What type of chess provides the most accurate rating?

Sorry if this has been discussed (even sorrier if it's been regurgitated), but which system provides the most accurate rating?  Be it online, blitz, standard, 960, tactics, or chess mentor, which do you think is most accurate? 

Obviously the answer varies depending on the individual, their game preference, and frequency of play, but what I'd like to know is which individual system, when played consistently with other systems, gives the most accurate description of progress?

I personally think tactics trainer is most accurate.  And chess mentor is the most inflated.  But I'd like to hear your thoughts.     

Standard chess. It simulates tournaments and combines with chess.com's accurate rating system.

I would guess that of those options standard would give the most accurate rating. It seems to me that a longer time control would just give more consistent results, because variables related to chance, such as blunders, would be minimized. I do not say online because the energy you put into the game might change from game to game. It is likely to stay roughly the same for non-correspondence time controls. This is just a guess of course.

How do we define accuracy here anyway? The accuracy of a rating may be based on how far it is from our "true" strength, but without a better system how can we know what our "true" strength is? Do we even have a true strength?

Each rating is the most accurate for what it measures (blitz for blitz, standard for standard, etc.).

IMO OTB (over the board) federation rated games (e.g. FIDE) are best for comparing overall chess skill.  This is where there are long time controls, and you know the players are serious (using their best openings, being focused and rested, etc).

Correspondence chess is for the chess scientists so to speak.  You get to exercise your researching skills and collaborate with a computer to test lines.

Tactics trainer / chess mentor ratings are just to track progress of those specific skills.  When you train with them you're not getting better at chess specifically just better at some of the tools used to play chess.  If tactics are your weakest area, then sure, your other ratings will go up.  If tactics are your strong point then upping your TT rating by 100 points probably won't make any difference in your actual games.

Blitz is blitz, it's a different kind of chess.

Standard games online may be long (say G/60) and are IMO best of the online ratings to determine your real skill (whatever "real skill" means).  As browni said in longer time controls there's less opportunity for stupid oversights and more of your chess knowledge/ability gets to shine through.

Still not as good as OTB IMO because there's no investment.  Go to a weekend tourney and these people are giving up their weekend (or weekend holiday) and money (gas, entry fee, hotel fee) to do nothing but play chess.  On average they're bound to be more serious and prepared.

Wafflemaster, I gotta disagree with that last part.  My TT rating increased about 100 pts recently, and I've been able to beat players that are rated higher than ever before.  So I think TT is a great tool for advancement in any type of chess, tbh.

Of course TT can lead to better results.  I even mention that in my post.  TT rating however is not in any way a good measure of chess ability overall.

Ratings measure performance, that's right, I forgot :p

But doesn't performance (over time) display strength?  How else can you measure it?

At first I was going to disagree until I started thinking about my own inflated standard rating here which is based on results inc a fair # of forfeit wins in games I should have drawn or lost. Still,  after someone has played several hundred rated games (blitz, turn-based, whatever) I'd think their rating is fairly accurate - maybe someday I'll recalculate mine by rating all the forfeit wins "correctly" ie adjusting my rating for the games I should've drawn or lost.

Sure, ratings measure performance (at least for now), but are they not an attempt to approximate strength? We could measure their accuracy as the confidence that our true strength represented as a rating falls within a certain number of points of our measured rating. It might be possible to measure or estimate ratings' accuracy mathematically but it would require certain assumptions to be made.

good point!

tt

We're dancing with words here, no?  Results are our only indication of strength.  Ratings measure results, results indicate strength, at least abstractly.  I understand there's no concrete measure of strength and that it's all really ambiguous, but one of these systems may provide a more accurate description of our ambiguously abstract rating .

The rating pools that are the most accurate judge of a player's strength (and that is ultimately the end here: using results to judge players' strength) are the ones that do not let players choose their opponents and have huge sample size of games amongst all the players in the pool.

good point, also!

I'm not exactly sure about tactics trainer. I have around a 2100 OTB rating and around a 2700 tactics rating, so does that mean that since I am 2700 in tactical skill then I am 1500 in positional play? Seems hard to believe.

Let's make it happen!

I think online chess....especially for old slow thinkers like me!

I believe that each form of chess is a different game. As such they require different skills in order to master them. For instance the main skill in bullet chess is the ability to move very fast. Probably a huge percentage fo those games are decided by the clock. Therefore your rating for each type is solely an estimate of how well you can expect to do at that type of chess. Queen_of_Knight, you are looking for something else. That mystical number that says accurately, how well you understand Caissa, the soul of the game, and right now, you believe that your tactical rating best defines that for you.

Well, for now, I would say that you are correct. Most of your games against 1800 and below will come down to finding a tactic, that your opponent has overlooked. I am in the same boat there, but have noticed that, as my opponent's rating goes up, I am offered less of an opportunity to capitalize and am forced to beware my own blunders which are immediately snatched up and thrown in my face. Eventually both sides have the same skill at tactics and nothing more can be accomplished with this weapon.

That is when the level of positional understanding comes to the fore. Or so they tell me. Games between experts and IMs are said by my higher rared friends to have examples of this. Who knows what differentiates the Grandmasters? That is far beyond my mortal musings.

I'm getting sick of agreeing with you MJ4H

Perfect example. Just finished this post and went to check on my games. I have 12 going right now, but there were only 11. This was strange because I was not in a great position on any board, neither was I expecting mate in one. Come to find out, one of my opponents had moved a pinned knight, allowing me to take his queen for a bishop. His rating was 1575. This type of tactical gift does not happen against someone rated 2000. There positional understanding gains precedence as a deciding factor in the game. It will still look like tactics to us grumpkins, because the fact of the well posted knight and better pawn structure which eventually forced the bad bishop to a square where it was forked will escape our notice. But it is there, just the same.

Winter is coming.