There's a difference between a session having a low accuracy because you're playing badly and making mistakes and a session with a low accuracy because you're playing quickly. Making more mistakes at your normal speed will obviously lower your rating but moving more quickly will (generally) gain you more points from the ones you get right than you'll lose from getting more wrong.
I have two accounts there. My main account's rating hovers between 1500 and 1550 with my recent accuracy at about 83-84% (my overall accuracy is lower due to much lower accuracy when I started). I started another to try to play at much higher accuracy, 98% was my goal. I fell a little short of this, 97% I think, and the rating of that account is about 70-80 points lower because of the longer time I have to spend to get that many problems correct.
Originally posted by heinzkatwell, firstly, one day is irrelevant. the fluctuation between your good and bad days is much greater than the difference you got. any possible effect would've been masked just for that alone. you need at least a 1000 problems over a number of days to get any meaningful data. and much more at high accuracy. if one day meant anything, my 'average' performance would be 1660 at 100%, but from practice I know I can't hold 95% at more than 1630 over thousands of problems (ie. I can't sustain the level, it was just a statistical anomaly).
To illustrate the point I have just done 300 positions, and quite badly, 52 were wrong 😕 Anyway that is 82,7 % (somewhat better than 18/20) and my rating dropped from 2051 to 2019. So how is a lower accuracy going to help achieving a higher rating??
and like I said, at 2000+ you have only a handful of peer problems, a minimal amount of higher rated problems, and multiple times more lower rated problems. the result is that your average problem is considerably lower rated than you. which in turn means that your penalty from failed problems on average is much greater than for someone closer to the average of the total problem set. you don't get a balanced sample of problems like the median guy at 1500, but a biased one. and that bias must be overcome by greater accuracy, so the 67-70% hotspot for most of us moves higher accordingly. still, the difference will remain, it'll remain below 100%.
come to think of it, the limit probably isn't at 100%, but at whatever is connected with the 'escape velocity function'. which would mean the 'escape rating' isn't the same for everyone, but instead relative depending on the individual rating AND individual accuracy. sort of like the triple point of water is related to both pressure AND temperature? on addition to that there's the memorization which will scrap any precise mathematical tries to express the relation. (well you could come up with a function of course, but it wouldn't include the human factor of memorization, which would make it worthless).
that's how I see it anyway.
and you aren't the first one to try it on the top levels. it wasn't that long ago when IM 'tomohawk' tried passing 1800 with 98%, and lost huge chunk of points from his 'normal' accuracy (which was relatively high to begin with).
but, but, there will always be the human factor of memorization. at some point you will memorize all your couple of hundred problems and ace them. at 1600 it's thousands instead of hundreds of problems, so it'll take (considerably) more time. so, in the end, none of the statistical math will hold. and the closer you get to complete memorization, the less the math will hold.
Why are the vast majority of puzzles on CTS only one-movers? 🙁 This makes the puzzles much easier - even some of the 'harder' ones are simple captures or knight forks!
Surely this isn't as beneficial as having a variety of anything from one to seven or eight movers, thus helping visualisation too?
Originally posted by skimsfor calculation training, chesstempo.com 's standard mode is much more suitable.
Why are the vast majority of puzzles on CTS only one-movers? 🙁 This makes the puzzles much easier - even some of the 'harder' ones are simple captures or knight forks!
Surely this isn't as beneficial as having a variety of anything from one to seven or eight movers, thus helping visualisation too?
but CTS's problem set gets harder as the rating of the puzzle gets higher, so there isn't only 1 movers there.
Originally posted by skimsDifficulty of the solution isn't that important - the higher the ratings get, the more complex and confusing the positions are...
Why are the vast majority of puzzles on CTS only one-movers? 🙁 This makes the puzzles much easier - even some of the 'harder' ones are simple captures or knight forks!
Surely this isn't as beneficial as having a variety of anything from one to seven or eight movers, thus helping visualisation too?
Originally posted by heinzkatIn what way do they get more confusing?
Difficulty of the solution isn't that important - the higher the ratings get, the more complex and confusing the positions are...
I had a look at some puzzles rated over 1800 on there, and they were still simple 2-move combos or capturing of pieces 🙁 I don't know if that's representative of all the puzzles, but I've seen that most of the puzzles are only one move long 🙁
Could anyone give an example of a complicated puzzle from the site?
Thanks!
Hi Ironmanrob I suppose you have to try it yourself to see how confusing they are. For a good example though there is a video by an (inactive?) International Master solving the puzzles on CTS where after a while he remarks how confused he is.
http://www.chessvideos.tv/forum/viewtopic.php?t=2004
And a concrete example, currently rated 1635
Six seconds. White plays 1. Qxd5. Your time is ticking 😛 The solution is easy, but do you see how confusing it is? In the initial position you want Ne5. Not sure how they generated the problems this way. It's great 🙂
Originally posted by CCNoobI'm not surprised at Ironmanrob's silence, I'm surprised fat lady and the gang do not walk the talk at CTS. Or is it too fast for someone with sausage fingers?
Ironmanrob, you should have no difficulty reaching 2000 or 2200 on CTS then? What is your handle there?
Originally posted by CCNoobI really struggle on the Chess Tactics Server. I just can't solve them fast enough. Nothing to do with sausage fingers, just a brain which has slowed down with age. I still visit there fairly often, but I just log-in as a guest. If I finish a session with a rating over 1700 then I count it as successful. However I almost always get 100% of the problems correct.
I'm not surprised at Ironmanrob's silence, I'm surprised fat lady and the gang do not walk the talk at CTS. Or is it too fast for someone with sausage fingers?
Originally posted by CCNoobyou have 25 problems under your belt, and you're running your mouth at strangers about it??
I'm not surprised at Ironmanrob's silence, I'm surprised fat lady and the gang do not walk the talk at CTS. Or is it too fast for someone with sausage fingers?
ccnoob
http://chess.emrald.net/tProfile.php?TacID=32797