InfinityChess Freestyle Battle 2014

InfinityChess Freestyle Battle 2014
Tour report by Werner Bergmans, MiG29
(Eindhoven, April 12th 2014)
First of all I would like to congratulate the organizers and the sponsor of the Infinity Freestyle Battle
2014, with this fantastic event. I really enjoyed the tournament and the professional way it was lead by
Arno Nickel and his team. I also want to thank GM Christopher Lutz for his insightful round reports.
Finally congrats to everyone participating and especially for those who made it in the prize money!
Who am I:
My name is Werner Bergmans, I became 43 years old during this tournament and am living in
Eindhoven, the Netherlands. I have a full time job as software integrator at ASML (the world's leading
provider of lithography systems for the semiconductor industry). This tournament I played under the nick
‘MiG29’, but most people will know me better under the nick I use on Rybka forum and Playchess:
Fulcrum2000. A long time ago, when I was young, I was a very mediocre club player. I used to analyze
my games with chess software like Fritz 3 and Chess Genius. After about 3 years or so I stopped playing
as it cost me too much time, however the interest in chess programs still remained. Ironically by
stopping playing chess OTB and focusing on chess programs my overall understanding of chess has
improved dramatically since then.
My freestyle background:
As I’m probably not as well known as the other freestylers, particularly the top3, I will give some
background on my Freestyle ‘career’. My first real freestyle experience I got with the 7th and 8th PAL/CSS
Freestyle Tournament played on the Playchess server (back in 2007 and 2008). Results were quite
sobering and I learnt how not to prepare for such events. I used downloaded books, and had no
reference database what so ever, used weak hardware and had no good idea how to interact with
engines. As a result I suffered from book losses and was regularly outplayed in the opening and early
middle game.
After this disillusion I knew something had to be changed. The next years I started to work on my own
book/database and worked on it almost daily ever since. Unfortunately the PAL freestyle events stopped
and it took till 2010 before I could finally participate in a big event again. The hard work paid of and I
was pleased with the results:
* Mundial Chess Tournament 2010 (1st International Online Advanced Chess Tournament):
3th place Mundial Chess Tournament - group players up to 2200 Elo FIDE
4th place Mundial Chess Tournament - group players without Elo FIDE and up to 1.500 Elo FIDE
* WBCCC World Blitz Correspondence Chess Championship 2011 - 2nd place (out of 35) * 1st Freestyle Chess Tour 2012 (Infinitychess) - 3rd place (out of 50)
Tournament preparation:
Before starting this (and any other) tournament several things have to be in place, notably the hardware
(PC’s), and software (chess engines, endgame tablebases, openings book and Reference database).
Before the tour I had one 6-core machine and an older Q-6600 quadcore machine. I decided I needed at
least a second 6-core machine to be at least marginally competitive. The whole tour I only used those
two 6-core machines, the quadcore was not used. My setup:
- Intel i7-3960X 6-core overclocked to 4.7 Ghz (16 GB memory, 6-men Syzygy EGTB's on SSD)
- Intel i7-4930K 6-core overclocked to 4.4 Ghz (16 GB memory, 6-men Syzygy EGTB's on SSD)
On the software side I used my own hand tuned opening book (in ctg-format) and a reference database
consisting of about 5.6 Million games. I try to keep my reference database between 5 and 6 Million
games by removing about the same number of games as I add. This keeps the content fresh as for
example 3 minute games played in 2007, with slow hardware and weak engines, will not add much
value to the database (I know some people disagree with me here, correct Nelson? ;-). The reference
database consist mainly on engine-engine games (from Playchess, TCEC, own testgames etc),
correspondence games, and top level GM games.
One of the choices with the most influence is the chess engines used to analyze. I personally think there
are currently only three engines to consider when taking into account the time control of the tournament
and the bunch of public rating lists: Stockfish, Houdini and Komodo. As I only had two (reasonably)
competitive machines and do not have much experience with Komodo I decided for Houdini 4 Pro and the
latest Stockfish Syzygy development version (changes every few days) as my main engines. Occasionally
I used Komodo for double checking some lines (mainly in the end game phase). As Stockfish engine I
used the special ‘Rockwood’ compile which adds Large Page support + better compile options on top of
the Syzygy support.
On both systems the full 6-men Syzygy tablebases of Ronald de Man were installed on a fast SSD. Both
Houdini 4 and Stockfish Syzygy are able to fully utilize these 6-men EGTB’s.
A picture of my Control Center, were all the action was taking place. On the left part of the left screen we
see the Infinity chess client, on the right side of the screen we see the Fritz 13 GUI with Stockfish doing
it’s magic and a notation window were several lines are being investigated. On the right screen we see
the reference database on the left, the board and opening book in the middle and the notation window +
Houdini 4 Pro on the right part of the screen. (the screenshot is not of an actual game)
Game preparation:
Before every game I tried to get some relevant information about my opponent, trying to use this
information to turn the game into my advantage. Main source of information was collecting (engine)
games played by the opponent (both in this tournament and on chess servers like Infinitychess and
Playchess). With these games I then tried to find weaknesses in their opening or certain other patterns I
could potentially make use of. One of my opponents for example always played e6 (2. d4 d5 3. e5 c5 4.
c3 Nc6 5. Nf3 Qb6 6. a3 f6) as black against 1. e5. This was valuable information and allowed me to do a
lot of preparation before the game.
I also tried to make an assessment how strong my opponent was (this was easier in the later stages of
the tournament). I normally prefer to play more solid lines against strong opponents, as in this
tournament my hardware was probably sub-standard compared to the top players.
In between rounds the database was of course updated with the games played and other interesting high
level games like the WBCCC and TCEC events which were running in parallel.
I played all my games as a sole centaur.
Some notable games:
Normally it’s a good tradition to take a couple of heroic wins and explain in detail how you pulled that off,
but as the report is already getting too big and I have way too many games/positions I want to show, I
will do it a bit different. I will start with making some comments on my lost game against Donkasand as I
want to show what happens if you don’t follow your own rules / best practices. For the nice wins I refer
to the round reports of GM Lutz who has done an excellent job annotating them.
In hindsight I have to admit I had a complete off day in my game against Donkasand (Michael
Glatthaar). Before the game I did not have the time to study my opponent and failed to notice he was an
engine only player (mistake 1). As I did not prepare anything and was still a bit rushed from work I
blindly played e5 against his e4, with the idea e5 is the safest way too play an unknown opponent. Again
I failed to notice I was playing a standalone engine, else I would have played c5 with much more
opportunities to outplay an engine (mistake 2).
In the game itself I made some unnecessary risky moves while still in book, notably 5 … d5(?) where a
simple castling move was the correct option and the inaccurate 10 … a6 (mistake 3 and 4). But the worst
part has still to come. What probably cost me the game was failing to bring my white squared bishop to
the kingside on time where it was needed to defend his king, and at least pose some threat to the enemy
king (mistake 5). The position quickly went downhill and I kept making mistakes like playing way too fast
while I had enough time on the clock (mistake 6). Near the end of the game one of my engines saw a
drawing line and I made that move without investigating it further or cross checking it with another
engine (mistake 7). After making the move, the engine saw its mistake and it was game over. With so
many mistakes you deserve too lose.
I had one other difficult game; it was against Alvin Alcala (Maximus), again with me playing black. I
made some small inaccuracies and Alvin was fast to exploit them, slowly but steadily improving his
position. Luckily I realized in time things were looking not so good and devoted a lot of time trying to find
a way out of the mess. With only minutes on the clock I secured the draw. Analysis after the game
showed black’s position was bad, but in every line I tried, I could get away with a draw. I’m curious to
read Alvin’s report to see if he found a winning line for white.
Because a chess report without at least one chess position in it isn’t a complete chess report I will post a
position from my game against Akhtar were a lonely engine has a hard time finding the key move.
In the position on the right (MiG29 [ct] – Akhtar [cryptic Heroes 1.1]),
white is to move and has a key move that can decide the game.
Engines (notably Stockfish for some reason) have a hard time finding
this move (h4!). It also took the engine of the opponent a couple of
moves more before it saw the severity of the situation. A centaur player
with a nice toolset at his disposal (multi-variations, multiple engines,
deeper manual investigation of promising lines etc) will have much less
difficulty in finding the strength of this move.
FEN: 1rb2rk1/3q2bp/2nNp1p1/p2pP3/R4B2/2P2N2/2Q2PPP/4R1K1 w - - 0 26
Conclusion:
I totally enjoyed the well organized tour and the stable infinity chess platform. The tour was
professionally led by Arno Nickel (JoJo) and his team, the server was super stable and the atmosphere
between the players was very good. I particularly liked the interaction with fellow centaurs after the
game. Often in depth information was exchanged about missed moves, critical lines and other useful
information.
The level of play was very high, this is partly due to better engines and faster hardware (compared to
several years ago), but also limiting the tournament to the best 30 players contributed to this. Winning a
game against a top 10 player in this tournament proved a hard and nearly impossible task.
After such a successful event it’s hard to come up with improvement proposals, but let me give at least
some things to consider for a next tournament as we all hope this gets a continuation.

The time span of the tournament was very long, more then 2 month. Shortening it might attract
even more players. It might also be an idea to play more matches in the weekend (e.g. 2 matches
on both Saturday and Sunday) so people have less difficulty planning around their normal day job.

Encourage centaur play over engine only play. Overall the engine players seemed a bit less involved
and less communicative. As I already mentioned I really like doing analysis of the game with my
opponent after the game. Did went very well with most centaur opponents, but I hardly got any
feedback from engine only players. I think this interaction is a big part of the fun.
Some things not to change:

Time control, 90m + 15s is perfect in my opinion

Breakdown of prize money. Relatively big group which can earn money but still with a clear stimulus
to end in the top3.
Werner Bergmans
Aka MiG29 / Fulcrum2000