
Round 8
2569 1
 
2569 1
Round 9
2602 1
 
2602 1
Round 10
2729 ½
 
2729 ½
Round 11
2626 ½
 
1200 ½
 
6/9 = 68.18%
Average rating: 2578
EC 2011 “Performance” = 2703
 
6½/9 = 72.22%
Average rating: 2551
EC 2011 “Performance” Rating = 2717

The first column is my actual tournament. In the second column I repeat the first 10 games in my tournament, but with my last round opponent replaced by a player rated as low as 1200. 1200 could be 2384 or anything in between as well.

Calculated according to the EC 2011 “Performance” Rating system, my Performance Rating rises 14 points (enough for qualification) by lowering my last round opponent’s rating from 2626 to 1200. This is due to the fact that despite scoring the same number of points in the overall tournament, the EC 2011 “Performance” Rating will be calculated with a higher percentage score, as now I have only 1 point removed from my score, not 1½.

Conclusion: Lowering the rating of a specific opponent improved the EC 2011 “Performance” Rating, which proves that the EC 2011 interpretation does not fulfil the fundamental criteria (B) to be an approvable Performance Rating system.

The examples showing absurd results as a consequence of using the EC 2011 “Performance” Rating are numerous. One of the most extreme cases is that of Paligras. The reason he did not qualify for the World Cup, was not because he lost the final round, but because he lost it to a high rated opponent! He was paired against Jakovenko rated 2719. In his situation the solution was to default the game or violate the zero tolerance rule. Then the game would have been counted as his lowest rated, and Parligras would have improved his EC 2011 “Performance” Rating more than enough to qualify for the World Cup!

The above given examples 1) and 2) are more than sufficient to prove, that the EC 2011 implementation does not match the standard of Performance Rating which is expected from the tournament regulations and does not calculate on the relevant number of games but leads to wrong, unfair and misleading tie-breaking results.

I therefore claim that: primarily the final standings of the 2011 European Championship to be based on a fair tie-break according to the regulations interpreted in accordance with what I wrote at the bottom of page 1, secondarily call for proper compensation given to me and other participants of this ECU tournament who have been harmed by the EC 2011 implementation of the Performance Rating.

For further comments on intentions, correspondence, interpretentions and timing please read the Enclosure.

Århus, the 10th of April 2011

Peter Heine Nielsen

Links


Discussion and Feedback Join the public discussion or submit your feedback to the editors


Discuss

Rules for reader comments

 
 

Not registered yet? Register