Missed ECO codes           :      0x after 13.120 games =  All ECO codes min. 1x on board

March 06th, 2021 (07:45) News-31
*** Tournament average changed from 3097 to 3094 Elo, other news ***

1. (Since round 21 out of 50)
After my new calculation I changed the tournament average from 3097 to 3094 Elo for Shredder 13 GUI *.html
tournament table. The calculation can be found in my *.zip file.

2.
I restarted the *.pgn database after 16.400 games on my start page, updated each 30 seconds with FTP client
software Fling-Plus 5.04. The complete databases can be found in my *.zip file.

February 04th, 2021 (13:00) News-30
*** Tournament average changed from 3095 to 3097 Elo, other news ***

1. (Since round 11 out of 50)
After my new calculation I changed the tournament average from 3095 to 3097 Elo for Shredder 13 GUI *.html
tournament table. The calculation can be found in my *.zip file.

2.
I restarted the *.pgn database after 8.200 games on my start page, updated each 30 seconds with FTP client
software Fling-Plus 5.04. The complete databases can be found in my *.zip file.

3.
Klaus Wlotzka is now ready with the new Excel files and wrote a bit about his work in CSS-Forum.

January 25th, 2021 (01:00) News-29
*** Excel files overworked / rewritten ***

Klaus Wlotzka overworked / rewritten his Excel statistics.
The six graphic examples will be updated round by round on my start page.

The new *.xlms file can be found in his *.zip file.

January 22nd, 2021 (09:00) News-28
*** Tournament average changed from 3080 to 3095 Elo ***

After a new calculation I changed the tournament average from 3080 to 3095 Elo for Shredder 13 GUI *.html
tournament table. I sent Klaus Wlotzka the information for his Excel tables.

Calculation I added in TalkChess thread: 76388

January 13th, 2021 (15:30) News-27
*** Minic 3.02 NaNi Skylake x64 = disqualified ***

After three x lost on time in round 3 out of 50 I disqualified Minic 3.02 NaNi Skylake x64.

I had the same problems with Minic 2.33 x64 for FCP Toruney-2020 and reported the bug. Minic lost often
on time around 2-6 moves before own mate (Shredder 13 GUI information: Analyze = empty).

That is s bad situation because Klaus and I must make the statistics for round 1-2 again. I did not test
Minic before I started the tournament. The problem is homemade!

OK, since I started the tournament two interesting engine updates are available. I added the updates. In
additional the current Stockfish dev. version.

01. Minic 3.02 NaNi Skylake will be replaced with Hannibal 1.7 x64.
02. SlowChess Blitz 2.4 NN x64 is replaced with SlowChess 2.5 NN AVX2 x64.
03. Marvin 4.0.1 POPCNT x64 is replaced with Marvin 5.0.0 NN AVX2 x64.
04. Sf 2020-12-31_18 NN BMI2 x64 is replaced with Stockfish 110121 NN BMI2 x64.
    Timestamp: 1610390551, Stockfish is running in Hybrid mode (default).

With Stockfish I am not sure?
I suppose NNUE will be load with Shredder 13 GUI only, if the full path for the NNUE file in UCI
configuration is given. Furthermore, the nn-62ef826d1a6d.nnue file should be add in Stockfish directory.

My configurations can be found under:
Engine-Configuration

462 games are to replay, these need ~ 2 days!

January 08th, 2021 (01:30) News-26
*** Anchor to Shredder 13 POPCNT x64 with 3125 Elo ***

Two users do not understand why I set for FCP Tourney-2021 an Shredder 13 POPCNT x64 anchor to 3125 Elo.
Performance for Shredder 13 POPCNT x64 is 3105 for FCP Toruney-2020.

1. I am using 5-man endgame databases. For the forerunner tournament I am using 4-man endgame databases.
   A. Elo- gain = 5 Elo
   B. Additionally 5 Elo for engines stronger in endgames with more knowledge = 10 Elo.

2. With the time-control 40 moves in 12 minutes "repeatedly" engines have much more time in endgames as
   with the time-control 20 minutes game + 5 seconds move "Fischer". The time control I am using for
   FCP Tourney-2021 have a higher value.
   Engines stronger in endgames with more knowledge will win 10 Elo.
   Engines weaker in endgames will be not lost or win Elo.

Simply, experience with own rating-list systems, tourneys I organized for more as 25 years.
Elo- gain for 4-man or 5-man tested out for many years. It's not a secret that the strength from Shredder
engine is the endgame.

January 06th, 2021 (20:00) News-25
*** FEOBOS opening book changed ***

It is too many works to check each of the 41.000 games for bad opening lines if eco.pgn is inside the
opening book. I switched after 774 out of 41.000 games to v2 without the lines from eco.pgn. Three
bad lines after 774 played games only are too much.

Never changed a running system!
OK, OK ...

Again: Please be informed that I checked the first 776 games for bad lines.
Three games are replayed and can be found in my *.zip file.
v1 and v2 from the opening book can be found in my *.zip file.

Settings for: feobos-v20.1-contempt_3-5_tuned-v2.bkt

readme-v2.txt

January 06th, 2021 (05:00) News-24
*** FEOBOS opening lines mixed with eco.pgn ***

A very controversial decision to added eco.pgn in the 100% clear FEOBOS database, free of blunders.
Please have a look in News-18.

Technicality ...

The advantages for eco.pgn:
- better data structure inside the opening book.
- important A00-E99 missed lined are inside.

The disadvantage for eco.pgn:
- not balanced gambit-lines inside the opening book.

With the book setting "Games = 2" 95%-98% of unbalanced new moves, comes from eco.pgn (most gambit-lines),
will be rejected.

This is not the ultimate troubleshooting!
I must check the complete 41.000 games.

I will do that with the universal-pgn-epd-tool v2 by Ferdinand Mosca with:
high-score fcp-tourney-2021.pgn +80 -60

The hight-score-fcp-tourney-2021.pgn output file will be check with the "Live-Book" from Chessbase 16.
The nice tool by Ferdinand Mosca can be found in my download selection.

From the first 569 played games I found 2 games with bad opening lines!

Quite easy to check because FEOBOS lines marked in database with 3000 Elo.
eco.pgn lines marked in database with 2800 Elo.

Located games will be replayed!
I will set in the *.bkt opening book "f" for delete!
The current opening book will be updated from round to round and is inside my *.zip file.

The rejected games (draws below 20 moves, games over 300 moves, bad opening lines) can be found in *.cbh
format in my *.zip file.

December 03rd, 2020 (23:30) News-23
*** The tournament director has to make two decisions?! ***

1. Koivisto
The current available release version "4.00" have small stability problems. I asked for an update and the
authors sent me directly Koivisto 4.19. Sources from 4.18 are on GitHub, sooner or later a new Koivisto
will be available.

Important for myself because I always gave the information:
No private available versions of engines or private engines for my tournaments.

I like Koivisto a lot and my decision are to add Koivisto 4.19 in the field. Stability is most important
for the long time FCP Tourney-2021 tournament.

Elo strength is around the same (v4.00 to v4.19).

I am incredibly happy with Koivisto 4.19, test it today a bit.
My thanks' go to the three authors of Koivisto.

Go Koivisto go ... a great and powerful new development!

2. Nemorino
The current available Nemorino 6.00 with Net_06 is the strongest Nemorino version. Net_14 is an own
trained net and not compatible to Nemorino 6.00. On BitBucked v6.04 sources are available.

The programmer Christian Günther sent me an Nemorino 6.04 PEXT compile and told me that is my decision
what I added in the tourney.

Thanks' to Christian for Nemorino 6.04 and the information I got.

Sleeping about it and I am thinking I should add Nemorino 6.04 with Net_14 in the tourney.

Go Nemorino go ...

Often all is not easy but with the current field of 41 engines I am happy now.

FCP Tourney-2021 will be start in 1 hour!

December 30th, 2020 (12:15) News-22
January 04th, 2021 (00:30) Last changes
*** FCP Tourney-2021 ***

Additional information:

01. FCP Tourney-2021 will be start January 04th, 2021.

02. List of participate engines can be found in News-01.

03. I calculated an average from 3080 Elo for 41 participate engines.
    - for Shredder 13 GUI *.html tournament table
    - for Excel calculation files by Klaus Wlotzka
    - 3080 Elo = Tournament category XXXIV (34)! 

04. Anchor: Shredder 13 POPCNT x64 with 3125 Elo
    - for the rating list calculation with Ordo, Bayesian and EloStat.
    - 3080 Elo average or 3125 Elo anchor to Shredder 13 ...
      End of the tourney both will be produced the same result!

05. Elo range = ~ 575 Elo
    - Topple 0.80 & pirarucu 3.3.5 are probably with ~ 2875 - 2900 Elo the weakest engines in the field.
    - Stockfish 2020-12-31 probably with ~ 3425 - 3450 Elo the strongest engine in the field.

06. LIVE mode with Fling Plus 5.04:
    - *.sto, *.html & *.pgn files will be updated each 30 minutes

07. Replay-Zone
    - from round to round I will add the short won games below 60 moves!

08. Wasp 4.50 is naturally most interesting for us and will be probably fight with:
    Winter 0.9, Arasan 22.2, GullChess 3.0 for place 20.
    We await around 960 - 970 of 2.000 points = 48,0% - 48,5%

09. I await for our opening book: FEOBOS v20.1 Contempt 5 tuned v2
    - 0,15% short draw games below 20 moves, if so ... a sensational result!
    - 1,15% short draw games below 30 moves, if so ... a sensational result!
    - 3,15% short draw games below 40 moves, if so ... a very strong result!

10. We wish all the participate programmers good luck and a happy new year 2021!

December 26th, 2020 (06:15) News-21
December 29th, 2020 (08:00) Last changes
*** FCP Qualify Tourney-2021 ***

The final results!

2845 Elo = Tournament category XXIV (24)!
The cacluation for Elo average can be found in *.zip file.

35. NED (06) (01/2019) TheBaron 3.44.1 (disqualified)

3x time loses for TheBaron after around 280 of 320 games.

December 23rd, 2020 (03:30) News-19
*** New Elo average for FCP Qualify Tourney-2021 ***

2843 Elo = Tournament category XXIV (24)!
The cacluation for Elo average can be found in *.zip file.

December 19th, 2020 (23:15) News-18
January 01st, 2021 (19:15) Last changes
*** Complete new and tuned opening book for FCP Tourney-2021 ***

The opening book for FCP Tourney-2021 is ready.

feobos-v20.1-contempt-5-tuned.bkt
can be found in *.zip file.

readme-v1.txt

December 17th, 2020 (20:00) News-17
*** 3 engines disqualified ***

07. GER (11) (09/2019) ChessBrainVB TCEC 3.74 w32 (disqualified)
08. USA (10) (02/2020) Crafty 23.6 WB (disqualified)
13. ESP (02) (02/2020) Gödel 7.0 (disqualified)

5x time loses for ChessBrainVB after around 170 of 320 games
5x time loses for Gödel after around 170 of 320 games
3x time loses for Crafty after around 170 of 320 games

Engines will play the FCP Qualify Tourney-2021 to end but I disqualified the three for FCP Tourney-2021.

I must do that or I will get to many problems for the next FCP Tourney-2021, 5 months runtime.

After so many years it seems ...
A typical problem for games, runs without resign mode I know from the good and old Winboard times!

After I reduced the participate engines from 46 to 41, the new Elo average is 2830.

50% of FCP Qualify Tourney-2021 is finnished!
My calculation can be found in *.zip file.

The Elo average is important for the Shredder 13 GUI *.html tournament-table.

2830 Elo = Tournament category XXIV (24)!

For John Stanback:

08. Booot 6.4 POP x64             : 3148.01 2000 64.3 55.3 14.88 3034.92 14.28 40.0
09. Lc0 0.26.3 x64                : 3147.87  160 85.0 26.3 52.01 2822.33 35.63 40.0
10. rofChade 2.3 BMI x64          : 3145.31 2000 64.0 51.5 14.78 3034.99 14.28 40.0
11. Laser 1.7 BMI2 x64            : 3121.30 2000 61.0 54.9 14.12 3035.59 14.29 40.0
12. Shredder 13 x64               : 3105.96 2000 59.1 52.6 13.93 3035.97 14.30 40.0
13. Fritz 17 (Ginkgo) x64         : 3105.76 2000 59.1 54.6 14.23 3035.98 14.29 40.0
14. Schooner 2.2 SSE x64          : 3103.39 2000 58.8 57.0 14.72 3036.04 14.28 40.0
15. Defenchess 2.2 POP x64        : 3096.90 2000 58.0 57.3 14.80 3036.20 14.28 40.0
16. RubiChess 1.7.3 x64           : 3093.18 2000 57.5 53.9 13.69 3036.29 14.31 40.0
17. Andscacs 0.95 BMI2 x64        : 3085.37 2000 56.5 54.0 14.43 3036.49 14.29 40.0
18. Fizbo 2.0 BMI2 x64            : 3069.49 2000 54.4 50.6 14.03 3036.89 14.30 40.0
19. Arasan 22.0 BMI2 x64          : 3039.00 2000 50.5 51.9 13.84 3037.65 14.30 40.0
20. GullChess 3.0 BMI2 x64        : 3035.94 2000 50.0 55.0 14.69 3037.72 14.28 40.0
21. Wasp 4.08 dev Modern x64      : 3033.81  160 75.0 38.8 42.81 2825.18 35.86 40.0
22. Fritz 16 (Rybka) x64          : 2999.21 2000 45.3 50.8 13.88 3038.64 14.30 40.0
23. Pedone 2.0 BMI2 x64           : 2990.94 2000 44.2 52.1 13.73 3038.85 14.30 40.0
24. Chiron 4 x64                  : 2988.82 2000 43.9 52.0 13.98 3038.90 14.30 40.0
25. Vajolet2 2.8 BMI2 x64         : 2987.08 2000 43.7 52.9 14.24 3038.95 14.29 40.0
26. Winter 0.8 x64                : 2986.50 2000 43.6 48.1 14.21 3038.96 14.29 40.0
27. Wasp 4.00 Modern x64          : 2985.92 2000 43.5 52.9 13.63 3038.97 14.31 40.0
28. Critter 1.6a x64              : 2975.07 2000 42.1 49.0 13.93 3039.25 14.30 40.0
29. Orion 0.8 NN POP AVX FMA x64  : 2973.87  160 68.4 48.1 39.49 2826.68 35.94 40.0

Results from FCP Tourney-2020 and FCP Qualify Tourney-2021:
With my other test-games (not included) +55 Elo for Wasp 4.08 dev.

John told me for some days, that Wasp 4.50 release version will be ~ 10 Elo stronger as Wasp 4.08.
~ 55-75 Elo gain to Wasp 4.0 ... Fizbo 2.0 is in the near!

Wasp 4.50 will be soon available!

Results after FCP Qualify Tourney-2021:
December 29th, 2020

17. Wasp 4.08 dev Modern x64      : 3050.89  320 74.8 38.4 30.30 2844.06 25.76 40.0
25. Wasp 4.00 Modern x64          : 2988.09 1900 44.9 53.9 12.87 3029.99 12.96 38.0

December 14th, 2020 (16:30) News-15
*** Reduced participate engines from 46 to 41 engines ***

1.
After round 4 out of 8 from the still running FCP Qualify Tourney-2021 I will reduce the participate
engines from 46 to 41. The last five will be taken out.

The reason:
I lost to many time in the first days.

2.
In the main time I send Stefan Meyer-Kahlen (programmer from Shredder GUI) a list of wishes and found
problems. Stefan is at the moment not working on Shredder GUI but for later improvement it can help.

3.
At the moment I am working on the opening book for FCP Tourney-2021.
Will be available in the next few days.

December 05th, 2020 (20:45) News-14
*** Small Shredder 12 / 13 GUI problem ***

Today in the morning I wonder, why the last 36 games from round one missed.
The next problem I have to solve?!

1.035 pairings pro round with 46 engines, each one against each other.
1.035 - 36 missed pairings = 999 pairings!

Shredder 12 / 13 GUI created for each tourney an *.sto file.

Have a look here:
The *.sto started with ...

[TOURNAMENT]
Title=fcp-qualify-tourney-2021
PGNfile=Q:\fcp-qualify-tourney-2021.pgn
HTMLfile=Q:\fcp-qualify-tourney-2021.html
NumPlayers=46
Chunksize=999
Cycles=8
Type=0
MoreVariety=0
PermanentBrain=0
BookLearning=0
OpeningDB=

Chunksize = maximal possible number of pairings!?
Chunksize = 999 is standard? For each new tournament, Shredder 12 / 13 GUI set automatically "999"?

With other words, Shredder 12 / 13 GUI is unable to set a correct "Chunksize" after a tournament is
configurated.

I stopped the tourney, edit "Chunksize" higher, started the tourney again and Shredder 13 GUI started
to load the last missed pairings from round 1 out of 8.

"Chunksize" should be calculated by Shredder 12 / 13 GUI!

OK ...
How many users have interest to start a tourney with so many produced pairings?
A small mistake!
But in the morning at first a new big chaos!
Never before, in all the years, I need the parameter "Chunksize".

December 04th, 2020 (20:00) News-13
*** Deuterium ***

With standard settings (tested on a second PC) or with PawnHash=8, Deuterium 2019.2.37.73 POPCNT x64
produced the same weaker strength under Shredder 13 GUI.

Now I have no ideas, cannot find anything more. Problem is unknown for me.
I believe it have to do with possible UCI settings in combination with Shredder 12 / 13 GUI.

I replaced Deuterium 2019.2.37.73 POPCNT x64 with SmarThink 1.98 AVX2 x64.

The FCP Quality Tourney-2021 produced since days a lot of work.
I hope that all works fine now.

December 04th, 2020 (15:15) News-12
*** Deuterium / Rodent Tal ***

1. Deuterium 2019.2.37.73 POPCNT x64
The results for the still running FCP Qualify Tourney-2021 cannot be right.
I identify Shredder 12 GUI changed defaults???

*.eng file: I changed PawnHash=8 and delete the book file only!

[ENGINE]
Name=Deuterium 2019.2.37.73 POPCNT x64
Author=Ferdinand Mosca
Filename=C:\Chess\Shredder_12\Engines\Deuterium\Deuterium_2019.2.37.73-POPCNT-x64.exe
[OPTIONS]
PawnHash=8 ... OK, what I do
BookFilename=<empty>   ... OK, what I do
KingShelterWeight=0
PawnStormWeight=0
KingAttackWeight=0
MobilityWeight=0
PassedPawnWeightOp=0
PassedPawnWeightEn=0
WeakSquareWeight=0

And Shredder 12 GUI set 7 defaults automatically from 100 to "0".
The reason for the bad results!

Solution:
No solution found with Shredder 12 GUI.
With Shredder 13 GUI it seems all works fine.

A) I changed Shredder GUI from v12 to v13 or I lose Deuterium!
B) 39 Deuterium 2019.2.37.73 POPCNT x64 games will be replayed.

2. Rodent IV 0.30 Tal BMI2 x64 (Personality Tal Test).
The results for the playing style "Tal" are very weak.

I replaced Rodent IV 0.30 Tal BMI x64 with Spike 1.4 Leiden w32.
Again 39 games to replay, sorry!

December 03rd, 2020 (11:00) News-11
*** Sting-SF ***

After 25 games 3x lost on time for Sting-SF 24 x64.

3x too much!

I replaced Sting-SF 24 x64 with K2 0.96 dev x64.
Again 25 games to replay, sorry!

K2 is an unknown engine for me.

December 02nd, 2020 (15:00) News-10
*** Bobcat ***

After 16 games 3x lost on time for Bobcat 8.0 Core2 x64.

3x too much!
No interest that a Trouble-Tourney-2021 are still running 30 days in live mode.

I replaced Bobcat 8.0 Core2 x64 with Quazar 0.4 x64.
Again 16 games to replay, sorry!

Quazar have a fine positional playing style, one of my favorites.
Vs. "modern engines" the results should be better as in the past.

December 02nd, 2020 (09:30) News-09
*** Scorpio / Mr Bob / Lc0 ***

1. I replaced Scorpio 3.0.10 x64 with Mr Bob 0.9.0 POPCNT x64.
"Neural Network" with Shredder GUI is for Scorpio not possible and without NN the results are under
expected performance.

2. I am not sure to set the right settings for Lc0 0.26.3 x64 and made two correction in engine
configuration. Games will be replayed!

3. I corrected the FCP Qualify Tourney-2021 average of Elo from 2840 to 2820.

4. 3x lost on time games (1x Hakkapelliita, 2x Bobcat) replayed.

December 01st, 2020 (10:45) News-08
*** Orion ***

Last-minute changes:

I replaced Orion 0.7 NNUE BMI2 x64 with Orion 0.8 NN POP AVX FMA x64.
The new version released hours later I started the tourney. 

November 30th, 2020 (05:00) News-07
*** Time control, stability test 2 ***

Beef 0.3.6 produced to many "Lost on time" games.
The same with the time control 40/10, 40 moves in 10 minutes "repeatedly"!

I replaced Beef 0.3.6 with Protector 1.9.0.

November 30th, 2020 (01:00) News-06
December 14th, 2020 (16:30) Last changes

*** Time control, stability test 1 ***

A bad idea ...
Fischer time control in combination with Winboard engines and WB2uci adapter!

Winboard Engines have problems with the time management if I set the time control to
20 minutes game + 5 seconds move "Fischer".

After ~ 60 moves, engines like ProDeo 3.0, Francesca 0.29a others have 10 minutes left on
the clock. 50 Elo can be lost with such a time management!

UCI engines like Beef 0.3.6, BobCat 8 lost continuously on time, other UCI engines have
smaller time management problems with the Fischer time control.

I have to change the time control, or I lost to many engines for FCP Qualify Tourney-2021.

A) Time control for FCP Qualify Tourney-2021:
   40/12, 40 moves in 12 minutes "repeatedly", 08 rounds

B) Time control for FCP Tourney-2021:
   40/12, 40 moves in 12 minutes "repeatedly", 50 rounds

November 17th, 2020 (22:15) News-05
December 04th, 2020 (20:00) Last changes
*** Elo calculation ***

The Elo range between the strongest 41 FCP Qualify Tourney-2021 engines is ~ 475 (Lc0 to Cheese).

I believe GreKo 03.2020 is the weakest (not sure because the playing strength for Open Rodent 1.0
is questionable) and Wasp 4.08 dev with Lc0 0.26.3 CPU the strongest engines in the field. 

I calculated the Elo average of 46 engines with results from CEGT, CCRL, FGRL and my own
with 2840 = Tournament category XXIV (24)!

11 engines will be qualified for FCP Tourney-2021!

1. 2820 Elo: I set for Shredder GUI tournament table!

2. Results from FCP Tourney-2020:
   2968 Elo: Nirvanachess 2.4
   2939 Elo: Protector 1.9.0
   2887 Elo: SmartThink 1.98
   2859 Elo: Monolith 2.01

    Can be use for the first Ordo 1.2.6 Elo-calculation!

November 15th, 2020 (00:30) News-04
December 14th, 2020 (16:30) Last changes
*** Opening book, Shredder GUI settings ***

For FCP Qualify Tourney-2021 I changed the opening book!

FEOBOS project

feobos-top-10500.bkt

based on:
FEOBOS TOP 10.497 positions

 additionally:
C38, best C38 position on ranking: 24.308
D84, best D84 position on ranking: 15.458
E27, best E27 position on ranking: 10.589

Positions = 10.500
A00-A99 = 2.274 = 21,658%
B00-B99 = 2.920 = 27,810%
C00-C99 = 1.604 = 15,277%
D00-D99 = 1.938 = 18,458%
E00-E99 = 1.764 = 16,800%

Move-Range = 4-17
Move-Average = 8,675

   Opening book settings for Shredder GUI:

    Note:
   1.c4 and 1.Nf3 should not have more as ~ 8% or to many A00-A99 games will
   be the results. Around 100 of 10.500 position will be lost with this
   setting but that's not important.

    Note:
   With "Num.Games" you can adjust it!

    Note:
   Shredder GUI main settings!