From: "World Chess Championship", INTERNET:newsletter@mark-weeks.com Date: 01/02/15, 10:56 Re: Chess History on the Web (2001 no.4) Site review - CRM II In the last review (2001 no.3) I started to look at Chess Records Management (CRM), a database of games for trade. The site, run by John Pope, is at address... http://users.imag.net/~lon.jpope/ ...Although there are no games available for download, the site's 16 index pages contain a wealth of data about chess competitions. I captured this data and loaded it into a database for further analysis. A table reproduced from the last review... Ct Type #Games 653 M 5.613 4000 RR 279.205 1363 SS 256.955 649 TT 171.034 ...shows the number of matches (type 'M'), round robins ('RR'), Swiss systems ('SS'), and team tournaments ('TT') in the CRM database, as well as the number of games for each type. This makes a total of 712.989 games in 6665 events of all types. The most recent events in the database were played in 1998. I'm going to include many tables and numbers in this review. I hope that they align properly in your email software. If they don't, I suggest that you extract ('save to file' -or- 'copy & paste') the text of this article and read it using a text editor configured for a nonproportional font. As an alternative, a copy of this article will be available in the list archive whose address is given at the end of this message. This should display properly formatted in your Web browser. Structured databases provide easy answers to questions like what is 'the most of this', 'the average of that', or 'the first of these'. To get started, let's look at some records. Here's a list of the longest matches in the CRM database:- Year Event #Games Winner 1971 Iceland (ICE-ch) 66 Kristinsson,J 1984 Karpov,A-Kasparov,G (Wch) (m) 48 Karpov,A 1927 Alekhine,A-Capablanca,J (Wch) (m) 34 Alekhine,A 1978 Karpov,A-Korchnoi,V (Wch) (m) 32 Karpov,A 1861 Kolisch,I-Paulsen,L (m) 31 Paulsen,L 1853 Harrwitz,D-Loewenthal,J (m) 31 Harrwitz,B 1845 Rousseau,E-Stanley,C (m) 31 Stanley,C 1845 Rousseau,E-Stanley,C (m) 31 Stanley,C 1935 Alekhine,A-Euwe,M (Wch) (m) 30 Euwe,M 1992 Fischer,R-Spassky,B (m) 30 Fischer,R Here we see that even though the CRM data may be well maintained, we have to be careful with it. That 66 game match ('ICE-ch') looks like a classification mistake, while the two 1845 Rousseau - Stanley matches look like duplicates. Ignoring the Iceland event, the three longest matches were world championship title events where the winner was the first to win six games. The 1984 match is, of course, the controversial match stopped by FIDE President Campomanes in February, 1985. What does CRM say about short matches? Curiously, of the 17 matches missing the number of games, all were played by Alekhine or Euwe, including two matches between the two. Ignoring these, the shortest matches seem to have been one game affairs:- Year Event #Games Winners 1866 Minchin,J-Trelawney,J (m) 1 Minchin,J 1958 Fischer,R-Matulovic (m) 1 Fischer,R 'Bobby Fischer's Chess Games' by Wade and O'Connell says that Fischer - Matulovic was a four game match which Fischer won with 2.5 points, and that three of the game scores are missing. Indeed, the header of the CRM column for match length is 'CRM Gms', which is not necessarily the number of games played in the match. CRM also tell us that there have been 22 two game matches and 35 three game matches. Many of these were undoubtedly playoff matches to determine the clear winner of a title event. The most common match length is six games:- #Games Ct 6 108 8 77 4 75 10 68 5 41 This table also shows that matches with an even number of games are more popular than matches with an odd number of games. These numbers don't show how many games were scheduled for the match; they only show how many games were played. Here's a list of the greatest number of players in a round robin:- Year Event #Players Winners 1905 Barmen - Supplementary (RR+) 74 [various] 1997 Groningen/Lausanne (Wch KO) 68 Anand,V 1912 Breslau - Congres C (RR+) 42 Kruger 1997 Elista (RUS-ch KO) 40 Svidler 1929 Odessa (URS-ch) 36 Verlinski,B 1998 Ramat Aviv (ISR-ch) (KO) 32 Liss,E; Mikhalevsky,V 1998 Elista - President's Cup (KO) 32 Dreev,A 1998 Elista (RUS-ch wom) (KO) 32 Prudnikova,S Most of these were not round robins, but knockout events ('KO'). For Odessa 1929, 'The Soviet Championships' by Cafferty and Taimanov says that the 6th USSR Championship had '36 contestants playing three games a day in four groups', so this was also not a true round robin. As for the fewest players in a round robin, there's no surprise here, it's three -- CRM lists 36 events. The most popular round robin format has been 12 players:- #Players Ct 12 920 10 860 14 500 16 445 8 195 The most popular format for an odd number of participants has been 11 players:- 11 125 15 109 13 91 17 38 I suspect that the CRM data for Swiss systems is less comprehensive than for the other types of events. Only seven Swisses are listed with more than 1000 games:- Year Event #Games 1995 Wurzburg (op) 1271 1997 Tallin (EUR-ch jr) - Boys 1187 1988 Belgrade - GMA (op) 1132 1987 Dortmund (op) 1123 1997 Germany - OIMB Bank Hofmann (op) 1112 1994 Berlin - Summer (op) 1041 1992 Dortmund (op) 1001 Of the 1363 Swiss events, 937 are missing the number of players and 700 are missing the number of rounds. CRM lists eleven events with 400 or more players. The two largest were:- Year Event #Players Winners 1998 Cappelle la Grande (op) 637 Glek,I 1997 Berlin - Summer (op) 542 Chuchelov,V [3 others] Of the other nine, also held in the 1990s, three were Cappelle la Grande (France) Opens and six were Berlin (Germany) Summer Opens. Seven Swiss systems are listed with 10 or fewer players, three of them Michigan events:- Year Event #Players Winners 1996 Groningen - Staunton Memorial 7 Smyslov,V 1992 Ann Arbor - Michigan (ch) 8 Tsemekhman,V 1991 Ann Arbor - Michigan (ch) 8 Lindsay,F 1998 Panormo (zt) 9 Cela,A; Kotronias,V 1993 Westland - Michigan (ch wom) 9 Skidnore,J 1998 Levekusen 10 Balduan,M; Mueller,D 1994 Cap d'Agde [A] 10 Gelfand,B The most popular length for a Swiss event is 9 rounds:- #Rounds Ct 9 332 11 91 7 74 5 55 10 35 [...] 15 1 72 1 The last two in the list (15 & 72 rounds) are the longest Swisses in the CRM database:- Most rounds:- Year Event #Games #Players #Rounds Winners 1998 Panormo (zt) 325 9 72 Cela,A; Kotronias,V 1974 Menden 270 36 15 Eising,J; Ostermeyer,P The 1998 event is almost certainly a mistake. The event may have had 72 players in 9 rounds, but even that is a big zonal tournament ('zt'). CRM lists 12 team tournaments with 2000 or more games. The most number of games are from Olympiads:- Year Event #Games Winners 1994 Moscow (ol men) (tt) 5128 [Russia] 1996 Yerevan (ol men) (tt) 3142 [Russia] Three of the remaining ten were German Bundesliga events. Only 151 of the 649 team tournaments are listed with the number of players. Of these, five events have more than 100 players:- Year Event #Players Winners 1996 Yerevan (ol men) (tt) 456 [Russia] 1996 Yerevan (ol wom) (tt) 296 [Georgia] 1966 Moscow - Spartakiad (tt) 143 [Army Team] 1997 Pula (EUR-ch tt) 136 [Great Britain] 1997 Pula (EUR-ch tt wom) 120 [Georgia] --- That's all very nice, but it's not particularly enlightening. The results are certainly skewed by the geographical location of Pope's correspondents. It's hard to believe that Michigan is a world leader in small Swiss systems. I decided to investigate further. To look at match results, I extracted the names of both players listed for each match. The top 10 match players of all time are:- Euwe 40 Karpov 30 Korchnoi 27 Alekhine 25 Timman 23 Lasker 21 Kasparov 21 Spassky 18 Larsen 18 Steinitz 15 Euwe has 34 matches listed with the number of games, making a total of 303 match games. All of Karpov's and Korchnoi's matches are listed with the number of games; Karpov's total is 380 games & Korchnoi's is 297. Five of Euwe's matches were apparently played with Alekhine:- Year Event Gms Winners 1926 Alekhine,A-Euwe,M (m) 10 Alekhine,A 1927 Alekhine,A-Euwe,M (m) Alekhine,A 1935 Alekhine,A-Euwe,M (Wch) (m) 30 Euwe,M 1937 Alekhine,A-Euwe,M (Wch) (m) 25 Alekhine,A 1937 Alekhine,A-Euwe,M (m) (exhibition) Euwe,M 'Alexander Alekhine' by Kotov list four Alekhine - Euwe matches, given here with the result from Alekhine's side:- 1927 +3-2=5 1935 +8-9=13 1937 +10-4=11 1937 +1-2=2 (exh.) Where is the 1926 match? CRM says that it was a 10 game match, which is also the number of games played in 1927. This is more than a coincidence. 'World Chess Champions' by E.Winter gives the years 1926-27 for the match, so it appears that it is duplicated in the CRM database. A look at Karpov's matches reveals seven with Kasparov. The generally accepted number is five, so where do the two extra matches come from?:- Year Event Gms Winners 1984 Karpov,A-Kasparov,G (Wch) (m) 48 Karpov,A 1985 Karpov,A-Kasparov,G (Wch) (m) 24 Kasparov,G 1986 Karpov,A-Kasparov,G (Wch-1) (m) 12 Kasparov,G 1986 Karpov,A-Kasparov,G (Wch-2) (m) 12 [drawn] 1987 Karpov,A-Kasparov,G (Wch) (m) 24 [drawn] 1990 Karpov,A-Kasparov,G (Wch-1) (m) 12 Kasparov,G 1990 Karpov,A-Kasparov,G (Wch-2) (m) 12 Kasparov,G CRM counts the 1986 London - Leningrad and 1990 New York - Lyon title matches each as two mini-matches of 12 games. Korchnoi's string of 18 candidate matches spans 24 years. During much of that time he was nearly invincible in match play:- Year Event Gms Winners 1968 Korchnoi,V-Reshevsky,S (M)(cq) 8 Korchnoi,V 1968 Korchnoi,V-Spassky,B (m) (cf) 10 Spassky,B 1968 Korchnoi,V-Tal,M (m) (cs) 10 Korchnoi,V 1971 Geller,E-Korchnoi,V (m) (cq) 8 Korchnoi,V 1971 Korchnoi,V-Petrosian,T (m)(cs) 10 Petrosian,T 1974 Korchnoi,V-Mecking,H (m) (cq) 13 Korchnoi,V 1974 Korchnoi,V-Petrosian,T (m) (cs) 5 Korchnoi,V 1977 Korchnoi,V-Petrosian,T (m) (cq) 12 Korchnoi,V 1977 Korchnoi,V-Polugaevsky,L (m) (cs) 13 Korchnoi,V 1978 Korchnoi,V-Spassky,B (m) (cf) 18 Korchnoi,V 1980 Korchnoi,V-Petrosian,T (m) (cq) 9 Korchnoi,V 1980 Korchnoi,V-Polugaevsky,L (m) (cs) 14 Korchnoi,V 1980 Hubner,R-Korchnoi,V (m) (cf) 10 Korchnoi,V 1983 Kasparov,G-Korchnoi,V (m) (cs) 11 Kasparov,G 1983 Korchnoi,V-Portisch,L (m) (cq) 9 Korchnoi,V 1988 Hjartarsen,J-Korchnoi,V (m) (cq) 8 Hjartarson,J 1991 Korchnoi,V-Sax,G (cm) 10 Korchnoi,V 1991 Korchnoi,V-Timman,J (cm) 7 Timman,J National championships are marked in the CRM database by a three character country code plus '-ch'; there are a total of 924 entries. The 10 countries with the most championships, plus the first year where CRM has a record of the championship are listed in the following table:- Tnmt 1stYear URS-ch 61 1920 USA-ch 46 1867 URS-ch sf 43 1931 CZE-ch 42 1940 YUG-ch 41 1945 HUN-ch 40 1931 NET-ch 39 1909 DEN-ch 35 1951 ARG-ch 30 1936 CAN-ch 27 1924 The USSR appears twice because semifinals ('sf') are also identified using same code as national championships. Many of the national championships held since 1970 are listed together with the FIDE category of the event. The following table shows countries with more than 10 national championships also having a category. The last column is the average category across all events:- Tnmt Ct Cat USA-ch 17 11.65 URS-ch 18 11.61 HUN-ch 13 9.54 YUG-ch 13 9.08 URS-ch sf 21 8.81 NET-ch 20 8.20 CZE-ch 14 8.14 The result is surprising. Have the USA championships really been stronger than the USSR championships? A look at the category of the championships by year puts this into perspective. Year CZE HUN NET URS USA YUG 1971 11 1972 11 9 1974 4 3 1975 3 1976 7 10 1977 12 1978 12 10 9 1979 7 6 12 1980 5 6 12 10 1981 9 12 12 11 1982 9 9 1983 9 7 14 10 8 1984 7 10 6 10 11 9 1985 9 7 11 10 7 1986 7 8 8 10 10 9 1987 8 8 12 12 1988 8 7 14 12 1989 11 8 8 13 12 10 1990 9 14 9 1991 8 9 1992 11 10 11 1993 11 10 14 1994 11 15 10 1995 11 11 12 10 1996 8 11 12 9 1997 10 15 12 10 1998 9 12 13 Here we see that in the 10 years with data for both countries, the USSR wins +7-1=2. The USA average is higher because its championships in recent years had a higher category than the USSR championships in the early 1970s. This may be due to rating inflation starting in the late 1980s. You may have noticed that the first table of championships showed the USA-ch with 17 events and the URS-ch with 18, but the second table shows each country with one less. This is because, for each country, one year appears twice:- Tnmt Year Cat URS-ch 1984 10 USA-ch 1998 13 This might be either a mistake in the data or that championships for two years were played at the beginning and the end of a calendar year. I didn't resolve this, as it makes only a small difference to my calculations. Other championships, marked '(ch)', are also recorded in the CRM database. The entries which appear the most frequently among the total of 199 championships are:- Event Ct 1stYear Arhus (ch) 13 1983 Canada - British Columbia (ch) 11 1972 Leningrad (ch) 10 1926 Moscow (ch) 9 1942 Copenhagen (ch) 6 1974 Toronto (ch) 4 1990 Paris (ch) 4 1989 Great Britain - BCF (ch) 4 1968 That last entry in the list exposes another anomaly. The 20 championship events marked 'BCF', where the first was held in 1946, are in addition to the 8 events marked 'GBR-ch', starting in 1974. I suspect that these 28 events are all in the same series, which would have vaulted Britain into the preceding list of national championships, ahead of Canada ('CAN-ch'). Title events and other FIDE competitions are well covered by CRM. There are:- - 97 Candidate events (marked 'ct', 'cf', etc.), of which 6 are for women, - 37 Interzonals ('izt'), of which 7 are for women, - 153 Zonals ('zt'), 12 for women, - 46 Olympiads ('ol'), 11 for women. The women's events are among the total of 139 events restricted to women. Of the 1850 events listed with a FIDE category, 13 are marked 'w', indicating a women's category. The strongest women's events on the database are:- Year Event Cat Winners 1997 Groningen (ct wom) 17w Galiamova,A 1994 Tilburg (ct wom) 17w Polgar,Z; Chiburdanidze The categories of unrestricted events are distributed as follows:- Cat Ct 21 2 19 6 18 13 17 27 16 38 15 50 14 71 13 84 12 118 11 141 10 261 9 303 8 303 7 178 6 53 5 57 4 68 3 69 2 7 1 1 This shows that the most common FIDE events have been clustered in the 8-10 category range. Why should category 3 events be fewer than category 8 events? This may be a bias arising from the type of events selected for CRM. The database's 900 opens ('op') break down as follows:- Event Ct 1stYear New York Open (op) 12 1984 London - Lloyds Bank (op) 11 1978 Copenhagen (op) 11 1980 Bad Worishofen (op) 10 1988 Cannes (op) 9 1989 Bled (op) 9 1991 Groningen (op) 8 1988 Valencia - Mislata (op) 7 1992 Cappelle la Grande (op) 7 1989 Biel (op) 7 1990 Berlin - Summer (op) 7 1991 This raises the question, 'Which venues have been the most popular for chess tournaments'? Excluding matches, which have no venue listed, CRM lists venues for 6012 events. The most frequent are:- Venue Ct 1stYear Budapest 271 1896 Moscow 132 1908 Hastings 94 1895 Buenos Aires 94 1926 Amsterdam 93 1889 New York 80 1857 London 66 1849 Copenhagen 64 1899 Arhus 61 1959 Wijk aan Zee 60 1968 Budapest is first on the list thanks to 152 First Saturday tournaments, which apparently started in 1993. Linares appears 24 times, of which:- - 3 are for the event of the same name in Mexico, - 5 are Opens, and - 1 is a zonal tournament. More than 5500 of the events in the CRM database are listed with a winner; 1300 have more than one winner. For the round robins, I derived a list of 3855 events with 5055 names. The names appearing the most are:- Name Ct Korchnoi 66 Karpov 60 Tal 51 Alekhine 50 Larsen 47 Portisch 40 Hort 40 Timman 37 Kasparov 36 Gligoric 36 A similar manipulation on Swiss systems yielded 839 events with 1196 names. Here the names appearing most frequently are:- Name Ct Tukmakov 12 Tsemekhman 9 Hector 8 Hodgson 8 Khalifman 8 Gurevich 7 Miles 7 Nielsen 7 Sokolov 7 The second name on the list was not familiar to me, but I quickly discovered that V. Tsemekhman has been a consistent winner in Michigan events during the 1990s. As for Tukmakov, I'll add one more table in case anyone wants to dispute his claim to first position on the list:- Year Event Winners 1985 Lugano (op) Tukmakov 1990 Amsterdam - OHRA B Polgar,J; Tukmakov 1991 Bern (op) Epishin; Tukmakov 1992 Palma de Mallorca (op) Gild. Garcia; Tukmakov 1992 Lenk (op) Tukmakov 1992 Helsinki (op) Neverov,V; Tukmakov 1993 Wijk aan Zee - Hoogovens (op) Tukmakov,V 1993 Bern (op) Epishin; Tukmakov 1994 Winnipeg - Canadian (CAN-ch) (op) Tukmakov 1994 Pula (op) Tukmakov 1995 Zurich (op) Tukmakov 1995 Portoroz (op) Tukmakov --- What can we conclude from all of these statistics and lists? It depends. They may be useful as a starting point for deeper researches; I'm presently using the CRM list of zonal tournaments for exactly this. Perhaps more importantly, they show how truly global is the game of chess and how it touches so many cultures. But we already know this, don't we? Bye for now, Mark Weeks