Stockfish 15 Elo?

Sort:
ChessSBM

I heard that stockfish 15 released recently, butI’m wondering what’s its Elo. Anyone knows?

Mattew

First 

Wait no second

 

But nah idk

Matthew_Slater

Stockfish website says v15 is better than v14 by 36 rating points but I’m not quite sure what the exact rating is

ChessSBM
BillBelichick01 wrote:

Stockfish website says v15 is better than v14 by 36 rating points but I’m not quite sure what the exact rating is

I heard that stockfish 14 is around 3750 elo, but when I search for it, I can’t seem to find its elo.

bradleytrex

wait what sf15 is out !?!?

ChessSBM
Menhattan wrote:

I assume around 3700

 

I hoped for around 4000, so that’s why I brought that question. I wanted to see whether stockfish was able to reach 4000 which means a really strong engine.

ChessSBM
BradleyTRex wrote:

wait what sf15 is out !?!?

Yeah, I was really surprised about that too. One of the reasons I created this thread is to inform people that stockfish 15 is released.

ChessSBM
pfren wrote:

Does it matter?

ELO is a statistical number extracted from results against other engines.

Outside that pool the number means nothing- e.g. the top Centaur players at ICCF play better chess than any existing standalone engine, and the top rating there is below 2700.

It matters for me since it helps me compare with other engines. For example I can know the difference between Alphazero and Stockfish 15.

ThaiViet41
pfren wrote:

Does it matter?

ELO is a statistical number extracted from results against other engines.

Outside that pool the number means nothing- e.g. the top Centaur players at ICCF play better chess than any existing standalone engine, and the top rating there is below 2700.

I am surprise that this is not clear here when many accord a lot of importance to the elo rating. 

Elo tell you where you are situated relative to the other peoples present in the group where you are competing. 

So let say that Elo 1000 correspond to the middle of the group (of course that's an arbitrary number as the middle will not be the same depending of the composition of the group) . If you have three different group of different level the Elo 1000 of each group will not have the same strenght. 

Or in Chess

The Elo rating of the same player is higher OTB, then on chess.com and lower in turn to Lichess. 

The reason : The average of the pool of player having an OTB rating is stronger than the average of the pool of player here which in turn in stronger than the average of the pool of player on Lichess. 

Elo is also not good for measuring the outlier, as by definition it operate by comparison not absolute measurement. 

tnk116290

idk

 

ChessSBM

Thanks @YazanShatnawi , can I ask where you find the list though?

ChessSBM

Thanks

herbievanbeveren

Based upon the 36+ for v15 versus SF14, ELO being 3549, it should be 3585.

flashlight002

@hwrbievanbeveren is spot on!! Ratings were: Stockfish 13 was clocked at elo 3546, Stockfish 14 at elo 3549 and Stockfish 15 at a whopping elo of 3585!!! 

When playing vs Stockfish via a good interface like Lukas Chess I find it helpful to go into engine setup and weeken the elo of Stockfish to an elo just a bit higher than my own! This way you won't get crushed time and time again! happy.png. When performing analyses naturally one wants engine at full capacity! 

According to the CCRL website the above elos are a bit high. Elos fluctuate all the time if judged using the CCRL testing framework that runs games constantly at 40/15 time frame and set hardware configurations. There is also a 40/2 framework, but under this very fast game format elos are much higher. I prefer to use the 40/15 table. 

Here is the link to CCRL site: https://bit.ly/3Pu3NmH 

Hope I have been of some help happy.png happy playing!

IncomeManagement

I remember seeing in multiple places, the elo of Stockfish is around 3700 (Could be more or less) 

spig0t

According to CCRL at 40/15 (40 moves in 15 minutes), SF 15.1 is 3531, SF 14 3529, and SF 13 3526 (and SF 14.1 for some reason is down at 3517) ... so, very small increases in ELO if you give the engine half a minute to think for a move (when these top engines are playing each other with this much time, it's 70%-80% draws, so very hard to gain elo)....

Bigger differences at 40/2 (40 moves in 2 minutes) - SF 15.1 is 4019, SF 14.1 3883, SF 14 3853, and SF 13 3784.

Note that CCRL also says it plays with random openings at 40/2 (maybe at 40/15 too, unclear to me from what it says?) - without this, presumably there'd be even more draws between the top engines and thus lower elos with smaller differences.

Pladzee

stockfish 15 has a total elo of 3650. but in the latest version of stockfish which people misunderstand and think they are the same thing (stockfish 15.1), stockfish now has 3700 elo including 1+ evaluation, that 1+ evaluation might not seem like a lot but now with that 1+ eval, stockfish has a 50% chance of winning against a player/engine that has an equal amount of elo/rating

GM_Blunderfish

Perfect chess is always a draw

nicoladc_1989

I think that a good estime of Stockfish Elo is an Elo rating big enough that the best human player ever won't lose points if he lose against it, because he don't have any odds of win. The higher human rating ever is 2882 so - according to this assumption - a good starting Elo for Stockfish should be around 4000.