Honestly, two rooks for a queen being a disadvantage is kind of a relief.<p>I’ve always been taught the two rooks are better and who can argue with 5 + 5 > 9? But also, I’ve also lost almost every game where I’ve had the rooks. I always thought that I just needed to be a better player to take advantage of it. Glad to know it wasn’t just me.<p>Just goes to show that these shortcuts, like the point system, are only heuristics, and pretty shallow ones at that. Knowing that being up a bishop gives you slightly more of an advantage than a knight is better than nothing. But learning in which sorts of positions a knight is actually better than a bishop will give you a much deeper understanding (and correspondingly more wins).
This thread makes me want to build a simple chess-practice tool that works like this:
1) Get a big database of positions<p>2) Use StockFish to evaluate all the positions<p>3) Pick a position at random, show it to the user, and ask for their evaluation<p>4) User gets a score based on the difference b/t their eval and StockFish's.<p>The idea being that this could allow you to rapidly hone your position-evaluation skill. That might be a faster way to improve at chess than just grinding through many games.
Am I misunderstanding something or is there a glaring statistical error here? I think this analysis works if the matches are all between two players of <i>truly</i> identical skill that will roughly win 50% of the time against the other, but that's not really the case here. They attempt to do that by controlling for Elo but I'm decently sure that you could pick any two players of the same Elo on lichess and one will win the majority of games.<p>Now if we assume there is often a slightly better player in these games, the better player will more likely get into an advantageous position early in the game <i>and</i> win more often, but not entirely because they got into the more advantageous position. What I'm trying to say is that someone who is up a rook will win partially because they're up a rook and partially because they are likely a better player in the first place and will continue to play better than their opponent.<p>I think to do this study correctly you'd need to place new players in a random position drawn from the dataset and actually evaluate the win rate without the confounding factor of having gotten into that position in the first place.
I'm surprised that in the higher ELO ranges, being up a clean pawn has only a comparatively minor impact - I thought that one clean pawn ahead and no other major positional or structural disadvantages was more or less "GG" among top players.
Neat results! I am very much a rank amateur, but it looks like these results roughly align with the traditional 1-3-5-9 point values for the pieces, with a couple exceptions. Advantages of a single pawn, minor piece, or rook have proportionate values similar to what you'd expect. In particular, bishops and knights are very close with bishops being worth a tiny bit more. Winning the exchange is a sizeable advantage. Trading a rook for two minor pieces is an advantage except at the lowest skill levels.<p>Two rooks turn out to be significantly worse than a queen instead of slightly better. But the most surprising thing to me is that having the two bishops seems to be worth almost nothing -- close to 50% odds across skill level. A single pawn advantage is more valuable! (The article says "4-5% more likely to win" at Elo 1200-1400, but that doesn't match the graph.) These surprising results were also more consistent across skill level, while the well-known advantages are worth significantly more to skilled players.
Really nice insights.<p>Though<p>"Even at lower Elo ranges like 1200-1400, you're 4-5% more likely to win if you have the only bishop pair. 1200 Elo players don't know how to take advantage of having the bishop pair, and yet it helps them win all the same [...] A 1000-1200 Elo player is only about 8% more likely to win when up two minors for a rook."
Nice article!!<p>But this is not what you're testing. You're not determining the effect of imbalance alone, but imbalance + particular position of pieces. To determine the effect of imbalance you should study how does not having a certain piece since the beginning affect the result of the game.<p>Also, being a pawn up in the opening is much more different than being a pawn up in the end game. It would be interesting to redo this analysis but splitting by opening, middle game, and end game.
Maybe I'm just a choke artist but if I win the exchange I'm probably losing the game. Opponent always seems to "turn on" and I just get blown away
I wonder how this would change if you only looked at highly rated chess engines. For example, is a rook pair really not as good as a queen or is it just that _humans_ aren't as good at using the rooks effectively?
Would be interesting to see material imbalances as a function of plies into the game. As in, we know an extra pawn is more important in the endgame than in the opening, but how much more?
"1200 Elo players don't know how to take advantage of having the bishop pair."<p>How exactly does one take advantage of having a bishop pair? Why is that not obvious to players under 1200?