Tactics rating calculation

I have been trying to understand how the ratings are calculated in the tactics trainer.  Lately it seems that when I get a tactic correct I get one point but if I make an incorrect move I get up to -16. 

I feel that even though I am understanding the tactics better my rating is steadily getting worse. Should I believe the rating and if so how can I improve it? Any insights would be appreciated. 

This is basically the way it works:

https://support.chess.com/customer/en/portal/articles/1444811-how-does-the-time-and-ratings-work-

If you get the puzzle correct but take too much time, you get 1 point.

I don't think this formula is being followed.  Slagathore has not given enough information to test that hypothesis, but here are two examples from my own "Recent Problems" record:

Yesterday I attempted one problem whose rating was 57 points greater than my rating at the time.  I got it correct, but took 4.4 times as long as the average time.  I received +1.

Also yesterday I attempted another problem whose rating was 27 points less than my rating at the time.  I got it wrong (0%) but took 3.8 times as long as the average time.  I received -11.

Perhaps the scaling with respect to time is being done when we get the problem right but not when we get the problem wrong?  Whatever it is, it seems grossly unfair, and it is very frustrating.

In 2016, that link was correct. The process has changed quite a bit since then.

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

and

https://www.chess.com/blog/News/chess-com-dev-update-may-26-2017

Those have information on changes to the tactics that started at the end of April, 2017.

As for points when you get it wrong, time doesn't matter at all if you get it completely wrong. You lose the full point value of the tactic.

I don't see either of those links address the time scaling being applied when you get the problem right but not when you get it wrong.  Can you point to text in either of those links that addresses that?
The link in the first response ("How does the time and ratings work?", which is undated and thus ostensibly still in effect) specifically says the opposite.  "If you make all the correct moves you can achieve up to a full point for a win, and if you fail then you get a partial score based on how many correct moves you made. This 'move score' is then adjusted based on how long you took to solve (or fail) the problem compared to the average solving time." 

Moreover, I stand by my comment that the mode you describe is grossly unfair.  The time scaling should be applied without regard to whether points are being added or subtracted.

The tactic is worth a certain number of points. If you get it wrong, that's how many you lose. It may not be specifically spelled out but that is the way it works (or, at least I'm 95% sure ). Some tactics will give you partial credit and there may be a time-scaling calculation in that part. It wouldn't make much sense to scale on completely missing the problem based on how long it took you to fail.

As to the help articles, I don't think any of them keep a last updated notice on them. There are a lot that are outdated after the version change.

But, if you wanted to get clarification, you could open a ticket and maybe a staff member could clarify.

https://support.chess.com/customer/portal/emails/new

Why in the world wouldn't it?  For one thing, I don't know what the target time is until after I have attempted the problem.  It could be less than 10 seconds, or it could be more than a minute.  Say I am looking at a 10-second problem, and the clock rolls to a minute or two.  Maybe this is because I am worried that it's a one-minute problem and I want to be sure I haven't missed something, maybe it is because I got distracted by my children and then came back.  Either way, I am now faced with the prospect of only +1 if I answer correctly or -15 if I answer incorrectly (I also don't know the problem's rating until after I've answered).  This is totally contrary to how ratings are supposed to work.  In such a case, it would be better not answer the problem at all, but really that defeats the purpose of having puzzles in the first place. 

A lot of the tactics expect one to anticipate really stupid moves on the part of the

computer - like the computer putting the queen in harms way for no gain. Or else, there

are more than one ways to resolve. Or else, you are supposed to forgo a checkmate in order

to grab a piece in one tactic, but in another you are supposed to checkmate and ignore the

piece freely available. Still learn from this thing but it has glitches for sure.

Getting it wrong is just getting it wrong. It shouldn't matter how quickly or slowly you failed it. While the tactic trainer does have a heavy speed element to get the most points, speed should come with familiarity with patterns.

You already have a huge benefit when presented with a position. You know there is something there where in a game you don't know if there is a tactic or not.

In your hypothetical, if you get distracted, just move on to the next problem if you don't want risk a point loss or only a gain of one point. I'm sure plenty of people do that. Me, I just try to solve the problem. The only reason for the rating is to try and present the player with a suitable tactic; that's really it. As long at the system is doing that, giving suitably challenging tactics, some easier, some harder, then it is working well and the point gain/loss is just a number to help get the next appropriate tactic.

That isn't true. The first move by the player should always be the best move, regardless of the reply by the other side. There are times the line may not be the one you were most worried about or find the most difficult to deal with. There are times when a specific pattern is being trained, so the line may go that route to try and present that pattern instead of going with a different line.

Heck, I've had a tactics that start from the same position and go through different replies. That is part of the process.

There are times where there may be a winning line but if checkmate is also possible, that is the solution. That may or may not be a good tactic, depending on how you think, but ending the game is usually better than a line that might drag out for many more moves, or with no immediate mate in sight.

This time Martin happens to be really right. It should not and does not matter how long it takes to get the problem wrong, the penalty should be the same. For us lowly under 1800 players tactics are most of what determines who wins or loses (other than obvious careless moves).  So if you get a tactic wrong and the opponent doesn't, or the opponent capitalizes on your wrong play the result is the same, a loss. So if you lose the game because of your bad play, what difference does it make if you took 10 seconds or 10 minutes to make the mistake? The result is the same, a loss. There are no points for style. A loss is a loss and the points reduction in your rating is the same regardless of how you arrived at that loss or how long it took. 

These arguments either work both for getting it right and getting it wrong, or they don't work for either.  You're trying to have it both ways.

Chess.com gives ratings for different time categories -- Blitz, Rapid, and Daily (i.e., no time constraint).  Perhaps the Tactics ratings should also come in timed and untimed varieties, or the user should be allowed to select.

Yes, exactly.  And if the rating is adjusted in a skewed fashion, it will fail to achieve that purpose.

Are you suggesting that -11 was too little for the time or too much. Do you think you should get -1 cause you took longer then average to fail ?

If the change in rating must be affected by the time that was taken (and it is not clear to me why this must be), then it would make sense for the problem's value to simply diminish with time, whether it is answered correctly or not.

The premise of your question is that chess is only worthwhile if it is played quickly.  I do not subscribe to that idea.

So you think that people going slower should get less points if they fail ?

+1

Accuracy first, speed is a by-product. 

Yes, a game of chess is trying to have it both ways. You are right about that. You have to not make a mistake, while at the same time, capitalize on your opponent making one. The point of tactic trainer is to get better, or to win. To win, you must have it "both ways".  That means you are heavily penalized for a mistake you make, but rewarded very little if you do something right. If I get a tactic wrong I expect to lose the amount of points as if I had lost the game. If I get it right, I do not expect to gain the amount of points as if I had won the game. I think that's what training is supposed to do.

If they get fewer points when they succeed, then it would seem only fair for them to lose fewer points when they fail.