Plummeting on tactics trainer

I've gone from 2300 to 1950 on tactics trainer in a few weeks. Just wondering has there been a readjustment to make it tougher as I generally get seven or eight points when I get a problem right and lose 15-18 when I get one wrong. Before I get a 'ratings mean nothing' lecture, let me stress I'm not necessarily complaining, just curious to know whether there has been a change, or if I'm losing what little tactical ability I had...

Campione: I too seem to be working harder just to stand still in the tactics 'ratings'.  So I share your interest.  The way that problem ratings are adjusted automatically should mean that they find their 'right' level. So a blanket readjustment of problem ratings would only change individual's ratings temporarily.  I think that a 'permanent' reduction in individual ratings would require a change in the way our ratings are calculated (e.g. fewer points for passing and harsher penalties for failing).  As you observe, the stand still criterion at the moment is (more or less) solving two thirds of the problems.

One problem with Tactics Trainer is that as time goes by, the amount of time given for a particular problem goes down.  This skews difficulty ratings.

I doubt though that this would make a difference as described in the op.

I find that my TT rating fluctuates pretty wildly.  The biggest factor that I can identify in that pattern is that if I really start caring about my rating, I find myself very conscious of the time bar running down and, not wanting to lose points by taking too much time, I hasten to jump at an answer before I've thought it all the way through.  This results in more misses, and hence lower ratings.  When I relax, and let the time run out if need be, I find the right answer more often, and then lose 8 or 10 points instead of 15.

I use TT a lot, but the truth is that I don't like it, and the reason is the time element.  i think it would be a better tool if every problem took two minutes, plus perhaps an "increment" for every correct move in a sequence.

My TT ratings chart looks like a very steep roller-coaster. I've reached mid-2000 and dropped down to 1600 in a day, and then climbed back up. I hit my highest at 2072 last night and dropped to 1700 this morning. I'm back up to 1921 atm. It all seems very variable.

I've had a few roller coaster rides myself.  Seat belts fastened !!

Yeah I've had fluctuations before, be it a bad day or whatever, but I do suspect that some adjustment to the way they are calculated may have occurred, as Iain suggests, as this has now been  a steady decline over numerous sessions. Probably no harm if that is the case, as I don't think I'm good enough to be rated 2300 at any aspect of chess, but it would be good to get the official line on this.

Meadmaker that's a good point about relaxing and taking your time, I find I have my most successful sessions on TT when I make a conscious effort to do that.

I don't know if the rating method has changed recently but I think it did a couple of months ago. I know for a time, I was hovering in the mid-1800s and over a couple of weeks bumped up into the 2000s, with a high of 2178. While I would like to think that I got better, I know I didn't get that much better that quickly.

In watching my problems, I am noticing a lot more +15 to +20 where in the past it was always +10 or less unless I got it close to 100% and I think the highest I had ever seen before was +15. Though, last week I had a bad day where I dropped back down into the 1700s and struggled back up. I was getting a lot more of the sub-40% solve rate problems. So I guess it's possible the problem selection algorithm was changed some too.

My favorite thing recently has been TT giving me the same problem immediately after failing it. It's a nice reinforcement but it does skew the problem rating and time somewhat as I can almost always get 100% the second time

I don't think I've ever noticed a +15 or a +20. It seems the highest I've ever gotten is a +14 or so. I've seen lots of -14s though.

My Rating: 1638 (+16)
Problem Rating: 1637 (-16)

I just got this for 100%.

Honestly don't know if this is out of the ordinary though - I almost never get maximums.

EDIT:

My Rating: 1643 (+13)
Problem Rating: 1576 (-13)

And I just got this for 100%. I don't think anything has changed.

Maybe it would be a nice idea to offer a non-timed (or long-timed) mode of the Tactics Trainer (with separate rating) ... with the short times now, it's great for pattern recognition and blitz training, but it would be nice to have time to really ponder about a position, to work it out mentally some moves ahead before even moving a piece ... at the moment it's kind of trial and error, and I make the moves mostly on intuition: I immediately make the move that looks best without calculating anything, if it happens to be correct then I look for the next move the same way ... I guess that's great for blitz but as training for long games it could be better with more time.

I don't think there have been any readjustments. We all have our ups and downs, and they often result in streaks of "solved" or "failed". Also, the magnitude is different. Some people manage to lose/gain a few hundred points. I am more stable in this respect, usually between 2700 and 2800.

It may be just the rating difference between my current rating and the puzzle rating. I have a couple of +21s, a +19, and a couple of +18s in my recent problems list. The +21s came from 100% and 100+ rating differences (problem 100+ higher than my current rating) and the +18s and +19s were with 90+% but the rating differences weren't that great (about 50 points).

I don't have a large point loss in the recent list, but I have had some where I lost points in that magnitude also. It could be that the problems were rated 100 points or more less than my rating at the time so I lost a lot more points for missing them. It would make sense based on how ratings work.

Add to that the fact I have recently been getting a lot of low success rate puzzles, and a wider range of ratings (from +200 to -200 over my current rating), it could just be that the algorithm that chooses the problems have been modified, and failures in a much lower rated, low success puzzle, will drop the rating a lot. Get a few of those in a row and it would be easy to get a big drop.

I never saw that kind of rating increase provided before a couple of month ago; though if you read the comments the ratings formulas have changed over the past couple of years. I think some tweaking is still being done.

One thing I do dislike about TT is being hit with -15 when the computer analysis shows rather subtle distinctions between my move and the 'right' move.  However, it's a great facility to have and certainly lets you know where you stand in that aspect of the game.

I recently joined Chess.com and began using TT.  Initially, I would receive 80% if I correctly solved the problem in the average time.  This week, I noticed that my rating was falling and that when solving in the average time, I was only earning 70% and that I had to solve in half the average time to get 80%.  My failures continue to cost me about 16 points.  I guess that it really doesn't matter what the system is, but it should be consistent IMO.  I would just like to understand the reason for the change.