Introduction The purpose of this study is to find out how the strength of play has risen in the course

of time. FIDE rating of 2008 serves as a basis of comparison. There exist two types of analyses: those that comprise players, and those that comprise certain time periods. The advantage of time period-based analyses lies in the fact that their results are not affected by style of play so much as it could sever the credibility of results. Tactical play by humans often contains moves that objectively are not optimal, but give good chances in practical play. If games are chosen randomly, the probability of either wild, or calmer and positional games happening to be within a selection is equal. Thus decades from 1860s till 2000s are taken as an object in the study. To ensure the coherence of data, players from the chessmetrics.com rating interval 2600-2700 were taken into consideration. The average rating stays therefore always near 2650. For the sake of interest, Carlsen's games from Nanking 2009 are also included where his performance rating was even 3002. It has been often described as the most impressive display of chess skill ever by human. Later on Fischer vs Larsen and Taimanov, and Karpov in Linares 1994 will be analyzed as well. Some theoretical principles. Every computerized analysis whose aim is to ascertain the playing strength of a player in comparison with today's FIDE-rated chessplayers, has to consider following theoretical principles: 1. All moves must be taken under scrutiny as long as one of the players does not have a won position, except opening moves. It's crucially important that all moves after leaving theory are analyzed, provided the position is more or less equal. If one picks an arbitrary selection of moves from a game, the whole picture will not be seen like in the case of all other measurings in the world. As long as the position remains more or less equal, a player supposedly mostly retains the objective and neutral opinion. Things are different when the position inclines decisively towards one or other side. For example, if a player is in a clearly lost position, there would be no reason to expect him to make every effort due to psychological reasons, or he would resort to swindling, or just go into complete passivity. In a won position pursuit of accuracy, on the other hand, is not the most essential, if there exist other, more natural for humans, ways leading to victory. It must be admitted, though, that in the case of data size being large enough it is not so important, as it would be cancelled out statistically. 2. It must be considered that time control may be different by games, whereas in earlier times they were longer than nowadays. Time controls in tournaments and matches have become shorter in the course of time, and due to the spread of computers, adjournments have disappeared. One of reasons must be making chess more attractive to the audience. Shorter time controls also decrease the number of number of draws via the lessened quality of play. 3. Longer time control leads to more accurate play. By doubling thinking time, the playing strength of humans increases further than computers (cs 5070 ELO). Also, the increase in accuracy is sharper at faster time controls. 4. Positions in a game of chess are of different degrees of difficulty that affect the accuracy of play. Generally, positions occurring in a game can be classified into two categories: tactical and positional/strategical. The more tactical elements there are in a position, the more difficult it is for a

human to maintain the accuracy of play. I prefer the terms 'higher accuracy of play' or 'lower accuracy of play' over 'easier/harder to find best moves' because in more than a half of positions the notion of 'best move' matters extremely little. It is possible to determine the difficulty of position with the help of various parameters whose sum total makes up the general factor of difficulty. 5. The level of play or the rating of a player is indicated by what his expected average error is at a certain degree of difficulty. The rating of a player is a reflection of his playing skills in a numerical way. Depending on the style of play, the expected average error of a player of the same level of play can be slightly different with respect to the extremes of the degree of difficulty, but the same altogether. The corresponding indicator of positionally playing chessplayer in simpler and more strategical positions is higher than that of a equistrong tactical player. The accuracy of play of a tactical player, nonetheless, is relativelt better in more difficult and tactical positions. 6. The expected error is a function of the actual error and the difficulty factor:
f expected error= actual error difficulty

The general difficulty factor is equal to the joint effect of all separate difficulty factors. How much a factor influences the accuracy of play can be determined from measuring how much relatively the actual error changes within an equal sized amount of positions. 7. The outcome of a game of chess is determined by the actual accuracy of play. Since the difficulty of positions is also influenced by moves players make, generally more threatening and aggressive moves tend to highen the difficulty for the opponent, it is theoretically possible that both equally good players with a similar style of play are playing at equal strength, but one of them loses due to his positions being more difficult. It demonstrates how important it is to put the opponent under pressure, to strive for the initiative. It makes move-choosing more complicated. The methods of analysis used in this study. 4 phases can be distinguished: 1. Find out the actual average error. It is calculated by taking the eval of best move by Rybka and substracting the evaluation of the move made by a player from it. 2. Next find out various difficulty parameters, including thinking time. 3. Derive the average expected error on the basis of difficulty factors and actual average error. 4. Compare the results with that of the modern FIDE-rated players to get an overview of the playing level of players being analyzed. In determining the average actual error Rybka 3 Default with default settings has been used. The GUI was Arena 2.0. Each position was analyzed for 5 minutes with 5 simultaneous PV-s on dual 2GHz Athlon. Since the amount of data was relatively average actual error
Carlsen Nanking 2009 1870s 2700 2400 1900s 1880s 2600 1860s 2500 1890s 2200 2300 2000 2100 0,000 0,200 0,063 0,119 0,122 0,129 0,135 0,146 0,147 0,172 0,181 0,184 0,185 0,211 0,225 0,228

scarce, stabilizing points have been used, otherwise the conclusions could have been misleading. The largest mistake possible is 2.00. In the case of large mistakes, it doesn't matter how large it exactly was, but the fact that a mistake was committed. An example can be instanced where player A in a game makes 5 mistakes 2.01; 2.68; 1.94, 2.08 and 2.44, but player B only one mistake withe the value of 13.34. If I didn't use the boundary value, we would arrive at the conclusion that both players had played at a similar level of accuracy, which would be obviously misleading. 2.00 is chosen arbitrarily and is not actually more valid than any other similar value. Only positions during which the situation of the game stays more or less equal are considered. Thus, positions where both the evaluation of move proposed by computer and the evaluation of move actually played are outside of the evaluation interval [2; -2] and are with the same sign, have been removed. Games whose blunder percentage exceeds 15% were removed, except when the number of valid moves is under 30.

There is no point in including opening moves if purely chess skills are being analyzed, not how much theory one has learnt by heart. Because opening theory is advancing every year, the start point where the analysis begins depends on a time-period as shown in the following table: 1860-1879 1880-1899 1900-1919 1920-1939 1940-1959 1960-1979 1980-1999 2000s 8 9 10 11 12 13 14 15

The minimal length of games is 20 moves + the start point of the analysis. So, it ranges between 2835 according to period. Since Rybka 3 is quite untrustworthy in properly evaluating endgames, positions with less than 10 pieces or pawns have been left out of consideration. An exception has been made for mistakes that really turn the situation around: a won position to a drawn or a lost one and vice versa. What actual average error indicates in the ideal situation is how much the accuracy of play deviates from the absolute point of view of perfect play. As mentioned previously, exactly this criterion plays a determining role in the outcome of a game; the fate of a game depends on to what extent one can bring his divergence from the absolute to a minimum. blunder percentage Unfortunately, a serious matter acts as an obstacle: how easily one Carlsen Nanking 2009 0,74% is able to lower his error rate depends not only on his playing 1880s 1,81% skills, but also on factors that make maintaining one's accuracy of 1870s 1,89% 1900s 2,23% play harder. This is why it is necessary to take those factors into 2400 2,52% account in the analysis. The more of them, the more trustworthy 2700 2,97% 2600 3,31% the calculation of the expected error will be. For the sake of interest, the frequency of blunders is displayed here. A blunder is a mistake valued 1.00 or more, but not over 2.00.
1890s 1860s 2500 2200 2300 2100 2000 3,90% 4,01% 4,85% 5,92% 5,97% 6,13% 6,55%

There are two types of difficulty factors: external and internal ones. External factors are thinking time, psychological pressure as a 0,00% 5,00% 10,00% result of some external influences, feeling, conditions at playing venue etc. Only thinking time can be measured and considered. Internal factors develop in the course of game, and are dependent on the placement of pawns and pieces that in turn depends on the style of play and openings. Unlike external ones, internal factors are also valuable indetermining the playing style of chessplayers. Internal factors can be classified into two groups: direct (complexity, difference) and indirect ones (material, evaluation).

The difference between direct and indirect factors lies in the fact that direct factors determine the the tacticality or the positionality of a position directly, indirect ones do not. Higher complexity and difference occur chiefly in tactical positions, a large amount of material on the board and a high evaluation do not reveal anything about the type of positions. A short description of the five difficulty parameters usen here in the study is as follows: complexity 1. Complexity. Complexity shows for what extent a position is tactically complicated, irrational and unclear where concrete calculation has greater prominence than positional evaluation. A high complexity can occur also in seemingly calm positions but where, however, some surprising and hard-to-see tactical shots are found. Another engine is used - Stockfish 1.4, because its evaluation figures are larger than Rybka 3, enabling more fine-grained distinguishing. Complexity is calculated this way: all instances of Stockfish changing its best move at each depth from 2 to 15 are determined. 0,000 0,500 After that all differences between first and second best move at each point of move change are added together. One has to take into account the fact that changes at greater depths are more difficult to see for players, so all differences within the interval 10-15ply were further multiplied by two.
1880s Carlsen Nanking 2009 1860s 2000 2300 1900s 1890s 2700 2100 2600 1870s 2200 2400 2500 0,560 0,546 0,530 0,519 0,507 0,504 0,500 0,483 0,482 0,458 0,458 0,455 0,435 0,426

difference 2. Difference 2500 0,555 2000 0,504 Difference is simply the evaluation difference between the best 1890s 0,504 and the second best move. Since all the players analyzed here 2200 0,488 generally play on a quite high level and do not make oversights so 2700 0,487 2600 0,472 easily, the maximal difference is set at 3.00 and all higher values 2300 0,460 are regarded as 3.00. Positions where best moves are forced have 2100 0,440 greater difference. It shows how important it is for a player to 2400 0,417 1860s 0,417 choose best moves. It may seem a bit strange that higher 1900s 0,391 difference is in correlation with lower accuracy of play, whereas Carlsen Nanking 2009 0,369 the probability of choosing the best move increases. But that's 1870s 0,343 1880s 0,338 precisely why chess is interesting, very often the best move is not obvious and clearly distinguished at all. In more peaceful and 0,000 0,500 strategical positions difference factor is relatively small, and enabling a player to choose among many more-or-less equal moves. Although it has not been considered here, a smaller difference is presumably in correlation evaluation with how moany PV-s are within a certain evaluation window. 3. Evaluation Evaluation describes how far a position has drifted from the complete equilibrium towards a decisive result. It is the same as the absolute value of the evaluation of an engine on the best variation. Also, it shows indirectly how tactical a position is, since the frequency of evaluation jumps in messed-up positions is certainly bigger either as a result of participants committing more errors or engines having troubles in orienting tin a thicket of variations. Secondly, in positions where evaluations are higher, deviations tend to have bigger values due to bigger proportions even if other factors remain same.
2300 2100 2000 2700 2500 1890s Carlsen Nanking 2009 2400 1880s 2200 1900s 1860s 2600 1870s 0,000 0,934 0,758 0,727 0,716 0,680 0,674 0,612 0,550 0,538 0,507 0,497 0,435 0,401 0,394 0,500 1,000

4. Material The system devised by Larry Kaufman with the help of computer was used in describing the quantity of material. Queen 9.75, rook 5, bishop and knight 3.25; bishop pair adds 0.5. Material describes the 'endgameness' of a position. In general, there is a tendency for the accuracy of play to be greater towards material diminishing.

material
1860s 2000 1900s 1890s 1880s 2100 2200 1870s 2600 Carlsen Nanking 2009 2500 2700 2400 2300 30,95 29,42 28,64 28,06 28,01 27,58 26,91 26,43 26,06 24,56 24,34 23,95 23,78 22,67 40,00

5. Thinking time There are numerous sites on the internet where information on time controls in various times at various tournaments and matches are available. In case no data on a particular event was available, estimates given by Encyclopaedia Britannica were used: 18800,00 20,00 1925 4 min; 1926-1945 3 min 20 s; 1946- 1985 3 min 45 s; 1986-... 3 min. One of the most annoying problems were adjourned games. How to measure them? There is no some thinking time uniform criterion, in earlier times there even was an unwritten Carlsen Nanking 2009 134 2000 144 statute that adjourned games were not analyzed. It is known that 2300 147 soviet masters used to help each other in analyzing them, 2600 153 especially against foreign players. In any case, it seems 2500 153 2200 158 impossible to find a decent solution, so I included an arbitrary 1 2400 164 hour to every adjourned game. 2100 164 Next, one of the most difficult problems needed to be solved: how changes in the thinking time affect the accuracy of play? Computer's playing strength increases by 50-70 ELO each time thinking time is doubled, slight diminishing returns can also be observed. In case of humans, these phenomena are more intensified. One possibility was to use clock simuls, as displayed below:
player 1 Kasparov 2 Kasparov 3 Kasimdzhanov 4 Kramnik 5 Tal 6 Fischer 7 Kasparov 8 Kasparov 9 Kasparov opposition Czech team Czech team Uzbek team German team Warsaw team Greek team German team Israeli team Israeli team year 2001 2001 2007 2004 1966 1968 1992 1998 1998 average
2700 1890s 1880s 1870s 1900s 1860s 0 167

262 264 265 265 342 100 200 300 400

time odds opposition performance player gap 3,3 2596 2609 2860 -251 3,3 2596 2621 2860 -239 5,0 2408 2446 2682 -236 4,0 2604 2617 2765 -148 8,0 2378 2441 2763 -322 5,0 2348 2398 2795 -397 4,0 2632 2657 2865 -208 4,0 2620 2658 2854 -196 4,0 2620 2658 2854 -196 4,511 2534 2567 2811 -243,7

The method consists of comparing the performance rating of a player giving a simul to his own actual rating and dividing it by mean time odds. In current case it appears that 4.5-fold time odds causes a player to lose his playing strength by ca 250 ELO points. according to the following graph, doubling thinking time means a rise of playing level by about 110 ELO.

6,000 5,000 4,000 time odds 3,000 2,000 1,000 0,000 -285 -255 -225 -195 -165 -135 -105 -75 -45 -15 15 45 75 105 135 165 195 225 255 285 -300 -270 -240 -210 -180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180 210 240 270 300 ELO

After determining the difficulty factors of every cohort, the average expected error has to be calculated on the basis of them. It shows hypothetically what the accuracy of play of chessplayers would be if each had identical types of positions and time controls. The average expected error is a function of playing accuracy in respect to all difficulty factors. Depending on the style of play and the nature of positions, the curve may be flatter or steeper. The accuracy of play of tactical players in positions that are more complicated and more familiar to them is relatively better compared to players inclined towards positional play. But it is lower in positions requiring strategical thinking and and generally in endgames. Their curve of the expected error is therefore flatter. On the other hand, the curve of positional players is steeper. The influence of each difficulty factor on actual error is individual. It is strongest in the case of the difference, and the lowest in the case of the material. complexity vs actual error
0,91 0,84 0,77 0,7 0,63 0,56 0,49 0,42 0,35 0,28 0,21 0,14 0,07 0 R = 0,14

actual error

0,04 0,12 0,20 0,28 0,36 0,44 0,52 0,60 0,68 0,76 0,84 0,92 1,00 1,08 1,16 1,24 1,32 1,40 1,48 1,56 0,00 0,08 0,16 0,24 0,32 0,40 0,48 0,56 0,64 0,72 0,80 0,88 0,96 1,04 1,12 1,20 1,28 1,36 1,44 1,52 1,60 complexity

difference vs actual error

0,91 0,84 0,77 0,7 0,63 0,56 0,49 0,42 0,35 0,28 0,21 0,14 0,07 0 R = 0,5

actual error

0,06 0,18 0,30 0,42 0,54 0,66 0,78 0,90 1,02 1,14 1,26 1,38 1,50 1,62 1,74 1,86 1,98 2,10 2,22 2,34 2,46 2,58 2,70 2,82 2,94 0,00 0,12 0,24 0,36 0,48 0,60 0,72 0,84 0,96 1,08 1,20 1,32 1,44 1,56 1,68 1,80 1,92 2,04 2,16 2,28 2,40 2,52 2,64 2,76 2,88 dif f erence

evaluation vs actual error

0,91 0,84 0,77 0,7 0,63 0,56 0,49 0,42 0,35 0,28 0,21 0,14 0,07 0 R = 0,16

actual error

0,04 0,12 0,20 0,28 0,36 0,44 0,52 0,60 0,68 0,76 0,84 0,92 1,00 1,08 1,16 1,24 1,32 1,40 1,48 1,56 1,64 1,72 0,00 0,08 0,16 0,24 0,32 0,40 0,48 0,56 0,64 0,72 0,80 0,88 0,96 1,04 1,12 1,20 1,28 1,36 1,44 1,52 1,60 1,68 ev aluation

material vs actual error

0,88 0,8 0,72 0,64 0,56 actual error 0,48 0,4 0,32 0,24 0,16 0,08 0 40,25 38,25 36,25 34,25 32,25 30,25 28,25 26,25 24,25 22,25 20,25 18,25 16,25 14,25 12,25 10,25 41,25 39,25 37,25 35,25 33,25 31,25 29,25 27,25 25,25 23,25 21,25 19,25 17,25 15,25 13,25 11,25 9,25 material R = 0,01

The green curve shows the amount of positions in percentages, the red curve is a statistical regression line, R2 is the correlation coefficent. The most influential and most trustworthy of them is the difference criterion. In creating all the graphs, the number of moves that included at least 95% of valid positions, served as a basis. To get the expected error values by each difficulty factor, measure ponts must be chosen first. For the purpose of this study, the points were chosen in such a way that they are standing at the transition spots between percentiles of 25%, 50% and 75%: Complexity: 0.00; 0.34; 0.72. Difference: 0.03; 0.13; 0.42. Evaluation: 0.18; 0.44; 0.88. Material: 35.0; 28.75; 18.25. expected error by difference expected error by complexity 0,036
Carlsen Nanking 2009 1880s 2700 1870s 2400 1900s 2600 2200 1860s 1890s 2100 2500 2300 2000 0,000 av erage complexity of 0.00 0,100 complexity of 0.34 0,044 Carlsen Nanking 2009 0,107 0,111 0,113 0,129 0,130 0,134 0,143 0,159 0,164 0,169 0,176 0,184 0,213 0,200 0,300 av erage 2700 2600 2400 1900s 1870s 2500 2200 1860s 1890s 1880s 2300 2000 2100 0,053 0,066 0,068 0,074 0,080 0,087 0,092 0,094 0,094 0,097 0,108 0,111 0,126

0,000 0,050 0,100 0,150 0,200 0,250 dif f erence of 0.03 dif f erence of 0.13 dif f erence of 0.42

complexity of 0.72

expected error by evaluation

Carlsen Nanking 2009 2700 2400 1870s 1900s 1880s 2300 2500 1890s 1860s 2200 2000 2600 2100 0,000 av erage ev aluation of 0.18 0,100 0,063 0,111 0,130 0,134 0,139 0,143 0,154 0,155 0,160 0,171 0,199 0,200 0,200 0,203 0,200 0,300 ev aluation of 0.88

expected error by material

Carlsen Nanking 2009 1870s 1900s 2700 1880s 2600 1860s 2400 2500 1890s 2200 2300 2000 2100 0,000 av erage material of of 35.0 0,100 material of of 28.75 0,200 0,064 0,121 0,136 0,139 0,147 0,148 0,171 0,182 0,182 0,185 0,186 0,212 0,224 0,229 0,300

ev aluation of 0.44

material of of 18.25

average expected error

Carlsen Nanking 2009 2700 1870s 1900s 1880s 2400 2600 1860s 2500 1890s 2200 2300 2100 2000 0,000 av erage by ev aluation 0,050 0,100 0,150 0,052 0,104 0,112 0,120 0,124 0,127 0,137 0,149 0,150 0,151 0,155 0,165 0,182 0,187 0,200 0,250

by complexity by material

by dif f erence

The average expected error is the mean value of all specific expected error values. Having determined its values, it is time to fit it to an identical thinking time; in this study 3 minutes per move (2 hrs per 40 moves) was chosen as a common denominator. Unlike the other difficulty factors, thinking time is taken into account on the basis of ELO rating. Unfortunately it was not feasible to directly apply this effect on the actual average error. The relationship between the accuracy of play and the rating is represented by the graph below. The tme control there is correlated to 180 s per move. The relation is best described by logarithmic trendline. It hints at well-known fact that the more far back one is as to his playing skill, the easier it is to raise the rating.

average expected error vs ELO

0,200 0,180 0,160 R = 0,79

average expected error

0,140 0,120 0,100 0,080 0,060 0,040 0,020 0,000

30 50 29 00 27 50 26 00 25 00 24 50 23 50 22 00 21 00 20 50 19 50 19 00 30 00 29 50 28 50 28 00 27 00 26 50 25 50 24 00 23 00 22 50 21 50 20 00

ELO

Since at a lower level of play similar changes in the accuracy of play should at least theoretically cause bigger relative playing level fluctuations, logically, it must be indicated in FIDE rating statistics. And that's how it actually is! The graph below shows that the standard deviations of rating changes increase in the case of the rating of a player being lower. The highest rating peaks of randomly picked 8 chessplayers by half-years within 2005-2009 were compared. standard deviation of rating changes vs ELO
80,0 70,0 60,0 standard dev iation 50,0 40,0 30,0 20,0 10,0 0,0 1900 2000 2100 2200 2300 ELO 2400 2500 2600 2700 2800 R = 0,33

After applying thinking times on figures represented by the graph 'average expected error', we get a hypothetical playing strength for each player of the past in comparison with today's players. Figures on the bars display the expected rating according to the chart 'average expected error vs ELO'. The red bars represent performances corresponding to rating peaks of the decades. As can be seen on the graph below, the results are quite unstable, some of them show incredibly unrealistically high playing level. It demonstrates that the study is still in an immature stage, and one is expected to make no far-reaching conclusions yet. First of all, there is a need to increase the amount of games and positions and figure out how to solve the problem of practical, opponenoriented play and take such aspect of chess into consideration. Computers currently show no understanding whether a mistake is caused by an oversight or insufficient positional knowledge, or a deliberate calculated risk that by raising the difficulty level creates problems for the opponent to solve.

Average expected error by thinking time + rating

0,000 Carlsen Nanking 2009 0,050 0,023 0,100 0,150 0,200 0,250

2970
0,102

2700

2760 2650 2830 2640 2590 2810 2580 2760 2580 2460 2390 2360 2590 2330 2440 2290 2040 1980
0,050

1870s

0,122

2400

0,124

1900s

0,131

1880s

0,133

2600

0,133

2500

0,147

2200

0,153

1890s

0,157

1860s

0,159

2300

0,162

2100

0,180

2000

0,185

0,000

0,100

0,150 expected performance of rating peak of a decade

0,200

0,250

average expected error by time

The following is a list of games included in this analysis. Except Carlsen's games, moves of both participants were taken into account. By 'valid positions', those that satisfy the criteria described above were meant.
1860s no White 1 Kolisch 2 Anderssen 3 Paulsen 4 Anderssen 5 De Vere average rating Black 2628 Anderssen 2611 Steinitz 2652 Kolisch 2660 Paulsen 2675 Steinitz 2645 2653 1870s no White 1 De Vere 2 Anderssen 3 Zukertort 4 Anderssen average rating result moves year opening number of valid positions 2645 0-1 45 1861 B40 2688 0-1 44 1866 C51 299 2624 0,5-0,5 29 1861 C00 2663 1-0 29 1862 C65 2688 0-1 76 1867 A10 2662

rating Black 2601 Steinitz 2648 Paulsen 2663 Blackburne 2638 Zukertort 2638 2644

rating result 2696 0-1 2624 1-0 2617 0-1 2663 0-1 2650

moves year opening number of valid 50 1870 C60 positions 54 1873 C41 265 61 1878 C11 49 1878 C65

1880s no White 1 Chigorin 2 Mason 3 Bardeleben 4 Chigorin average

rating Black 2631 Schwarz 2689 Winawer 2651 Tarrasch 2636 Paulsen 2652 2656

rating result 2657 1-0 2676 0-1 2664 1-0 2645 1-0 2661

moves year opening number of valid positions 55 1882 C01 37 1883 C45 276 82 1888 D32 44 1881 B32

1890s no White 1 Bird 2 Schiffers 3 Burn 4 Cohn 5 Mason 6 Teichmann 7 Blackburne average

rating Black 2603 Blackburne 2662 Walbrodt 2669 Marco 2616 Schlechter 2670 Marco 2638 Charousek 2655 Cohn 2645 2651

rating result moves year opening number of valid 2665 0-1 30 1892 A02 positions 2664 1-0 42 1897 C00 308 2643 0,5-0,5 57 1898 D53 2697 1-0 32 1899 C44 2653 0-1 45 1894 C27 2660 0-1 61 1897 C77 2616 0-1 77 1899 C26 2657

1900s no White 1 Lasker 2 Mason 3 Alapin 4 Burn 5 Forgacs average

rating Black 2675 Marshall 2634 Brody 2658 Blackburne 2625 Swiderski 2669 Bernstein 2652 2645

rating result moves year opening number of valid 2658 0,5-0,5 45 1904 B40 positions 2602 1-0 55 1900 C45 269 2656 0,5-0,5 42 1901 B22 2623 1-0 34 1906 D60 2654 1-0 53 1909 B15 2639

Carlsen Nanking 2009 no White rating 1 Carlsen 2 Carlsen 3 Wang 4 Jakovenko 5 Carlsen 6 Leko 7 Topalov 8 Carlsen 9 Radjabov 10 Carlsen average

Black

rating result 2762 1-0 2812 1-0 2783 0,5-0,5 2782 0-1 2757 1-0 2791 0,5-0,5 2791 0,5-0,5 2741 1-0 2796 0,5-0,5 2740 1-0 2776 2774

moves year 44 41 58 63 25 64 43 69 34 38

opening number of valid positions 271

2772 Leko 2777 Topalov 2738 Carlsen 2743 Carlsen 2787 Radjabov 2757 Carlsen 2808 Carlsen 2792 Wang 2747 Carlsen 2796 Jakovenko 2772

2009 C45 2009 E90 2009 D83 2009 B92 2009 B30 2009 D72 2009 B33 2009 D17 2009 D86 2009 D31

2000 no White 1 Luodonp 2 Buehler 3 Falkowski 4 Boucek 5 Achereiner 6 Schaad 7 Miciak average

rating

Black

rating result 2022 0,5-0,5 1992 1-0 2015 0,5-0,5 1978 0,5-0,5 1979 1-0 2023 0,5-0,5 2023 0-1 2005 2003

moves year 57 41 116 50 40 45 39

opening number of valid positions 336

1991 Pohjala 1992 Fiedler 2002 Stolarczyk 2017 Straka 2025 Kirk 1996 Rickenbach 1987 Stric 2001

2008 A12 2008 D13 2008 B26 2008 B26 2008 E97 2008 A08 2008 E99

2100 no White 1 Potze 2 Kanyadi 3 Weber 4 Crombleholme 5 Li 6 Milonakis average

rating

Black

rating result 2121 0,5-0,5 2081 0,5-0,5 2084 1-0 2096 1-0 2099 1-0 2082 1-0 2094 2096

moves year 52 45 42 50 45 50

opening number of valid positions 310

2105 Galje 2103 Rakaczki 2078 Volkov 2101 Gustavsson 2108 Sermier 2099 Bras 2099

2008 B22 2008 B01 2008 D40 2008 D63 2008 D11 2008 B14

2200 no White 1 Kozak 2 Tsganova 3 Shytaj 4 Masse 5 Wyss 6 Jagodzinski average

rating

Black

rating result 2212 0,5-0,5 2182 0,5-0,5 2208 1-0 2213 1-0 2205 0-1 2179 0-1 2200 2202

moves year 63 83 42 45 53 43

opening number of valid positions 321

2195 Toma 2221 Korchagina 2212 Zoldan 2196 Arsenault 2181 Kojima 2216 Klim 2204

2008 D93 2008 E61 2008 B07 2008 E32 2008 B33 2008 B22

2300 no White 1 Geenen 2 Barreto Filho 3 Petersen 4 Limontas 5 Franciskovic 6 Pachta average

rating

Black

rating result 2321 0,5-0,5 2298 1-0 2316 0,5-0,5 2277 0-1 2288 0-1 2278 1-0 2296 2297

moves year 35 89 48 41 44 35

opening number of valid positions 318

2304 Ringoir 2300 Prates 2309 Jacobsen 2289 Tvarijonas 2275 Kantorik 2312 Aschenbrenner 2298

2008 B96 2008 D11 2008 B06 2008 D02 2008 A21 2008 E68

2400 no White 1 Marinovic 2 Lffler 3 Housieaux 4 Brumen 5 Huss 6 Saptarshi 7 Zvara average

rating

Black

rating result 2393 0-1 2415 0-1 2378 1-0 2396 0,5-0,5 2389 1-0 2423 0,5-0,5 2409 0-1 2400 2401

moves year 37 38 38 41 40 68 50

opening number of valid positions 317

2408 Nestorovic 2411 Pinter 2411 Debray 2383 Rogulj 2378 Gerber 2404 Himanshu 2409 Kanovsky 2401

2008 C85 2008 D80 2008 D25 2008 C26 2008 E14 2008 D01 2008 B43

2500 no White 1 Jakubowski 2 Porper 3 Ushenina 4 Vovk 5 Bhat 6 Mainka 7 Jakubowski average

rating

Black

rating result 2506 0-1 2524 0,5-0,5 2501 0-1 2495 1-0 2488 0,5-0,5 2525 1-0 2511 0,5-0,5 2507 2498

moves year 73 47 43 43 46 35 35

opening number of valid positions 330

2497 Onischuk 2476 Friedel 2484 Carlsson 2488 Sulashvili 2498 Lima 2482 Firman 2497 Bachmann Schiavo 2489

2008 B15 2008 E13 2008 A60 2008 B19 2008 D53 2008 B50 2008 A48

2600 no White 1 Smeets 2 Malakhatko 3 Atalik 4 Macieja 5 Guseinov average

rating

Black

rating result 2610 0,5-0,5 2604 0,5-0,5 2607 1-0 2599 0,5-0,5 2614 0,5-0,5 2607 2606

moves year 60 37 55 98 35

opening number of valid positions 302

2604 L'Ami 2612 Mchedlishvili 2585 Pantsulaia 2606 Rozentalis 2617 Rodshtein 2605

2008 B19 2008 E11 2008 A30 2008 E40 2008 C12

2700 no White 1 Van Wely 2 Bologan 3 Alekseev 4 Alekseev 5 Jakovenko 6 Cheparinov 7 Wang Hao average

rating

Black

rating result 2707 0,5-0,5 2678 0,5-0,5 2708 0-1 2711 0,5-0,5 2687 1-0 2720 0,5-0,5 2716 1-0 2704 2699

moves year 37 66 45 45 50 41 37

opening number of valid positions 337

2681 Polgar 2682 Naiditsch 2708 Dominguez 2711 Jakovenko 2709 Cheparinov 2687 Gelfand 2684 Grischuk 2695

2008 E21 2008 C55 2008 B91 2008 E26 2008 C67 2008 C42 2008 C89