lishogi.org/3euOsU6Q
Apparently this is my best performance so far: no inaccuracies, no mistakes, one blunder only and the lowest ACPL I've ever got.
Well, this is good but I cannot help but wondering how reliable this analysis is.
Also, how are the win/loss points calculated? I'm just curious about that, not complaining about the number of points I just lost. In fact I expected more.
Thanks for reading me and your answers. 🙂
don't think that the machine is perfect with its analysis; but generally, it is very good
on a different note: I'd recommend delaying when you choose to resign so that you can get more out of the game
after all, in this case, you resigned a winning position. but even if you are losing, please do not be too quick to give up
I wish you well with your next game
Yeah that's a pretty good game, nice job! You definitely should have played on though, you had a really strong advantage. That lance the opponent dropped at the end is just going to get trapped.
As far as the computer analysis goes, I think it's really good, obviously stronger than a human player would be. The only downside is that sometimes its middlegame plans are hard to understand, as it has such a deep look-ahead. When I use the engine after a game, I mostly use it for blunder checks and to see if I missed any tactics. If it gives me an inaccuracy because I moved a silver wrong, and the follow-up move from the opponent is just some pawn push on the other side of the board, I don't worry about it all that much. But if I missed a fork or a trapped piece, then I take note.
For the win/loss points, lishogi uses the Glicko 2 algorithm:
en.wikipedia.org/wiki/Glicko_rating_system#Glicko-2_algorithm
TLDR if you don't want to look at the math:
Beat an opponent a little stronger than you, you get a little points.
Beat an opponent a lot stronger, you get a lot of points.
Losing works the same way, but you lose points instead of win, of course.
You start off with a very "unstable" rating (your points will go up and down a LOT)
The more games you play, the more "stable" your rating becomes (you gain and lose less points)
If you suddenly start playing much better or worse, you stability goes back down (your points will go up and down a LOT again)
When you start playing consistently again, your rating becomes stable again.
❝ @Papermask after all, in this case, you resigned a winning position.❞
❝ @LizardWizard Yeah that's a pretty good game, nice job! You definitely should have played on though, you had a really strong advantage. That lance the opponent dropped at the end is just going to get trapped.❞
Thank you both. I definitely lack confidence. I saw that I had an advantage but I didn't see how to use it. 🤔
Thanks for the explanation about the rating. 🙂
lishogi.org/mmLEZ7Sw
My brains have been in "cheese mode" all day long. This is the best I've been able to do.
I'm still wondering about the analysis engine though. On what criteria does it flag this move or that one "inaccuracy" or "mistake"? Blunders are generally obvious. After using it extensively lately, I think I'm going to stop.
Do you play with the analysis engine in mind? Wondering "what would it say about that?"? 🤔
@ladyontheshogiboard
The inaccuracy, mistake, and blunder tags relate to how much the move you made deviates from the engine's best choice.
So if you are in a position where you have a +3.0 advantage move but you make a different move that puts you in, say, a -2.0 position instead, that would probably be tagged as a "mistake" or even "blunder".
And no, when I'm in game, I don't think about the analysis or anything like that. The engine is so much stronger than I am - and indeed stronger than pretty much any human player - that there's no point in me trying to "play like the engine". I wouldn't even know where to start.
Like I said above, I use the analysis post-game and look at mistakes and blunders. I usually don't worry too much about inaccuracies, honestly. Inaccuracies (in my experience) is when the engine doesn't hate your move, it just thinks you're "missing the point of the position", so to say.
I'd recommend you look through the engine's recommendations, and use the "detect threat" option in the corner, and play out some of those scenarios. Then ask yourself "is this something I could have reasonably saw during the game?" If it's something totally beyond your understanding (a mate in 17 or something crazy!), I wouldn't worry too much. But if after looking at the variation you say "Oh, of course, why didn't I see that?", or like "I totally missed that but now I see it", that's really where the engine analyses' use shines.
Just a follow up, you can also get a lot of info just from the quantity and placement of inaccuracies and mistakes/blunders:
Inaccuracies in the early game? I would say learn a bit of opening strategy or theory/joseki.
Blunders in the early game? You might have fallen for an opening trap, and it would be worth it to memorize
Inaccuracies in the middle game? Maybe your game plan or attack plan was not ideal.
Blunders in the middle/late game? I would recommend playing some tactical puzzles.
Missed checkmates in the end game? For sure tactical puzzles, as well as tsumeshogi (checkmate) puzzles. Even pro players grind these by the thousands.
You can't post in the forums yet. Play some games!