Wide discrepancy between online and blitz rating. Why?

I'm a newcomer to online chess. I've only played less than 10 games, so my rating doesn't mean anything. I'm only asking about others, more established players who've played hundreds of online chess games.

I've noticed that a lot of players rated at 1800-1900 online chess actually have pretty low blitz rating. As in 1400-1500 blitz, some even as low as 1200s. Can anyone familiar with online chess explain for me why there can be such wide differences?

I'm curious because having played blitz a lot, I know the difference between 1200 and 1800 in blitz is like night and day. There's a big difference in all departments: from recognizing pattern, developing long-term plan to making less mistakes and exploiting mistakes. It's simply another level of chess understanding. So I would think that if you're rated 1800 in any format, you would still have that same understanding in you, and I don't understand how such person can simultaneously have essentially a really poor understanding as to have 1200-1300 blitz rating. Is it because at shorter time control, he/she can no longer think and calculate and retain all the chess insights? (Remember that if s/he is short on time, his/her opponent is also short on time). Or is it because such person will always have to take a definitive amount of time for each move, so in a blitz game they would always lose on time even though they can make superior moves?

I know an easy answer is people cheat and use engines etc but I don't want to think it's that prevalent. Perhaps there's some another reason?

When I look at the (well-known) players in the real world, usually there's no such discrepancy. Most of time the blitz and OTB rating are close to each other. If anything, for the fast players the blitz rating is even higher.

Online chess here allows the use of databases, opening books, and you can move pieces around on a analysis board with one or many days per move. Obviously you can't do that in blitz and you have only a few minutes for the entire game.

Another factor causing the difference is the fact that lots of strong players rather playing blitz.  This means that the weaker player end up with less points .

It's because the pool of weaker players tend to not play blitz because they simply get slaughtered, so the blitz ratings are essentially absent the pool of weakest players. I'd wager that few folks rated <1000 in standard chess even bother with blitz because they just get massacred due to gross blunders.

Here's the most recent ratings survey by chess.com - there's a big difference between blitz and standard ratings up to 1500, and then it starts to equalize and even reverse as you get over 2000.

The take home message is that you shouldn't make the COMMON error here of thinking that all your ratings here (blitz/bullet/standard) should be equal - they shouldn't be, and it's no indicative of your weakness in blitz if your blitz lags your standard even by 200 points.

 There's also the fact that there are way more blitz games played than correspondence games (several orders of magnitude more), and glicko has shown a pattern of deflating over large samples (average rating going down over time).  Glicko calculations do not conserve rating points.  In Elo calculations the winner gains the same amount of points as the loser loses.  This is not true in glicko calculations because the players' RDs are often different.

You're talking about glicko. Is glicko the system used in chess.com for rating?

Overall what I take from the answers is that blitz rating is generally lower compared to online chess for the same person. Perhaps 200-300 point would be the norm (as is the case of Alvin_Cruz)?

What I've just realized is that there is at least one example where someone has blitz rating in 1900-2000 range, and online rating of 1200-1300 range. I only know one such person so far, and I don't want to list his ID since I don't want it to look like I'm criticizing (I'm not). But honestly I cannot explain how that can happen. Let's call him A, and think about the case when he plays with the typical player (B) who is rated 1800 online and 1600 in blitz. A will be able to beat B in blitz 7/8 of the time (meaning his chess level is clearly superior), and yet when both are given a lot of time to think, A will mostly lose to B! Odd, isn't it?

Yes, chess.com uses Glicko.

Some people play their "online" games using only seconds to make a move... some are basically required to do this because they're playing 100s of games at once.

Some people have low "online" ratings because they've timed out a large number of games.

If the 1900 blitz player didn't cheat to get 1900, he should be rated well over 1300 "online" even if playing all his online games as if they were blitz.

Exactly. That's what I felt too. I don't want to just accuse randome people of cheating though, hence the puzzle.

Another possibility I've thought of is maybe there are 2 people sharing the same account, one playing blitz (the "adult"), and one playing online (the "kid"). I was thinking that because the avatar shows a very young kid playing chess.

Ii have been accused of cheating because I have a low blitz rating. a 1600 player doesn't think I SHOULD BE CAPable of beating him. I get disconnected a lot and have serious lag on days. aLso many lowly rated players are just pushing wood so you get a lot strange positions with no time to think plus kazikame pilots who believe the greatest innovation in chess theory is a sac on f3/f7.

1700-1800 in correspondence and 1599 blitz is pretty close to the same percentile rank. 2100+ correspondence and 1599 blitz is suspicious, however.

Could you elaborate how you come to that conclusion? In particular how do you find the percentile rank of a given rating?

I tried to give it a go using data here, but it doesn't seem the difference is that dramatic. Blitz ratings do seem lower than turn-based, but only by about 100 points or so. Here's how I do it (all data from http://www.chess.com/echess/players)

First I go to Live Blitz (total 1798445 players), then hovering at 1750-1850 range, it says there are 19662 players. I then move to the next higher ranges, all the way to the end and add all the numbers up, and I got about 46K. These are the players rated 1750 and higher. So the percentile rank of someone rated 1750 blitz is about 97.4%

Then I go to Turned-Based Standard (total 485908). At 97.4% percentile, the number of players above should be 12443 (roughly). I then go from the top range down, adding the number of players in each range, and I reach 12443 at the low end of rating range 1850-1950. So I'd say 97.4 percentile would be around 1860-1870. It is indeed higher than blitz, just not the 300-400 point difference that seems to be the consensus of everyone's opinion. Unless the data is biased in some way.

PS. Wow as a side effect I just realized I'm actually at 97% percentile for blitz. I know it means nothing and one shouldn't take it seriously etc, but I've always only thought of myself as being average. 3% of the chess-playing public (who are way better than me) is many many thousands of people, but still, that does massage the ego just a tiny wee little bit :)

The stats are readily available to premium members.

Some blitz stats:

Similar correspondence stats:

The blitz player's correspondence stats:

my brain explodes if i play bliz chess.

Coresspondence players or those who don't play blitz, have no idea how different it is and how much skills a player need to have to reach same rating as correspondence. I would say CC is far more away than blitz compared to OTB play.

I have played one 7 game Blitz tourney and got a rating of 1350 - over 700 lower than my online rating. The reason for that, even though I had nearly every game won, was that I got in so much time trouble because, together with the thinking, I can't move the mouse that quickly in the shorter games. I can't move it nowhere near as fast as my hand in OTB Blitz games. I knew that would happen before I played the tourney.

One is in fahrenheit and the other is in centigrade.

Huge differences in ratings between live play and correspondence play? Could mean an excessive use of reference materials and/or engine assistance.

But not always. Some players may lack positional/tactical intuition in live play, but are very good at analyzing and evaluating lines when you give them several days to think and try out moves.

Either way, I tend to prefer live chess, just because I'd rather face my opponent's brain with no assistance, fair and square.

That said, I definitely see the appeal of correspondence chess. Being able to ponder a position for days is a nice luxury that can bring about some great positional insights.

Just beware of the engine users. :D

!!

Its online chess.  I dont take it seriously all the time, and im sure others feel the same.  I wouldnt get caught up in why someones ratings differ so much.