2200 vs 2700

I've been thinking. At the lower levels 500 points is a huge difference. An 1800 can win virtually every game against a 1300 without breaking a sweat.

But is that also true at the higher levels? Can your average 2700 really sweep aside a 2200 like it's nothing? To be 2200 I guess you'd have to have a really profound knowledge of the game, perhaps that gives you decent drawing chances if you play safe?

If any 2200 (or close) is reading this, how would you fancy your chances? If you're that good, do you still feel like there's a whole higher echelon of players you have virtually no chance against? 

Edit: I mean FIDE ratings, not online ratings.

Thats the way rating works. 500 points is the same all over the scale.

1100 vs 600 is the same difference as 2700 vs 2200

Actually, it might evem be more accurate at higher level, because the 2700 is less likely to make a game losing blunder.

You hugely over estimate a 2200 player.  "Profound knoweldge of the game" I'm sure they'd be the first to laugh out loud at that comment.  Sure they can destroy a 1500 player, but against strong competition they're clueless and guessing most of the time, having no idea which moves are good... at least that's how it's been described to me.

Kasprov famously played some super simuls.  He played grandmasters 5 to 8 at the same time (the Czech national time, the German national team, and maybe others too).  Sometimes he lost 1 game, sometimes he lost none...  Ratings don't suddenly mean nothing after some point.

2200 isn't even worth considering.  Kauffman calculated Kasparov would nearly have to give knight odds to make it an even game between him and a 2200 player.  And because ratings move more slowly at the GM level, if anything there's a much better chance of an 1800 dropping a game to a 1300 than a 2700 to a 2200.  Also as rooperi notes there are less game losing blunders.

http://www.chessgames.com/perl/chesscollection?cid=1014241

http://home.comcast.net/~danheisman/Articles/evaluation_of_material_imbalance.htm  (near the bottom, Kasparov would be favored at knight odds vs a 2100 player).

I knew a 2200 player who mentioned playing regularly with a 2500+ player that could squash him like a bug most of the time.  

Maybe we use different definitions of "profound knowledge of the game". As a 2200, you certainly do have that. Or else, you would not be 2200. You need to be a well rounded player overall to get to that rating. Of course, they do not stand a chance against a 2700, but that doesn't mean that they do not know at all, what's going on in a chess game.

It's not unusual for a 2700-player to drop a draw against someone rated 2200-2400. If I remember well there were several such examples in the most recent word cup.

While it is true that a 500 rating usually means a huge gap in knowledge, understanding and calculation ability, in chess it's not always easy to get a position in which these factors can be put to use, especially with black. I'm reminded of the Game Meulders-Kasparov at the Swift tournament 1987, and Kasparov's analysis. Although the world no.1 had taken his FM opponent very seriously, including some deep opening preparation (a pawn sac in the Nimzo), he had to work extremely hard in the endgame and only won because Meulders missed a few opportunities to save himself.

The other danger in chess: Everyone miscalculates occasionally. Stronger players less often than weaker players, but they still do it from time to time. And after a miscalculation, anything can happen. If the GM is lucky, there is no consequence and he stays in control. If not, he might have a hard time, or even a hopeless position, that no elo points can fix.

Well ok, each player has different standards.  If you asked 100 different players to say at what point a player has "profound knowledge" or is "well rounded" they'd probably mention players 300-400-500 points higher than themselves.

To a true beginner, a well rounded player is probably 1200-1300... and they're not wrong of course.  At 1300 (I'm thinking USCF) you know some basic openings and endgames and all the basic tactical themes and many strategic themes.  Opening, middlegame, endgames, tactics, strategy, calculation, they have it all!  How well rounded they must be!

But ask the 1300 and they'll tell you they're clueless when they play.  Sure they beat the beginner, but the beginner gave away all his pieces.  They'll tell you how 1700 rated players are the well rounded ones.

Similarly the 2700 player would say "sure I beat the 2200 player, but that's only because they played terrible moves"  And they'll go on to tell you they're better at this than that... endgames or calculation or tactics are my weakness etc.

So ok, yes, to you 2200 is well rounded.  But from the point of view of 2200s and 2700s they are not well rounded (much less have profound knowledge).

Yes. Absolutely. It is massive, it is huge. But above some wrote that the lower rated players "do not stand a chance". And considering the occasional upsets that do happen, I think that is not quite true.

Wow, Kasparov took an FM seriously?  Was it some 12 year old who had gained 600 points in the last 2 years or something?  I can't quite imagine him taking an FM seriously.

Nope, Richard Meulders was a 36 year-old FM rated 2360. He got into that tournament because he was Belgium's number 2 (after Luc Winants), and the organizer Bessel Kok wanted to throw two countrymen into the super tournament shark pond.

2700 will win about 6,5-0,5 in a 12 game match. A 2300-player from my chess club once drew with GM Malakhov with 2700.

I've studied upsets (set Chessbase to max 2000 minimum 1400) where the 1400 defeated a 1900 and have a couple such games saved and studied.  The 1900 as expected outplayed the 1400 through most of the game but the 1900 made a dumb endgame mistake that the 1400 capitalized on. 

Generally though you're correct, 500 points lower generally isn't much effort to defeat.  I have defeated people USCF 1600 and even beat (but more often lost to) 2000s so it all comes down to knowledge gaps. 

Someone who is 2200 knows everything there is about weak squares, pawn structures, color complexes, tactical motifs, plans associated with different pawn structure setups (yours and the opponent's such as Marcozy Bind vs. e6 + d6, Carlsbad, Botvinnik, c5 + d5 vs. c3 + d4 +e3 or vice versa, etc.), knows when to adventageously trade down, swapping and squeezing, or when to accept that all winning prospects lie with the opponent and fight for the draw, and on top of all those basics knows a good deal of opening theory.  They also calculate the same as experts but due to their pattern recognition abilities don't need to calculate as much.

The Gap between 2700 and 2200 is due to having more patterns stored to memory: a 2200 may have digested and memorized FCE, Chess Strategy, BCE, and Dvoretsky's Endgame Manual, but the 2700 would know everything from those text but memorize the entire ECE for good measure too.  Heisman says that NMs have to think on their own for 15 moves per game whereas a GM only 5 times per game.  This isn't just memorizing some stupid variations (though they're included in the 200,000 positions that GMs memorize) but recalling patterns and viewing their relation to the chessboard as a whole. 

Personally I think at the very top differences of less than say 100 points is not necessarilly proof of who the better player is.   I would rather say a match between the players is better evidence.  

Once you get 200 elo points difference though then I think even if someone won a match the higher rated player just had a bad event.  (except in some cases of young people who are improving faster than their rating)

The rating system is just one of many possible reasonable rating systems.  For example if draws did not count at all at the highest levels I think, in theory, we would have a better predicter of who would win a chess match between the two.    The problem is as a practical matter I think averaging in all the draws gives us more data and therefore in practice to some extent gives us a better prediction.  The current system counting draws gives a better prediction of who will win a tournament.  

But I do think the k factor for draws should be reduced.  But whatever. 

I looked in my database for games played between players rated 2200-2250 and players rated 2700+

Here is the breakdown :

• 253 games
• 16 wins for the underdog (8 in simultaneous exhib.) (6%)
• 39 draws (15%)
• 198 wins for the stronger player (~78%)

So you could say the results of the 2200s are somewhat better than what the elo tables would give (~96% for a 500 elo points gap)

"The Gap between 2700 and 2200 is due to having more patterns stored to memory: a 2200 may have digested and memorized FCE, Chess Strategy, BCE, and Dvoretsky's Endgame Manual (...)"

Hmmm, not sure many have, but I may be wrong

Simultaneous games shouldn't count, that's a big advantage for the lower rated player.

Having digested all of these books may be of less use than simply being a strategically limited, but very tough player.

true, also I was 2200 at one point and I have no idea what BCE and FCE are.

Here's one of those upsets I told you about:

One's chess understanding makes one a tough player though.  Though thought process and weeding out bad habits are also critical.  It takes work and any honest author like Heisman, Soltis, Dvorestky, or Aagard explicitly states such.