Tactics rating dropping down

I have been doing tactic trainer for daily basis for over year now. In total over 2800 tactics. (5 per day) My rating have been steady 1550-1650 for almost all the time, slowly increasing. But suddenly for last couple of weeks raiting have gone long way down. At start of the May rating was something +1600 and now little bit under 1500.
I don't understand why.
For comparison my ratings elsewhere are steadily ~1600 (I do those daily too) and tactic practice ~1900 (much overrated)

Is there possibly some changes in rating algorythm or I am suddenly getting worse in one site while getting better in others? I see that I'm getting problems out of my rating range, for instance today:

Getting -11 for losing 1614 rated problem. And:

Getting +1 points for easy 1084 problem. I know I was slow but still faster than avg time. I see this happening quite often; Some problems much easier (-400p of own rating), some much harder (+100 of own rating)
It's looks little weird and therefore my overall tactic rating have dropped over 100 points in short time.

And no, I don't play tactics for ratings to make it clear. Just to improve my game. This just feels weird and I'm interested. Any explaination for this?

Yeah I was around 1575 for a bit but then dropped to like 1200

Hello,

 

Yes, it seems that the algorithm was changed. I was ~1800 then I ended up to 1400 in a couple of days. More and over, I have done about 13000 tactics so I can tell if something changed. However, I believe that the change is a major improvement for the chess tactics algorithm.

 

Here is what could possibly be changed:

  - It is skimming through our tactics database and giving us what we are bad at. (The tactics puzzles are tagged so that is possible)

  - Every couple of tactics, the algorithm will throw a tactic that is ~200 higher than your rating. (This will stop you from playing passively)

  - Last of all, there also seem to be a change in the rating system as well. Getting a tactic right or wrong seems to change your rating in a more reasonable/fair way.

 

All in all, a change was definitely made, and it is certainly making us be more cautious and proactive while playing. So hopefully you can make the best out of it.

 

U Eid.

The problem selection algorithm changed and there were some scoring changes too. My rating actually shot up 200-300 points after the change but my last few sessions have been terrible and I have since dropped back down.

 

 

https://www.chess.com/blog/News/chess-com-dev-update-april-28-2017

 

 

There are supposed to be some additional scoring changes happening but don't know if they have been added. I need to catch up on the comments on that blog.

 

 

 

Thanks for the information. That update explains this rating drop.

I liked more of old tactic trainer because all tactics were challenging in right level. (Not too hard, not too easy) Maybe I just need to get used to this change.

I stopped using Tactics Trainer about a year ago when I noticed my scores steadily falling after they had been consistently rising. What the site didn't explain is that the tactics problems gradually get harder, so I had developed bad habits (tricks) to solve the tactics problems quickly, and those tricks didn't work on harder problems. In other words, I was more focused on getting a high TT score than learning tactics, which hurt me in the long run. Therefore I no longer use Tactics Trainer. I don't know if that is the same problem you're having, but it's another possible explanation in addition to the explanations other people offered above.

TT is beneficial if you don't take them too seriously. It's good for developing a sense of when a tactic can exist, as well as pattern recognition. 

You shouldn't use it for calculation practice though. 

Also, the new system rewards speed, which is stupid, but it's the way it works now. 

whatever was modified yesterday has made it so my rating doesnt constantly plummet (+1, +1, +1, -11; rinse repeat)

i have moved up a bit and have gotten access to some better puzzles and it "feels like it's working better?" the +/- fluctuations are greater but the best fix was "correcting" any wrong move (especially 1st wrong move) -11 even if puzzle's rating exceeds the player's

i have done more tactics trainer in last two days than i was doing since the changes in april and i feel more engaged and much less frustrated

That isn't new. The changes didn't impact the speed aspect at all. Everyone is misunderstanding the Target Time metric ... the old system also required quick solving to get 100%, on many tactics, only now you can see what that target time is after solving, where before you couldn't. I also think the time to get 100% was slightly expanded in the updates.

 

The TT has been heavily influenced by speed solving since I've been here. It's the reason I turn off the timer since I don't want to be influenced by it.

 A player need to develop the ability to calculate at least three moves deep, any player can do this. Linear calculation is important skill to develop and second is tactical pattern recognition. The op is missing simple tactics and mates and a possible reason could be can't calculate three moves deep, must see it in your mind clearly like a walk in the park. My tactical rating is going up because I put a lot in tactics study, and developing my tactical pattern recognition. Don't get discourage and keep at it, when you get it wrong review it and keep reviewing until you learn this pattern. I don't care how fast I solve a problem but I care if I get it right. Four years ago my tactical rating was at high 1900 to low 2000 and now I am at 2251 and trying for 2300.

I understand what's going on pretty well with TT. Before, you could only get +11 (for me, it was only like +7 for 100%) if you solved it very fast, and risked missing stuff and going -9 or something. So you would take your time and get +1, and do some of the easy ones quickly. 

Now, you can get +18 or some other high amount on most tactics, but you need speed. So speed is heavily rewarded in this new system. 

Did this change yesterday?  I went up about 50 points in one day after several days of not changing much.

 

I haven't tested enough tactics recently but +18, under the new algoritms, should only be likely for tactics that are much higher rated than your current rating.

 

I got +20 on a tactic 200 points higher rated than me when solved on the target time and +13 for one 200 lower rated. 

 

But my primary point was that the TT hasn't really changed when it come to the speed aspect to maximize rating gain. Just that it is somewhat easier to get the max than it used to be.

I really dislike that they are giving me lower rated problems and expecting ridiculous target times.

this...they said we will get both higher and lower rated puzzles than our current rating so it's more similar to real game but i often get 4/5 puzzles rated lower with target times 4 sec which is almost impossible to hit and get +1 for solving even if i take quarter of an average time to solve it and if i make a mistake it's -12 or something, it's getting absolutely ridiculous

I think the problem is that the Target Time does NOT require "quick solving," it requires "quick guessing."  HOW can you see a solution in 3-7 seconds in complicated position with playing Hope Chess?  Players are encouraged by the Target Time to abandon solid chess thought processes.

i agree w/ mickinMD that to score well in tactics trainer there is an element of faith with how one move might trigger the correct combination

over the last several weeks i have adopted using "that site which should not be named" to "train/learn" the patterns and tactics as there is no time factor and the fluctuations between right and wrong answers are not as drastic

chess.coms's tactics "tester" is where i see if i have gotten better and can try to apply what i have learned; it means recognizing a pattern quickly and acting on it-

i have gotten tired of the +1, +1, +1, -15 scoring when i try to really "size up" the situation here so i really try to spot and execute quickly and if i miss a few i quit and go to the other site where i can plod along (incidentally my rating there has been in the 1200-1250 range for weeks where here my rating can vary +/- 200 points easily)

i try to ignore the "number of the rating" but i cant totally- there is an element of satisfaction when it rises and frustration when it plummets

 

The TT here is more of a pattern recognition trainer but it is still very possible to solve in the Target Time and calculate the more pertinent lines in many of the tactics. That isn't really playing hope chess, though I'm sure a lot of players just go for the move that looks best after a quick glance. My guess is a lot of players would do that, no matter what the target time is or if the tactics were untimed. 

 

I generally try to calculate what I feel are the more pertinent lines and while I don't always look at every single reply or always calculate to quiescence, I calculate to the point I feel I have a good solution. Sometimes it works and sometimes it doesn't 

 

But the gist of some of the discussion points comes down to a fundamental design decision the site made on tactics, one that has been present since I have been here and isn't really a new change in how the tactics operate. I like the way it works, in general. I wouldn't hate it if they added an untimed and still rated mode or made the ability to customize how wide the range is, in the new method, but I also wouldn't care to lose the TT in the essential way it is now. I think it works well for pattern recognition and pointing out flaws in my move selection and calculation processes. 

 

Looking through your last 25 (and mine for that matter recently), it does appear that algorithm might be a bit weighted towards the lower end now. They may have also tightened up how much higher rated the problem selection goes. I had been getting at least around 20-30% 200-500 points higher rated and now it appears that I'm only getting them at about 100 points higher and a probably 60-70% 200 below my current rating. Problem selection should probably be 60-70% withing 200 points with the remaining as evenly split as possible above that range on either side. 

 

I don't think that is a terrible thing, and it does seem to be keeping my rating growth in check over the long haul. I still get  a very good selection of tactics that I miss and ones I get right but take a while to solve (fewer points). That is the primary thing I want out of the trainer, though I understand some want something different.

I've dropped from 1025 to 670 because of this emphasis on speed.  If I take the time *I* need, it's several +1s followed by a -15.  If I try to hit the time limits, I just go down faster.