9

It is generally considered that white has better chances of winning in chess. This is due to white's initiative, since white gets to move first. The difference in chances seems to get more significant, as the level of play gets higher.

So, I suppose most grandmasters will choose to play white, whenever they are given the choice (especially in crucial games).

(1) What is the general view on the Armageddon tiebreak in that respect? Is there a clear preference in tournament play, or are white and black equally liked in Armageddon? What are opinions on this at the top level?

(2) Take the extreme example. In case of a tie after all classical, rapid and blitz games in a WCh match, there would be a drawing of lots, and the player who wins, gets to choose the colour in the Armageddon game. Which colour would be the choice of most GMs?

Remark. There are many variants of the Armageddon tiebreak. To avoid ambiguity, let us fix the time control used in the rules for the Anand-Carlsen WCh match 2014:

The player with the white pieces shall receive 5 minutes, the player with the black pieces shall receive 4 minutes whereupon, after the 60th move, both players shall receive an increment of 3 seconds starting from move 61. In case of a draw the player with the black pieces is declared the winner.

1
  • 1
    I'd love to see some statistics. Someone with "big data" and a good filter could do it very easily. Please, if you fell like sharing, and have the means, consider post it here. Thanks!
    – DrBeco
    Apr 6 '15 at 16:03
5

It is generally considered that white has better chances of winning in chess. This is due to white's initiative, since white gets to move first. The difference in chances seems to get more significant, as the level of play gets higher.

That is completely correct.

But in an armegeddon game the expected draw ratio is just as important as the white advantage and much more variable over playing strength and time controls. So the stronger the players and the longer the time control, the better it is to take black, because a draw becomes more likely.

To give a real life example: A few years back Gata Kamsky became US champ by playing the black pieces with 25 min on the clock against his opponent's 60 min. That's quite a different ratio compared to the Carlsen-Anand tiebreak time control … Kamsky's choice was justified by the rather long time control and the high level of competition. With a 2:1 time advantage I (a 2100 Elo player) would have taken white any day.

So there is a tradeoff between time control, playing strength and personal style in as much as they influence the expected draw ratio and the expected white advantage.

3
  • 1
    The longer the time control, the more advantage black has. In the field of computer chess, it's axiomatic that doubling the time adds 100 to the effective rating. For humans, the effect is slightly greater. And a +200 in Elo yields an expected score difference of 75-25 (+100 is 64-36). The first might be enough to tip it to White, the second definitely doesn't.
    – Arlen
    Dec 2 '14 at 3:55
  • @Arlen do you have references for what you are saying? Thanks Nov 12 '18 at 0:31
  • The win probability difference comes from the table in Arpad Elo's "Rating of Chessplayers, Past and Present" but I honestly forget where the doubling of thinking time equates to +100 Elo. I suspect it was from Levy's book on computer chess. I remember it from years ago -- a postal master was playing half a dozen computers at once. He referenced it when explaining why he was winning; the brain is able to use the extra time more efficiently than the computers were. It's possible advancement in algorithms has affected that projection since then.
    – Arlen
    Aug 7 '19 at 14:02
9

Armageddon Chess is a fair tie-break system only if the two players bid for how much time Black should have.

Foolishly most times I see Armageddon specified there is no bidding. Instead the Tournament Organizer gazes into his magic oracle which tells him the exactly correct fair amount of time to give to Black (the T.O. is careful to never let us see the oracle for ourselves). This no-bid foolishness amazes me.

2
  • FYI, I expanded on your answer in a standalone answer because my expansion was too long for a comment. Oct 13 '20 at 7:05
  • Should note that while armageddon chess with bids is fundamentally a fair game, that's including the bidding as part of the game, which being good at chess doesn't make you good at. What I mean is, while it's a fair game, it may not be a fair chess game
    – Cruncher
    Oct 21 '20 at 4:31
5

If you ask this question to any FIDE officials they will answer you that nobody has advantage. Because ultimately, tiebreaks have to be fair. If it's not fair, then no one will want to play it.

However, in practical black has an advantage due to the psychological advantage to just hold the draw. In fact, to compensate this, there is a new system of players bidding for time in order to get the sides.

For example, if A and B is to play off against each other. They will have to offer the time they are willing to take in order to get black, A offers 3 minutes for black against white's 5 minute; B offers 4 minutes to play as black. Hence A wins the bid and the game proceed with 5 minutes for white, and 3 minutes for black.

Anyway, I haven't came across any statistics for armageddon tie breaks. But I am pretty sure that it depends on your style of playing to say that which side is advantageous. For an attacking player, white is obviously better, while for a defending player, black is probably a good choice.

4

There is a recent academic paper claiming that time controls in Armageddon of the 2018 World Chess Championship "greatly favours black"

1
  • 1
    Interesting paper, but with a complete lack of comparison to any real data it's hard to say to what extent its conclusions are valid. In particular, the assumption that human strength can be compared to engine strength just by switching the engine time controls, and then that the ratio of time controls between engines can be directly transposed to humans would require more investigation IMO. One could also argue that using a single blitz tournament might lack statistical significance.
    – ATLPoly
    Jan 16 '19 at 21:08
2

I expand on GeneM's answer correctly noting that "Armageddon Chess is a fair tie-break system only if the two players bid for how much time Black should have.". (My additions are too long for a comment to that answer.)

An armageddon tiebreak with an arbitrary, externally imposed time penalty for the Black/draw-odds player is not reliably fair because it may grant an advantage to one of the players because there is no guarantee that the arbitrarily chosen time penalty is the correct amount to avoid undercompensating or overcompensating White for accepting must-win odds. (If the time penalty for Black is too high, it overcompensates White; if the time penalty for Black is too short, it undercompensates White.)

By having the players bid for the right to choose one's color, the arbitrariness is removed and the time penalty the Black player pays is one that that player voluntarily offered to pay. Thus armageddon with bidding is fair.

Bidding has been used in the past (e.g, in at least each of the 2013, 2014, 2015 and 2016 US Championships).

For example, in the regulations for the 2016 US Championships:

An Armageddon Game is defined as a game with base time of 45 minutes for each Player. Black will have draw odds. Each Player shall bid an amount of time (minutes and seconds, a number equal to or less than 45:00) with which they are willing to play in order to choose their color. The Player who bids the lowest amount of time chooses his color and begins with that amount of time; the other Player receives 45:00. If both Players bid exactly the same amount of time, the Chief Arbiter will flip a coin to determine who shall choose their color.

As an example of how the bidding can unfold, in the 2013 US Championship the base time control was 45:00. Both players bid aggressively in hopes of getting the right to draw odds: GM Alejandro Ramirez (19:45) and GM Gata Kamsky (20:00). Ramirez' lower bid won, and he received Black with draw odds. Kamsky got the White pieces with the full 45:00 base time control but would have to win to survive. (Kamsky won the game and, with it, the US Championship for that year.) (Note that in this 2013 US Championship apparently the rule was that the winner of the bidding automatically gets Black. In later years of the US Championship, the rule was, properly, that the winner of the bidding could choose his color.) You can watch commentators GM Yasser Seirawan, GM Maurice Ashley, and WGM Jen Shahade discuss the bidding procedure and their thoughts about what each player should bid beginning around 30:32 of this YouTube video of coverage of the playoff round. The sealed bids are opened around 48:20.

It's not hard, nor does it take much time, to administer or implement such a bidding procedure, so I'm baffled why it's not used more in important tournaments.

Some additional notes:

  • I say that an arbitrary time penalty is not reliably fair because, of course, it's possible that—by accident, in the same way that a broken clock is correct twice a day—the arbitrarily imposed time penalty may turn out to be the same penalty that bidding would result in.
  • Some implementations (and even some of my examples and explanations above) assume that the player winning the bid automatically gets the Black pieces with draw odds. But that assumes that it's always the case that the benefit of draw odds for Black outweighs the disadvantage of moving second. There's no need to assume that. Instead, the winner of the bidding should be allowed to choose between (a) White with must-win odds and less time than Black or (b) Black with draw odds and less time than White.
  • It would be superior if the rule were, instead of (a) the winning (i.e., lower) bidder gets his own bid on his clock (i.e., in this case Alejandro getting 19:45 on his clock), that (b) the winning bidder receives the other player's bid on the winning bidder's clock (i.e., Alejandro would get 20:00 minutes on his clock vs. 45:00 on Gata's). This would be a form of a second-price auction, which has the desirable property that each player has the incentive to bid his true value. (Otherwise, as in the 2013 US Championship, each player A has an incentive to try to outguess his opponent player B and just barely undercut B's bid, even if in fact player A would have been willing to accept even less time in exchange for Black/draw-odds.)
3
  • 1
    Thanks, your answer gave me a lot to think about. I want to ask if you could explain the second-price auction portion a bit more? I mean, what's stopping Alejandro from bidding a super-low 1:00 minute or even 0:30 to win the bid and still get Gata's 20:00 on the clock?
    – rhyaeris
    Apr 28 at 3:37
  • 1
    @rhyaeris, nothing is stopping Alejandro from bidding lower. However, note that Alejandro didn't know in advance what Gata's bid would be. if Alejandro did bid 30 seconds, as you suggested, he'd be in bad shape if Gata then bid 31 seconds, because Alejandro would then have only 31 seconds to Gata's 45:00. The system is designed to elicit "truthful revelation" in the bids. It's an optimal strategy to bid the lowest time you'd be willing to accept in order to get draw odds. Any lower and you wouldn't want that time with draw odds; you'd rather have 45 minutes with must-win odds. Apr 30 at 0:54
  • Game theory is not my forte but you explained it well. Thanks
    – rhyaeris
    May 1 at 2:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.