Reverse Komi; New Players Welcome! Registration OPEN through the 7th

Lagrange Points: A Running Reverse Komi Arena (New Players Welcome!)

Lagrange Points is here to provide you with the opportunity to arrange games with full reverse komi with like-minded individuals. With flexible commitment and scheduling requirements, it is my hope that a wide selection of people who may not be inclined or able to commit to a more rigid structure, can still participate in Lagrange Points.

Reverse komi is an elegant alternative to handicap stones: instead of the handicap taker receiving extra moves at the beginning of the game, the handicap taker receives extra points at the end of the game. Since this applies cumulatively with komi, either the handicap giver or the handicap taker may be Black in any given match. Regarding its impact on gameplay, this has both advantages and disadvantages compared to stone handicaps.

Each player gets to choose each round how many games they wish to sign up for and will be paired accordingly insomuch as is possible at discretion. Once paired, each player simply finds a time when their opponent(s) are also available, plays the game with the correct settings (correct komi value and color assignment will be listed next to each pairing), and submits the result.

If both players of a given match agree, they can play the relevant game with any Fischer, byo-yomi, or Canadian time control in lieu of the default; just make sure the other settings match those listed in “Game Rules”. Also keep in mind that the game needs to be completed by the end of the round and that there is neither obligation nor prohibition to respect pauses, so set time controls accordingly.

Each round lasts three weeks, with registration being open for a week during the end of the previous round and closing when I generate the pairings for the new round on a Tuesday. All players wishing to sign up for a round must do so explicitly and unambiguously regardless of previous participation or lack thereof. Forum pm @Samraku is the most reliable way to do so, closely followed by OGS pm @Samraku. As the rounds are completely independent of eachother, you can stop signing up for them at any time: Lagrange Points may be thought of as a new mini-arena every three weeks.

More Info

Anyone who wants to participate in an upcoming round, regardless of whether or not they are playing in the current round, must explicitly sign up by contacting @Samraku with their OGS username and the number of games they wish to play. You can sign up for as many or as few games as you wish, though if you sign up for too many games relative to the total participation, you may end up registered for fewer in order to avoid pairings against oneself which would be… a bit uninteresting to most.

There is one restriction on the number of games a player may sign up for if they lost at least one game by forfeit in the previous round in which they played. In that situation the player in question may only register for up to the number of games in the previous round in which they played for which they obtained a result other than loss by forfeit. In the event that a player loses all games in a round by forfeit, they may still register for a single game in a subsequent round. Only the last round in which a player participated counts for this, so once a player successfully completes a round, all restrictions on registration are lifted.

Every third Tuesday (US ET), pairings and correspondent komi values will be posted in this thread, signalling the beginning of a new round. Players then have the next 20-odd days to arrange their match(s) with their opponent(s), play them, and submit the results.

The pairings and results are here. It’s a wiki, so please feel more than free to edit in your session times and results so other people can watch the game either live or after the fact. Please be sure after each match to replace the word “result” in the relevant pairing with the correct value to avoid the game accidentally being counted as a loss by forfeit for the purposes of registration.

This arena is named after Lagrange Points, points at which the forces acting upon an object in space roughly cancel out allowing the object to apparently just sit there unmoving, just as the reverse komi perfectly cancels out the skill differential of two players.

It is directly inspired and based off of clyring’s Endless Endgame Event which I would highly encourage anyone interested in Arimaa to check out.

All rules subject to change: please comment in post or pm if you believe I have overlooked something in these rules.

Mirror Go?

It has been brought to my attention that mirror go could be very easily abused in reverse komi games. The handicap taker could force the handicap giver to spend the better part of the opening, when they really need to start regaining the reverse komi, breaking the mirror go. In light of this, mirror go is not allowed in any Lagrange Points games. If you suspect your opponent of using this strategy, please bring it to my attention either during or after the game, and I will evaluate it using the following objectively evaluable rules. It is my aim here to prevent mirror go without unduly restricting other possible strategies.

  1. def.: A move is said to create a 2-axis mirror symmetry if, after it is played, it is the case that for every stone on the board and not on tengen, there exists a stone of the opposite color such that a straight line segment may be drawn connecting the two stones which is bisected by tengen.
  2. def.: A move is said to create a 1-axis mirror symmetry if, after it is played, for either or both of the 2 central lines bisecting the board and parallel to at least one edge of the board, it is the case that for every stone on the board and not residing upon said central bisecting line, there exists a stone of the opposite color such that a straight line segment may be drawn connecting the two stones for which the central bisecting line is a perpendicular bisector.
  3. def.: A move is said to create a mirror symmetry if it satisfies (1) and/or (2).
  4. This section shall impose no restrictions upon the first 9 moves of the game, that is, 5 Black moves, and 4 White moves. Put another way, the restrictions herein apply beginning on move 10, which is a White move. Thus, players may play all star points if they wish, among other various openings which would be unintentional victims of this section if not for this rule.
  5. This section shall impose no restrictions upon a move for which it is the case that the second to last move, that is to say, the previous move of the player to play, did not create a mirror symmetry. In other words, players may create mirror symmetries past move 9 so long as they play at least one move which does not create a mirror symmetry between each move creating a mirror symmetry. Thus, players may achieve symmetrical positions, perhaps through identical joseki choices.
  6. For any move not covered by (4) or (5), it is forbidden to play a move which creates a mirror symmetry.
Pairing Method

This section subject to frequent change as Lagrange Points spins up and finds itself.


  • Match all players for the number of games they registered.
  • Avoid pairing the same players twice in a round, especially with the same colors.
  • Provide a variety of rank differences and thus reverse komi values, while not being too extreme. What is too extreme? Should matches of very similar rank be avoided as they do not provide much that any other game on OGS does? Should it be completely random and only disallow flatly impossible reverse komi values? That all remains to be seen: leave your feedback in the comments section.

Current algorithm:

  1. Randomly pair players who have yet to be paired for their desired number of games until there is only zero or one players left who haven’t had all their games matched.
  2. Repeat until this is zero players (all players have been paired for all their games) (for now I’ll prevent parity by registering for an even or odd number of games myself to keep it even).
  3. Manually check that there’s nothing too crazy going on in the pairings, like the same two people playing eachother more than once, or impossibly high komi.

This algorithm can stand to be improved immensely; feel free to suggest improvements (aside from automating step 3, which is obviously on the TODO list, but might sit there a while since it’s not too much work to just run the pairings a few times for now).

Game Rules

  • Unranked
  • Board Size: 19x19
  • Rules: New Zealand
  • Time Control: Fischer
    • Initial Time: 30 minutes
    • Time Increment: 15 seconds
    • Max Time: 1 hour
    • (or other Fischer/byo-yomi/Canadian controls by agreement)
  • Handicap: None
  • Komi: As Indicated in Pairing
  • Your Color: As Indicated in Pairing
  • Analysis Disabled
Allowed Aids


  • Game Chat/Malko Log
  • Paper Kifu
  • Pencil/Paper, or equivalent (these may hold the current or earlier game state(s), but should not contain any future game states)


  • Playing out variations on electronic and/or physical boards
  • Consulting with other humans regarding the game
  • Using AI (including but not limited to LeelaZero, Katago, and score estimators)
  • Using position and/or game databases (including but not limited to Waltheri and joseki or fuseki dictionaries)

How to Schedule and Play a Game

  1. Contact the opponent. (pms on OGS and/or these forums are recommended)
  2. Discuss when and with what time settings you and your opponent can play.
  3. Come to an agreement regarding both a time and date, and time settings.
  4. Post the agreed time by editing the pairing post.
  5. Show up at the agreed time, send or accept the agreed challenge, and play.
  6. Report the result by editing the pairing post.

All games need to be completed and reported in the pairing post by the Monday (in US ET) three weeks after the beginning of the round in order to be guaranteed to be counted.

If the two players in a given match are unable to arrange a mutually agreeable game time despite reasonable effort on the side of both parties, they can request an adjudicated draw. This will not limit the registration rights of either party in future rounds.


In order to join, pm me (@Samraku) either here (most reliable) or on OGS, or post in this thread during the registration week leading up to a round. Make sure to include your OGS username (link to profile is appreciated, especially if you have non-ASCII characters in your username) and the number of games you’re registering for.

Round 2021-5 (5) is open for registration through Mon., 7 Jun., 2021 and begin on the 8th.


Pairings and Results

Round 2021-1 (1) (Mar. 16 — Apr. 5)

Samraku [2k] (B) vs. teapoweredrobot [6k] (W; 63 komi) (B+100)
vyzhael [9k] (B) vs. He Who Walks in Shadows [2k] (W; -85 komi) (W+R)
riiia [5k] (B) vs. wurfmau3 [4d] (W; -106 komi) (B+R)
Kaworu Nagisa [5d] (B) vs. illusory_deceit [10k] (W; 207 komi) (result)
DashaTabasco [13k] (B) vs. LittlePebble [21k] (W; 124 komi) (result)
KoBa [2k] (B) vs. DashaTabasco [13k] (W; 161 komi) (result)
He Who Walks in Shadows [2k] (B) vs. wurfmau3 [4d] (W; -74 komi) (W+R)
riiia [5k] (B) vs. KoBa [2k] (W; -31 komi) (W+R)
Samraku [2k] (B) vs. KoBa [2k] (W; 0 komi) (B+31)
sbk96 [9k] (B) vs. wurfmau3 [4d] (W; -165 komi) (W+16)
riiia [5k] (B) vs. teapoweredrobot [6k] (W; 32 komi) (B+26)
wurfmau3 [4d] (B) vs. vyzhael [9k] (W; 180 komi) (B+20)
wurfmau3 [4d] (B) vs. DashaTabasco [13k] (W; 236 komi) ( B+R)
Samraku [2k] (B) vs. wurfmau3 [4d] (W; -75 komi) (B+R)
wurfmau3 [4d] (B) vs. pdg137 [8k] (W; 167 komi) (B+2)
He Who Walks in Shadows [2k] (B) vs. Samraku [2k] (W; 8 komi) (W+R)

link to KoBa

Round 2021-2 (2) (Apr. 6 — Apr. 26)

B W komi result
wurfmau3 [10d] Samraku [4d] 89 W+R
Samraku [4d] wurfmau3 [10d] -75 B+R
pdg137 [9k] teapoweredrobot [10k] 15 W+34
Samraku [4d] teapoweredrobot [10k] 187 W+R
teapoweredrobot [10k] wurfmau3 [10d] -255 optional

Round 2021-3 (3) (Apr. 27 — May 17)

B W komi result
wurfmau3 [4d] Samraku [1d] 45 B+R
Samraku [1d] wurfmau3 [4d] -31 W+R
teapoweredrobot [4k] wurfmau3 [4d] -92 W+R
teapoweredrobot [4k] vyzhael [8k] 51 W+2
pdg137 [8k] wurfmau3 [4d] -135 B+4

Round 2021-4 (4) (May 18 — Jun. 7)

B W komi result
Samraku [1d] pdg137 [8k] 122 B+129
martin3141 [5d] wurfmau3 [4d] 18 B+51
wurfmau3 [4d] vyzhael [8k] 151 result
wurfmau3 [4d] Groin [1k] 66 result


Relevant links will be posted here to aid in navigation to relevant resources.

OGS Groups

I’m up for this. I’m building up to trying live games again and like reverse Komi in principle.
I’m not completely clear about the number of games thing though. I’d be surprised if I can manage more than one honestly but with a fair wind, maybe I could do one a week. Do I need to commit at the start of the round? Or can I just say I’ll do one and then see about adding games of it seems possible for me?

1 Like

You do have to choose at the beginning of the round so that pairings can be made accordingly. That said, you can always sign up for more or fewer in subsequent rounds.

Ok, thanks. Better sign up just just one game at first then!

[As an aside, would it be better to change the title of this thread to include the words “live game, reverse Komi” or some such? I nearly didn’t click on it as I couldn’t figure out what the thread was about from the title. Maybe just me though]

1 Like

I’ll play. I much prefer reverse komi.

1 Like

Cool idea. I wish I had time to play but sadly I do not. I will keep an eye on the thread though as I may be able to join future rounds. Is that possible? To join future rounds if you have not played in previous rounds? And is there some kind of leaderboard or divisions from round to round?

Also, as a physicist I appreciate the Lagrange Points metaphor (that’s actually why I DID click) but I’m afraid that I have to agree with @teapoweredrobot that it does not enable people to know what it’s about. Might be better off named just Reverse Komi Tournament or similar.


Absolutely! Rounds are completely independent of eachother and are signed up for individually as well.

There is not. I’m not opposed to the idea in principle, but anything of that sort would have to be something which is secondary to the primary purpose of Lagrange Points to have reverse komi games with roughly equal chances for both sides, to have rounds which are signed up for individually without commitment to multiple, and to let players choose how many games they play each round. It’s the last point which really made me decide to not have rankings of that sort.

I do think there’s room for something of the sort, but I don’t know the specifics. Some sort of achievement system rewarding players who review a large portion of their own and others’ games in a given round might be a good way to go, and if anyone else wants to throw some ideas on the brainstorm pile feel free. Or, if anyone has a really good idea, feel free to make a post in this thread introducing it and giving the necessary records-keeping tables, ask a mod to wiki it (or wiki it yourself if you’re a regular) and I’ll link to it from the Index if there aren’t glaring flaws with it and people can either run with it or not.

Alright; title v3 coming up. :smiley: It’s entirely possible I’m holding onto the name “Lagrange Points” far beyond the point of reasonability due to my preference for slightly poetic and meaningful titles.


What kind of physics do you do? I used to work on DEAP3600 for a bit before going military.

Nice! I bet that was fun :grin: I’m working on magnetic confinement fusion.

1 Like

Oh, nice. I have a friend who went to LLNL for his work on plasma.

1 Like

This sounds like a fun idea, I’m in for one game and PMed you.

How about making an OGS group and inviting everyone who signs up so we can have some sense of the scale of it and find games in progress to watch?


Good idea. It’s still pretty bare-bones, but the fundamentals are there.

How will you determine the komi?


Short answer is “make run” slightly longer answer is as follows:

I use the new OGS rating to rank conversion formula (ln(rating/525)*23.15 iirc) to get the participants’ strengths in terms of stones, and then multiply by 14 and round off to the nearest integer (I suspect there is currently a bug causing it to round not to the nearest integer, but to either even or odd, but I’m looking into that today; should be a simple fix) to get the participants’ strengths in terms of points.

Then, when two players are matched, I add 7 (for komi) to the result of the above paragraph for Black, and subtract the result of the above paragraph for White.

This is equivalent to doing 7 - strength gap when White is stronger, and 7 + strength gap when Black is stronger.


So for, say, 8.0k black vs 7.0k white, you would get a “strength gap” of 14 and give -7 komi? I think OGS would call that a “handicap 1” game with no extra stones and 0.5 komi.


That is correct.

A “1 stone handicap” actually only accounts for a 0.5 stone difference in strength: stone handicap games being played with -7 komi would be more consistent.

Imagine a no-komi game between equal players: Alice and Bob. Alice takes Black and wins the first game by 7 points, as expected. They switch sides with Bob taking Black and winning by 7 points. The swing between these two results is not komi, as would be implied by taking a full stone difference in strength to be worth a “1 stone handicap”, but twice komi.

Katago also gives roughly twice komi as the value for an extra handicap stone when evaluating in Japanese rules:


I think you are both right.

1 handicap stone without komi, falls short by komi or half a stone, in a game between players separated by 1 full rank.

But OGS chose to use traditional handicap without correcting for the half-stone advantage for white in its rating calculations.


Got it. And OGS apparently rounds the rank difference down, so e.g. a 3.1k vs 5k player would still get just the 1 handicap. This tournament’s system seems like it could be a lot more accurate.

1 Like