Russell Sherwood

Back in the Engine Room

Russell Sherwood  Thursday, December 13, 2018

It’s been an interesting couple of months in terms of engine development. Probably the two biggest news items are Stockfish 10’s release and the advance of LCZero.

 

Recently SF10 was released, with a significant ELO gain over SF9. One thing to watch for CC players is that the contempt value has changed and the setting of this should be considered prior to use.

 

LCZero has been gaining strength in self-play recently but it remains to be seen if this will pass through into useful improvements. The significant news of recent days is that another AlphaZero paper has been released with a significant amount of actionable information. This is being put into practice by the Alpha Zero team at the moment. Unless you have suitable hardware LCZero remains interesting rather than competitive for CC players but this situation requires keeping close attention on.

 

Development of a CC-specific variation of Sugar has commenced. This is a very interesting project and one which could prove significant benefits.

 

Players’ favourite Raubfisch continues to develop incremental improvements coming on a regular basis.

 

Sting, whilst not a frontline engine, continues to develop and is of use in cracking difficult positions.

 

Komodo has released a new version of the MCTS equipped version with around a 100 elo gain being observed, coming close to the standard Komodo version.

 

CFish and CiFish continue to keep pace with new development versions of Stockfish, however, AsmFish has fallen significantly behind the pace. A spin-off version AsmFishCP has been developed. This has a number of Stockfish improvements implemented but is still significantly behind the pace.

Finally, for now, SHashChess has been updated. Whilst still only "interesting" in terms of output, it is still a useful addition to any stable.

 

New Title Tournaments

Russell Sherwood  Wednesday, December 12, 2018

On January 1st the biggest shake-up in ICCF tournaments for a number of years is taking place with the Introduction of the New Title Tournaments.

Historically ICCF offered two "open entry" events - the Master Norm for players over 2300 and the Grandmaster Norm for players rated over 2400. These tournaments served well for a number of years but an increasing frustration was seen in recent years driven by a number of issues:

  • The introduction of the CCE and CCM Titles, which whilst some federations took a lead to organise their own qualifying events, were not available for open entry.
  • Modifications made to the GM Norm tournament to facilitate the obtaining of SIM Norms had the negative effect of making GM Norms much harder to obtain.

Following an in-depth analysis of the situation, a proposal was presented at the Llandudno Congress which was accepted and comes into operation on Jan 1st.

So what has changed? First, instead of 2 Tournaments, there are now 10 events, 2 each focused on GM, SIM, IM, CCM and CCE Titles.

The first thing to notice is that the entry requirements for the GM events are higher than before. A GM Norm is possible in the SIM-A and possible SIM-B event but these will be much more difficult to obtain, the same being true for each subsequent lower tier.

So what is the benefit of this? For each Tier, we have a much narrower rating, which for the 'A' events means that, typically, the event is a Category higher and hence requiring a slightly lower score (generally one less win) for the Title Norm. This will still be difficult to achieve but this reorganisation should make the obtaining of titles slightly easier. 

Another benefit, which may not be immediately apparent is that events in which "elo leeching"  is not really possible in these events!

 

So have a look and make your choice! If you want any advice on these events, please get in touch.

  Low High Mid Cat Games
GM-A 2485   2485 10+ 15
GM-B 2455 2484 2469.5 9-10 13
SIM-A 2420 2454 2437 7-8 15
SIM-B 2380 2419 2399.5 6-7 13
IM-A 2350 2379 2364.5 5 13
IM-B 2300 2349 2324.5 3-4 13
CCM-A 2250 2299 2274.5 1=2 15
CCM-B 2200 2249 2224.5 K-L 17
CCE-A 2125 2199 2162 H-I 15
CCE-B 2050 2124 2087 E-G 15

2019/1 Rating List Review

Russell Sherwood  Wednesday, December 12, 2018

The 2019/1 Rating list has thrown up some interesting moves......one of the most interesting is a comparison to 5 years ago

2019/1   2014/1
4 2400+ 0
10 2300-2399 2
11 2200-2200 3
4 2100-2199 4
4 2000-2099 2
33   11

Back to the 2019 List

Top Gainers

810256   Jones, Phillip 42 1926 1857 69
810222   Smith, Andrew 118 2175 2128 47
810220   Davies, Jonathan 26 1727 1688 39
810251   Soar, Timothy 28 1744 1711 33
810212 CCE Sherwood, Alexander 352 2337 2308 29
212421   Robinson, Nicola 20 1699 1672 27
810257   Burrows, David 36 1675 1655 20
810259   Denton, Sean 45 2145 2126 19
810180   Nettles, Paul Eric 56 2216 2198 18
810139 CCM Adams, Mark 167 2361 2348 13
810198   Thomas, Dennis 24 1595 1582 13
810199   Dahlgren, Paul P. 37 1480 1467 13
810163 CCM Keevil, Paul 128 2329 2318 11
810156   Ricketts, Martin 44 1828 1817 11
810195   Bevan, Peter M. 97 2148 2138 10

 

Biggest Sliders

810181   Hallwood, Reg 22 1470 1544 -74
810075   Wakeham, Marc R. 467 2051 2121 -70
810102   Guy, David T. 234 1580 1639 -59
810255   Morgans, Justin 41 1982 2027 -45
810254   Jones, Steve 39 1400 1439 -39
810202 LGM Sherwood, Helen 645 2337 2372 -35
810133   May, Adam J. 40 1845 1872 -27
810165   Bailey, S. 118 2104 2127 -23
810217   Richards, Michael 231 1322 1344 -22
810192   Baron, Margaret 184 1826 1846 -20
810176   Chugg, Malcolm 122 1301 1318 -17
810161 CCM Sherwood, Russell 761 2336 2348 -12
810177   Wilding, Arthur 60 1499 1511 -12

 

(Personally, Helen and I are happy to see almost the back of our "Annus Horribilis" and commencing a Dunlop manoeuvre!

 

Congratulations to Adam Davies on breaking the 2300 barrier  and Eric Nettles the 2200 barrier

 

 

Time Machine

Russell Sherwood  Sunday, October 14, 2018

Occasionally in a game, we can have the problem of an opponent wishing to delay the outcome,  In more recent times the laws of CC were changed to introduce the "Code of Conduct". What constitutes a valid COC claim is quite complicated and the player is advised to review the ICCF Rules on the matter

A tool I knocked up at the time is a timeline visualiser - this is a simple excel sheet which can simplify the determination of one of the key parts of COC.

1) Copy your timeline into the sheet

2) Copy/ delete the  formulas

3) Add any vacation into the sheet

4) Review the timeline - some basic methods used people to delay results become very obvious on these charts.

 

 

Download

WCCF Template

Russell Sherwood  Thursday, October 11, 2018

WCCF Powerpoint Template

Download

29th ICCF World Championship

Russell Sherwood  Sunday, September 30, 2018

This event finished recently https://www.iccf.com/event?id=52852

Much has been made of the draw% but I wondered what other conclusions could be drawn?

9 Wins for White, 123 Draws and 4 Wins for Black

  Games Game% White Perf%
B90 24 18% 55% Sicilian Najdorf
D43 12 9% 55% Semi-Slav
C65 8 6% 42.50% Spanish- Berlin
C89 7 5% 50% Spanish-Marshall
E97 6 4% 58% King's Indian
B97 5 4% 50% Sicilian Najdorf -PP
E54 5 4% 50% Nimzo-Indian
E15 4 3% 50% Queen's Indian

 

This is slightly surprising, especially the King's Indian and Spanish Marshall and the much lower level of Spanish Berlin than expected.

There are plenty of other interesting nuggets that can be extracted but that is your job!:

 

 

Keep an Eye

Russell Sherwood  Sunday, September 30, 2018

Something worth keeping an eye on!   A neural net-based focused on playing 12 pieces or less. Whilst not challenging yet it may become "something"

 

https://github.com/dkappe/leela-chess-weights/releases/tag/Ender38

Slayers

Russell Sherwood  Monday, September 24, 2018

One of the first dividends of working with Welshbase has been an examination of results against Grandmasters

First against we have 8 results - all draws! Almost all of these came in an India-Wales Friendly in 2015 where the general approach was to take easy draws for the rating points!

Moving onto ICCF-GM - we have 68 games with a combined record of 7 wins, 49 draws and 10 losses, so 48% in Total. However, this hides a number of other facts:

5 players have positive scores against Grandmasters, with a further 4 on 50%

3 players have more than 10 games against Grandmasters (when FIDE games are taken into account) 

From this the conclusion, the potential for a Welsh ICCF-GM at some point in the future looks positive!

 

 

 

 

 

 

SICALM #3

Russell Sherwood  Sunday, September 2, 2018

It has been an interesting month, ICCF Congress notwithstanding (For which I will produce a separate report with some of the interesting ideas I picked up there!).

7 Piece Tablebases.

These have now been completed for download! The only issue is that they are about 17 TB in size. This is about the size of 9 typical laptop hard drives! At the moment unless you have superfast fibre or a team effort takes place, these would be difficult to download (and not all chess engines will work with them yet) but you can pick selected ones which fit the positions you seek!

TCEC and CCCC

These engine v engine tournaments are interesting to the serious CC player. Whilst the addition of a human changes things significantly these give a good indication of the improvements and issues of the metal monsters. The two most interesting developments to date have been around Ethereal and Neural Network engines (LCZero and Deus X). Ethereal as it is a new kid on the block and may add something to our analysis. Deus X was also interesting as whereas LCZero "learned" chess from scratch, Dues X learned from Human games. From our perspective, two possible research avenues jump out: (1) A NN based on CC games and (2) a NN based on a certain type of position. I would expect to see both of these appear in coming months.

CorrChess.ru

Turn on your translator for this! Early days yet but some interesting thoughts from some very strong CC players

www.tccmb.com

Even earlier days yet but this revival of an old CC discussion board could be interesting!

ICCF Congress decisions 2018 for Welsh Players

This article is available to federation members only; please log on to view the content.

Welsh Correspondence Chess FederationBritish Correspondence Chess AssociationSchemingMind Internet Correspondence Chess ClubSocial Correspondence Chess AssociationWelsh Chess UnionInternational Correspondence Chess Association