Lichess4545 Ledger 088

Lichess4545 Ledger

Issue #088 - February 12, 2019


Survey summary

by @mn8


To wrap up Season 15, I'll be recapping the End-Of-Season Survey, which saw eighty respondents. As always, these are (mostly) all much appreciated.

Nearly 70% of people found the league "very enjoyable" this season, up a whole 10.9% from the previous edition. Meanwhile, a whopping 98.7% of survey participants rated its enjoyability a 4 or greater.

Meanwhile, opinions regarding team balance/size, league size/length/format do not appear to have seen much shift. (In spite of this, the mods have decided to go with eight-player teams this season, but that's another story entirely).

The league's favourite team name this season, in a landslide, was "Caruana leaked our prep", which managed to avoid repeating worn-out puns by going for a topical meme. "What is Luft? - Baby don't back-rankme", "Magnus, when did u understand it was a draw?!", and "Resign Against the Machine" came second and equal third, respectively. Personally, I'd like to shout out ""Ruy Lopez Dispenser" for being a bizarrely specific pun I don't think I would ever have come up with.

Worth noting, I think, is a recurring complaint about alternate life: not always being invited to the team channels. Personally, as a former/future (?) alt, I can attest to this (occasionally) being a problem. It's been suggested that this process be automated, should this is possible and/or convenient.

Reddit remains the most common source of participants for our league, but it seems over time fewer and fewer people learn about it from there. "On lichess" is holding steady at about 28-29%, but "through a friend" and "streamer or reviewer" are becoming increasingly common.

It also seems there's reasonably high demand for the ledger to come out more frequently, so submissions are always appreciated!


Regardez Lichess 45+45 Season 15 | Game Analysis | {} de Astaneh sur

Quality of Play in Season 15

by @Erinyu


Hey, my name is Erin. I investigate stuff nobody that bothers with. Let’s get into it.

How high rated is every board?

Player rating, obviously, increases with the higher boards. Glancing at Figure 1, the rating spread is fairly consistent for boards 2-5. The middle 50% of players stand within 50-80 rating of each other, but something’s slightly awry with boards 1 and 6. For board 6, the interquartile spread is 119 rating, and for board 1 it’s 144 rating. This doesn’t surprise me; those two boards catch the players on the ends of the rating distribution.

Figure 1. Box and whisker plot of player ratings from every game played on board 6 (left) to board 1 (right).
“X” represents the mean of each board. Outliers excluded.


Amusingly, sometimes 1700-rated players play on board 6, 1800-rated players play on board 1, and there’s quite a bit of overlap between all the other boards due to players playing up/down and simply having imbalanced teams. But, in general, the board ratings are consistent. It'll be interesting to observe how season 16's rating distributions compare to this one.


How good are the moves?

As expected, board 1 players play the best. But according to Figure 2, the difference between board 1 and board 6 players is about 7 top three choices every 100 turns. Is that so significant? It definitely helps to play good moves, but you can still win without them; just look at @lovlas.

Figure 2. Percentage of top 1, top 2, and top 3 moves, when distinguishable, in roughly even or unclear positions for every board.


What is significant, however, is the number of blunders per game. In Figure 3, the difference between board 1 and board 2 is just about 1%; that means board 2 players will, on average, make one extra blunder per 3 games, which precipitates a huge hit to their rating. This reflects in the ACPL scatter; board 2 players will find themselves down 1 pawn in ~20-25 moves.

Figure 3. Percentage of moves that are blunders exceeding 100 centipawns, and average centipawn loss, in roughly even or unclear positions for every board.


What I’m trying to say is, board 5 and 6 players can expect to be board 1 players by halving their blunder probability. However, I’m not sure how easy that is.


How strong are our players?

And since people feigned interest last time, Table 1 is my Classical→FIDE rating correlation. I used several thousand FIDE-rated games from various playing levels to map the level of play in our League, from seasons 10 to 15, to its estimated OTB level. There exists a curvature around 2000 Classical, so the upper ratings were partially regressed with the verified FIDE ratings of the strongest players in the League. These numbers are probably good to +/-200. I suspect there are some systematic biases that deflate the estimate for lower ratings, for example the speed of 45+45 relative to 90+30.

Table 1. Erin’s Classical→FIDE model. The USCF and ECF estimates are a function of FIDE rating and are derived from Glickman’s correlations found on the internet.

I would greatly appreciate any feedback you have to give relating to this model or anything else in this segment. Thanks for reading!


Special thanks to @lakinwecker for guiding me through some of the data reporting and usage.

Special thanks to @colwem for proofreading and prompting me to investigate the curvature.

Special thanks to @toni4127 for proofreading and egging me on to some of my ideas.




Click "Open" to discover the experience of the multi-chaptered and well worked interactive analysis of BigEvo8 following the journey of his team.



Click "Open" to follow the games of okei.



Weekly Stats

by @kraaft and @Somethingpretentious



Overall Season 15:

  • The fastest mate was white on move 14 found in Gamelink White: kilderis, Black: infiniteflash, Gamelink White: pepepibote, Black: shirovgambit.
  • The fastest draw was found in Gamelink White: nondrama, Black: sentinelcrush.
  • The fastest resign was on move 9 found in Gamelink White: spiteknight, Black: fischersofmen, Gamelink White: luckyara, Black: footpath.
  • The biggest upset was 223 points in Gamelink White: board_13, Black: asn8745.
  • The longest game ended with white on move 101 Gamelink White: ben_dubuque, Black: nategreene.
  • 140 was the highest ACPL in Gamelink White: fiveknights, Black: mathijshuis.
  • 6 was the lowest ACPL in Gamelink White: sentinelcrush, Black: f1nn33, Gamelink White: dracoxvitae, Black: lakinwecker, Gamelink White: somethingpretentious, Black: tushar20008.
  • Combined maximum ACPL was 252 in Gamelink White: fiveknights, Black: mathijshuis.
  • Combined minimum ACPL was 12 in Gamelink White: sentinelcrush, Black: f1nn33.
  • The longest "think" was 34 minutes 54.0 seconds on move 9 in Gamelink White: kostasvl, Black: mysonnevercallme.
  • The most time left was 80 minutes 53.0 seconds in Gamelink White: ben_dubuque, Black: nategreene.
  • The most time spent was 111 minutes 21.0 seconds in Gamelink White: halfright, Black: erinyu.



AWARDS (You only qualify if you played 5 games.)

  • Accuracy King: with an average of 18.7 ACPL, Erinyu.
  • Fast Finisher: with an average gamelength of 40 ply/20 moves, itakeback.
  • Marathon Man: with an average gamelength of 127 ply/63 moves, Razorneck.
  • Giri Award: 5 draws, roso97 and mattyb.
  • Biggest rating gain: with 186 rating points, kschmigel.
  • Biggest rating loss: with 120 rating points, ask_the_dust.
  • Fun fact: The most popular opening played was:
  1. A45 Indian Game (28 played)
  2. D00 Queen's Pawn Game (22 played)
  3. B01 Scandinavian Defense (20 played)


MEDALS To see how you stack up: (B1 - B2 - B3 - B4 - B5 - B6)





Board 1 ErinYu 2373 Assassin_in_White 2361 mysterious_expert 2352
Board 2 bufferunderrun 2173 NoNameK 2156 lumenian 2140
Board 3 DracoxVitae 2604 RealRobbert 2095 stinsonite 2093
Board 4 calcium237 1923 Lethgo 1921 TraderLynch 1918
Board 5 RodrigoDiazDeVivar 1969 Jobwoge 1933 Mahmut69 1928
Board 6 oddsox 2305 Jokes1724 1904 Lalaze 1898


Database Analysis of Team4545 #15

                   Games     1-0     =-=     0-1    Score
 All report games    673     299     104     270    52.1%


Average Rating and Performance

White rating: 1809 White performance: 1825 (52% vs 1811)
Black rating: 1811 Black performance: 1795 (48% vs 1809)


Frequency of Themes in the First 20 moves of each game

   Same-side castling:             62%    White Isolated Queen Pawn:      12%
   Opposite castling:              10%    Black Isolated Queen Pawn:      10%
   Kingside pawn storm:            18%    White Pawn on 5/6/7th rank:     51%
   Queens exchanged:               47%    Black Pawn on 2/3/4th rank:     31%
   Only one side has Bishop pair:   6%    Open c/d/e file:                57%


Opening Move

    Move   ECO       Frequency    Score  AvElo Perf AvYear %Draws
 1: e4     B00a      360: 53.4%   52.7%  1810  1829  2019   16%
 2: d4     A40a      246: 36.5%   51.6%  1803  1829  2019   17%
 3: Nf3    A04        29:  4.3%   51.7%  1799  1832  2019   14%
 4: c4     A10        26:  3.8%   46.1%  1784  1730  2019    8%
 5: f4     A02         6:  0.8%   33.3%  1893        2019    0%
 6: g3     A00t        4:  0.5%  100.0%  1841        2019    0%
 7: b3     A01         1:  0.1%    0.0%  2326        2018    0%
 8: Nc3    A00l        1:  0.1%  100.0%  2301        2019    0%
TOTAL:               673:100.0%   52.1%  1809  1825  2019   15%


See the stats page for win percentages for each board by color and ratings distribution.



Stats for Season 14 Rounds 1 to 6:

  • The fastest mate was black on move 8 found in Gamelink White: kamekura, Black: isachess.
  • The fastest draw was found in Gamelink White: merkava777, Black: cerulean.
  • The fastest resign was black on move 10 found in Gamelink White: tcgrif, Black: glbert.
  • The biggest upset was 331 points in Gamelink White: daimonion, Black: darubaru.
  • The longest game ended with black on move 98 Gamelink White: felipekaox, Black: sirnatealot.
  • 205 was the highest ACPL in Gamelink White: quanlentpyke, Black: scv-chess.
  • 1 was the lowest ACPL in Gamelink White: mad_shaving, Black: daimonion.
  • Combined maximum ACPL was 330 in Gamelink White: rooooks, Black: coastalowl.
  • Combined minimum ACPL was 22 in Gamelink White: gross_patzer, Black: atcfaust.
  • The longest think was 21 minutes 34.0 seconds on move 18 in Gamelink White: qrs, Black: izzie26.
  • The most time left was 50 minutes 59.0 seconds in Gamelink White: lcur15, Black: drodnat.
  • The most time spent was 68 minutes 38.0 seconds in Gamelink White: glbert, Black: ringelnatz.



Chess Puzzles

Click on the images for the solution.


aebe (1705) - xRJGDx (1615)
◯ White to play.
aebe (1705) - xRJGDx (1615)
⚫ Black to play.
dmavro (1841) - MattyB (1848)
◯ White to play.
Lelouch_Vi_Brittania (2179) - SheepNL (2094)
◯ White to play.
lapetitemort (1617) - Lukhas (1725)
◯ White to play.
robertswa (1653) - halloway (1708)
◯ White to play.
jobwoge (1715) - Showboater (1768)
◯ White to play.
CaptNCarter (2007) - urjah (1969)
◯ White to play.
HalfRight (2087) - eie24 (2065)
◯ White to play.
robertswa (1626) - Ontheruy (1682)
◯ White to play.


Hoikka (1787) - Carschi (1621)
⚫ Black to play.
sanganakam (1570) - xEmpCC (1733)
◯ White to play.
kostasvl (1710) - SouzaMoraes (1546)
⚫ Black to play.
EricTheodoreCartman (2049) - kajoM (1848)
◯ White to play.
a12233 (1481) - Delpire (1676)
◯ White to play.

Shared by the community (can contain imperfect puzzles.)

ebrankoftis (1812) - Tranzoo (1724)
Shared by @Tranzoo.
⚫ Black to play.

Chessboard images provided by lenik terenin.

Please feel free to join #lichessledger on slack if you would like to contribute towards the ledger in any way, or provide any feedback. Both are highly encouraged and appreciated. Thank you for reading.


Creative Commons License

Lichess4545 Ledger #088 ©2019 by Thienan Nguyen is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License

Thanks to @bigevo8, @erinyu, @forhavu, @kraaft, @mn8, @Somethingpretentious and @SpiteKnight.