Shredder for Android ELO ratings: Accurate?

I'm a brand spanking new premium member of Chess.com. Got it for myself as a birthday present. :) I'm a rank beginner hoping to improve my game. I've found the Mentor feature and the video lessons to be marvelously enlightening.

I have a Nexus 4 that is pretty much my main chess playing computer. I've found that my favorite apps are Mobialia and Shredder. I'm inclining toward Shredder, since it actively ELO rates you and updates your rating.

This brought me to the question of how accurate it is. I can easily beat Shredder set at 1000 ELO. After several games my ELO rating is somewhere around 1300.

However, on Mobialia, I am quite hard-pressed to beat the phone at ELO 1000. I'd say my win/loss/draw rate there is probably about 20/30/50 percent, respectively.

Is there anyway to know if it is accurate?

Thanks!

Well, that's good news if so, but how do you figure that? Just curious, thanks.

Im on 1954 right now and my opponent is 1952.

that,s my favorite chess app.

I don't think chess engines provide an accurate rating at all. I beat easily Shredder and Rodent iv all the way to 2300. From 2300 to 2500 I use to get draws and some lost game from time to time. And I'm a crappy chess player.

In Chessmaster 11 I got 2100 rating in the graph and still growing, beating all the way to 2500 personalities and getting draws from 2500 up to 2700.

So I don't think it's accurate at all, i never went to official tournaments and never set a foot into a chess club (I read some chess books from time to time, though). But i don't think those ratings reflect my real strength. 1300 players crush me here on chess.com.

how can you beat 2300 in shredder?!

maybe your cpu is weak. i can win 1950 rated shredder hardly

my daily chess rating is 1533 and blitz chess rating is 1170

I beat shredder in my tablet (Samsung galaxy Tab A). "Shredder chess" app

Maybe the cpu in my tablet is too weak compared with a computer one, maybe that's the reason why 

Its quite unbelievable that you are managing to beat and draw strong engines, while you struggle against even beginner human players on this site.

Yes, that's the reason I think chess engines ratings are very wrong. Or my opponents here are too strong (despite of low rating), in rapid games they are almost always getting 3 or 4 inaccuracies, no mistakes, no blunders, and high accuracy, more than 90% most of the time. And they are supposed to be beginners. 

And my opponents here are always 1 or 2 months old accounts (in rapid games, not in daily). Maybe chess.com put me in a strange pool of players whenever the computer searches for opponents. 

That's the reason now I play mostly daily tournaments, to avoid that strange players pool in rapid. 

On the other hand when I play engines I concentrate much more, and I play longer games. That could be affecting results. But the difference is too big, so I automatically think engines rating must be wrong. I think they are measured by CCRL list, and those ratings are not similar to human ratings. 

My rapid games experience (I don't use to play blitz) here is strange. When I play new accounts I use to lose, even to 1200 players. When I play old enough accounts I win. For example I use to lose vs 1200, but for example my last rapid game I crush easily 1395 player because his account was 3 years old.

That's the reason i use to play engines more seriously, because the strength is more consistent. Even though ratings are not human comparable, at least I know I'll win when I play against 2000 rating, for example. And I know I'll find resistance vs 2400. Here in chess.com 1100 or 1200 will crush me and I beat easily 1400.

Here is my progress in Chessmaster 11 for the last months since coronavirus confinement (i didn't play chessmaster before confinement). My username in Chessmaster is Arahatree:

It's an exponential growth, and at the moment it's still growing. At the same time in chess.com i have a flat boring graph. It's strange that at the same time my chess is improving in chessmaster, my chess is deteriorating in chess.com, because my chess skills and knowledge is always the same. But, that's how it is.

The three weird-looking down peaks are the three times I drew vs chessmaster highest personality (1 won, 3 draws and 7 lost). I think that's a glitch in the game, everytime you draw vs Chessmaster you lose 500 points, and at the next game vs other personality you recover them automatically.

Engine's rating not very accurate. Human is able to beat 2000-2400 elo engine, or 2500.

I also can beat a 2556 elo chess engine.

Te pongo a jugar con mi chessmaster te viola

I just read this website, where the engine Hiarcs 13.1 (rated 2873 in CCRL list) won Mercosur Cup 2009 in Argentina, beating all GM, IM and FM, only one draw:

https://www.hiarcs.com/Games/Mercosur2009/mercosur09.htm

https://en.wikipedia.org/wiki/Pocket_Fritz

Hiarcs ran on a HTC Touch HD (ARM/528Mhz) mobile device, which is thousands times worse than any of our computers/smart phones.

So it looks like the engine's rating is pretty accurate.

Does this mean that beating Hiarcs 13.1 is the equivalent of beating a GM, right?

So, Chessmaster 11 Archangel is 2702 in CCRL list, only 171 points less than Hiarcs 13.1, if you beat Chessmaster 11 Archangel is like you are beating GMs and IMs?

@p8q that’s strange that you have such a success against shredder and chessmaster, I looked at your profile and you don’t have the same success again chess.com engines. Maybe you play better against a certain type of play, but shredder style is far from the one of chessmaster, so I have no clue about this…

Is this not 'advertising' the usb of both these apps ? Can it be called 'spam' ?

Maybe you didn't search in appropriate way... because you didn't take under consideration time controls: it's true lately i didn't get good results vs strong engines here in chess.com, because lately here i've been playing only blitz vs c.c bots, so my games are mostly lost, this is the result:

Still, according to my c.c rating i shouldn't get those few won and draws. In the lost games i got to equal position in the endgame, which shouldn't be possible according to my rating either.

For the last two months (more or less) in chess.com bots i've been playing only blitz, that's why i played a lot vs low rated bots. Lately at long time controls i only played vs Rodent IV and other engines on computer offline.

When i was playing longer time control games vs chess.com bots this is the result (same filter as before):

So i thought chess.com bots ratings were very low, because i was getting good results. But i compared with other engines and they are all well calibrated (i didn't save those games in my c.c archive because i don't want to have games that i didn't play (that are bot vs bot) in my archive).

I don't know, this has always being very confusing, my real life friends here don't know why either.

I chose not to think more about it, cause i feel like a waste of time keep thinking about this. Now i only play vs engines and humans OTB (a few times humans online, though) and i got more interested in blindfold chess -> first without pieces, now without board, very cool.

Problem is that i can only play blindfold when i'm with closed eyes and too much focused. My goal now is to do it the same with opened eyes while pretending i'm listening to someone, hehehe. I still need too much focus blindfolded, i want to keep developing on this path until it's more natural and feels more light.

I don't know why my chess offline vs engines and vs humans OTB improves dramatically while my chess online goes down or barely maintains.

(From my point of view)

Do not compare elo human and elo AI, two different things.

In many chess program the level 2000 has relevance because beating it (if we play same time, without take back, without anti-computer technique) means that we have a good enough technical level. 

To be good against AIs will not be enough that we will necessarily be good against humans because by dint of playing against a chess program we understand better its algorithms, how the AI works. It is the technical elements like the tactical level, knowing how to win finals etc. that work the same.

I don't have any idea how are engines algorithm, no idea about how engines work.

I never used anti-engine strategy to play vs engines. No help of any kind, no ods, etc. However, i can beat 2300 rated engines and humans 1100 rated in chess.com beat me constantly.

Ok, we agree that we will not compare engines rating with humans rating. However, an improvement playing vs engines is an improvement in chess. And an improvement in chess should also translate into an improvement playing vs humans online.

I'm not saying i'm a good chess player or anything, just saying that i improved at chess, OTB and vs engines offline, but online i got the reverse result. i don't know why. I suppose people online play different style or something like that, i have no idea why this happens.

I find that hard to believe...

We start at 900 elo, AI opponents taken at random and as for the site it gives about the same elo.  To pass a level (1400, 1500 etc. and pass a level means a 1400 pass a 1500) it is generally necessary to beat a player who is classified with +200 elo (example if you are 1900 to pass the 1900 "is necessary" to beat a 2100) and we can also fall against a -200 (example 1900 it will be 1700).

In time 20 minutes of memory I was stuck in the 1800 to 2000 range and I beat AI.Dobie (2100) by an anti-computer technique (exchange of Queen to reduce the tactical difficulties because he can be very strong in tactics, wait for him to play clown or weak moves and give him no chance by reducing the material). I know we can beat AI.Natalie (2300) if we play 1) e4 because she plays bad openning ( 1.e4-e5 2.Nf3-f5 for example 3.Nxe5-Nc6) and if we play her against AI.Kenji (1900), he wins with white in this opening. There is AI.Wally (2200) who values a Rook more than a Queen where there is a way to beat him, AI.Buck (2400) who can make counter-performances and maybe AI.Jade (2300) but after it seems hard to believe if we are not gosu.

Right now I'm doing an AI rating in time 5 minutes (i have of course removed the alarm clock).

I lost against AI.Thorian (1100) by the time elapsed (while being completely winning but to say that even while being winning there can be a technical difficulty to finish his opponent), almost made a stalemate against AI.Miaranda and lost to AI.Buddy (1200) for a tactical gaffe. I am currently at 1318 for 10 games (8 wins, 2 losses) but  even if it's the beginning, Chessmaster's AIs pose technical difficulties that make me not believe too much in this "I'm stronger than AI.2300 or + of the Chessmaster program but less strong than 1300 or - of a chess.com player" story.