Average Time/Move Limits

Many tournaments, such as the chess.com Rapid one currently open for registering, have a maximum average time/move of say 3, 6 or 12 hours. This all totally makes sense as it stops all the slow players from getting in and ruining it for everyone else. I get that. Trouble is, my average time/move is currently 18 hours, so I am excluded from many tournaments. From when I became a premium member up until several months ago, I was in way too many games, and tended to almost let the time run out before making a move, just to get by, as I was in 250 games and didn't really have enough time. This obviously made my time/move rocket up, to around 21 hours. But my problem now is that even though for the past 3 months I have been playing really quickly, probably under 3 hours per move, and I've got my games down to an easily manageable number, my average time/move is calculated over the entire time on chess.com, so it remains huge, at 18 hours. So now, even though I make my moves really quickly and could keep up with these tournaments, I get excluded, and end up stuck in the tournaments with the sandbaggers! It would take me years of moving quickly in say 50 games to make up my time of moving slowly in my 250 games...

My main point is that I think the time over which average time/move is calculated should be 3 months, like the timeout ratio, rather than my whole time on chess.com; this would be far more accurate of a player's speed!

Thanks, would welcome any response

I agree because I have the same problem

If you take a look at how the time/move is calculated, I think you'll see why it isn't done over a 90 day period.

https://support.chess.com/customer/portal/articles/1444878-how-is-average-time-per-move-calculated-

To only have it done over 90 days would be fairly intenstive and the system would have to track the time/date of every move and check for aged out moves after every move. Then, the sytstem would have to recalculate the average based on all the changes.

I'm sure the system could be make more efficient (maybe not doing the calculation after every move, maybe once a day for example) but it is still a lot more intensive than the basic method from above.

One way to decrease you average is to make conditional moves whenever possible, since they take no time and can only decrease the value.

I don't have a problem with it but currently I have a time/move of 4 hours 30 min. I want it to get down to less than 3 hours.

6 min 45 sec is my time/move. I don't have problems anywhere.

Thanks for that link Martin, I can see why it would be complicated. Maybe if they were to do the last 100 completed games or something, though that could be difficult also. In my particular case it would be useful if the time/move counter could be reset so it would have a more true and recent value, but I guess that'd but pretty difficult too... I will try to make more conditional moves but it's a big hassle on the iPad, so it sounds like I'm stuck with this problem for some time!

Interesting to see I'm not the only person with this problem, would be handy for us if something could be done :)

I agree. Although I'm at 3 hours for my average I like the reset idea despite the work and difficulties

it takes ages to decrease time per move to be able to play tournaments. Stupid calculation.

Hi all,

Please add me to the, "let's re-think this calculation method."

My current move time is 3 1/2 hours, been trying to knock that down below 3 hours for what seems forever and can't get the rate to budge. 

There's got to be a better way to ensure people of good will don't get locked out of events because of a program issue.

-Cheers

Agree that the average of the lifetime number of moves is very inflexible. I have gotten to play faster, but my average move time diminishes very slowly because of the thousand early games where I moved slow.

I suggest an exponential averaging which is not harder to implement than the current linear one. It does not require to keep track of extra information.

Some other chess sites compute the average time over the last 100 moves, but that requires to keep track of the last move times - which is why this is avoided here I assume. Exponential averaging does not have that problem.

@SageBen: Yes, your proposed formula would be much better than what is implemented now! I hope the current averaging gets changed to what you propose.

My current average time per move is down to 3h45min over 1250 games. If I play now at an actual time per move of 2 hours, I only need to play 1875 games of the same average length to get my average speed down to 3 hours per move. I better get playing...

Find someone that likes to play correspondence style chess fast. Play multiple games per day. If your moves come in at minutes per move, you can probably shave a number of games off your estimate and the time it will take

Yes, but not everyone that plays correspondence has the ability to immediately be ready to move. I am sitting around 5.5 hours/move after a only a few dozen games because my opponents choose to play their moves at inconvenient times. The first 5-10 book moves are usually pretty quick, then someone moves right as I go to bed and even if I respond immediately when I wake up, it's still an 8-9 hour move that increases my time. It happens while at work too. 

I had dropped my time/move down to about 4 hours but then after my sister's wedding, I had 6 games with over 36 hours since the last move. I gave up on getting under the 3 hour mark for tournaments since it would require me to check my phone app or computer 2-3 times an hour. That is not feasible.

I normally think about a position for 15-20 mins after I make my move and have a few responses in mind, but some positions take a little more time. When my opponent makes an unexpected move, I assume few of us are able to quickly calculate and find the right move in a matter of minutes on the chess.com app while driving home from work. 

I would like the formula to start the clock from the "move seen" point. So if my opponent plays his move at 1am my time, and I don't see it until 7am, I do not get penalized for those 6 hours I'm sleeping. There are loopholes to this, but possibly prevent any spectators from seeing the last move until both accounts have viewed it. Obviously your move clock does not wait, but the timer that checks your move/time should. 

That's why I suggested finding someone that likes to play fast. The idea being you ideally would have the same online hours and basically treat the games as a live one. Just an idea on how one could go about getting their time down.

Or just wait for a glitch ... my time/move is 4 hours and I'm pretty confident I don't move that fast. At one point, it was around 24 hours or so and one day I got on and I was around 3 hours.

my average time is "a few minutes" lol!!

@skotheim2: Impressive!

Three months later and my average dropped from 3h45 over 1250 games to 3h30 over 1396 games. So sometime next year might get the average down to 3h, needed to join many fast tournaments. Well, at least I have a chance of ever getting there in spite of the sad implementation (livetime average instead of current average)

At 1597 games my average came another quarter down to 3h15.
That allows me to estimate my recent actual average AA move time:

1597 x 3h15 = 1396 x 3h30 + 201 x AA
1396 x 3h15 + 201 x 3h15 - 1396 x 3h30 = 201 x AA
201 x 3h15 - 1396 x 0h15 = 201 x AA
(201 x  13 - 1396) / 201 = AA in quarter hours
13 - 1396 / 201 = approx. 1h30

If I keep the current rate of 1h30 / move up, then I will reach my goal of a life time average response time of 3h/move with x extra games 

12 - 1597 / x = 6
x = 1597 / 6 = 266 games

So there is hope that I can join the chess.com fast tournaments after a few extra months of playing.

As said, it is unfortunate the chess.com uses the lifetime move response time to allow one to enter tournaments rather than an measure based on recent games. For some folks it means that they will only be able to join the fast tournaments by creating a new account. A pity when it would be easy to change the implementation to for example what is proposed by SageBen in his post above.
 

i have the same problem : i'd like to play quick tournaments but my av. time/move came down to 3h30 so i can't participate anymore. Hope it will change one day. i will follow this post.

Doing the calculations based on a time standard would be difficult.  However it would be very simple to use the existing system to weight newest moves more.

After each move multiply the moves and move time by, say, .995.  After 900 moves, your first move is basically only 1% as important as your last.

Under the current system if you played your first 1,000 moves at 7 days average, then played your next 1,000 at 2 hours 45 minutes average, after 2,000 moves you'd only be down to just over 3.5 days.  Under this method you'd still be at 7 days after 1,000 moves (if all your moves were uniform) but you'd be down to 3 hours 50 minutes after 2,000 moves.

Since your most recent performance is more important (backbone of the Elo / Glicko ratings after all) this would presume your most recent 20ish games are more indicative of your current speed.  Obviously you can change the multiplier factor to make it whatever number of games / moves you want.  (A .999 would take prior example after 2,000 moves from 7 days down to 2 days which is still a good improvement.  It would take that person with 1,000 7 day moves about another 6,000 moves / 150 games to get under 3 hours playing at 2:45 pace.)

This would require the database be changed from int to float for those columns, otherwise you get some weird rounding errors once moves hit a threshold.  I'm sure there's a way around that though in calculation if needed.