Please turn on javascript in your browser to play chess.
Site Ideas Forum

Site Ideas Forum

  1. 28 Jul '10 14:23
    Not sure if this has been suggested, but it would be nice to be able to see the average number of days/hours that a user takes to complete games. This may be a more accurate representation than the move frequency choice in the profile.
  2. 28 Jul '10 15:39 / 1 edit
    Originally posted by micahchu
    Not sure if this has been suggested, but it would be nice to be able to see the average number of days/hours that a user takes to complete games. This may be a more accurate representation than the move frequency choice in the profile.
    How would you measure this meaningfully?

    For instance for myself, Game 7259276 started way back in March 2010 is still ongoing however Game 7588023 started on the 9th July 2010 and is finished already.

    A better idea if the total moves made in the month but again this is flawed as

    Player A with 800 games moves once per day in each game however Player B only plays 10 games at a time and moves 30 in each game per day so makes 300 moves.

    Player A is the most active but the volume of games means he is actually a slower player than Player B who is likely to complete each game in under 2-3 days.
  3. 28 Jul '10 15:43
    Originally posted by adramforall
    How would you measure this meaningfully?

    For instance for myself, Game 7259276 started way back in March 2010 is still ongoing however Game 7588023 started on the 9th July 2010 and is finished already.

    A better idea if the total moves made in the month but again this is flawed as

    Player A with 800 games moves once per day in each g ...[text shortened]... is actually a slower player than Player B who is likely to complete each game in under 2-3 days.
    Moves per game per month maybe?
  4. 28 Jul '10 16:00
    What about number of moves during the last 30 days, divided by number of games in progress?

    Now we have number of moves since the 1st this month. Rather meaningless in teh beginning of the month. The last 30 days would be much better.
  5. 28 Jul '10 16:51
    This is exactly the issue with stats, you can get multiple meanings out of them if you choose to look at them with no context.

    I would always look deeper into their profile to see that there were examples like the ones you provided, and would not go exclusively by the game length stat, but it would obviously be a useful stat.

    I do agree though, that the number of moves per game per day could be useful too, possible more so, and definitely if both stats were added, we would really have useful information.
  6. 28 Jul '10 19:10
    Originally posted by FabianFnas
    What about number of moves during the last 30 days, divided by number of games in progress?

    Now we have number of moves since the 1st this month. Rather meaningless in teh beginning of the month. The last 30 days would be much better.
    Lets say you made 600 moves in the last 30 days.

    At the start you have 30 games in progress. Of these 27 were in closing stages and have now finished.

    So your stats would show

    600 moves / 3 games in progress = 200

    If your games hadn't finished until the following day your stats would show

    600 moves / 30 games in progress = 20

    So the figures are pretty meaningless
  7. 28 Jul '10 21:41
    Originally posted by micahchu
    This is exactly the issue with stats, you can get multiple meanings out of them if you choose to look at them with no context.

    I would always look deeper into their profile to see that there were examples like the ones you provided, and would not go exclusively by the game length stat, but it would obviously be a useful stat.

    I do agree though, that t ...[text shortened]... sible more so, and definitely if both stats were added, we would really have useful information.
    If you're looking to gauge how quickly the player moves, an average time per move is a good stat. Weight it toward the most recent months and you might have something there.
  8. Subscriber Kewpie
    since 1-Feb-07
    29 Jul '10 03:24 / 2 edits
    From the FAQ of another website:

    Average time per move is calculated and adjusted each time you make a move in an online chess (correspondence) game. When a move is made, the total number of seconds passed since your opponent moved is added to your "total move time" and your "total move count" is incremented by one. (Note: Seconds passed while on vacation are ignored.)
    Note that conditional moves count towards the average time per move. A conditional move counts as 0 seconds and therefore will lower your average time per move.
    Your average time per move is simply "total move time (seconds)" / "total move count".

    Example. Assume Joe has a "total move time" of 0 and a "total move count" of 0.
    Joe makes a move in one of his games 18 hours into his clock. Joe makes another move in another game 1 hour into his clock. Joe makes another move in another game 48 hours into his clock.
    "total move time" = (18 + 1 + 48) * 3600 = 241200 (seconds)
    "total move count" = 3
    "average time per move" = 241200 / 3 = 80400 = 22.3 hours


    Seems to me that the resulting number would be extremely useful when choosing opponents. Some people move as often as they can, others move as infrequently as they can. Both of these behaviours can be annoying to others. This would make it clearer if you belonged to one or the other group.
  9. 29 Jul '10 04:23
    Originally posted by adramforall
    Lets say you made 600 moves in the last 30 days.

    At the start you have 30 games in progress. Of these 27 were in closing stages and have now finished.

    So your stats would show

    600 moves / 3 games in progress = 200

    If your games hadn't finished until the following day your stats would show

    600 moves / 30 games in progress = 20

    So the figures are pretty meaningless
    Well of course they are meaningless if you calculate them that way. I hope you don't play chess the way you reason

    Try looking at the problem more than one way. Each game has a start time and a end time, which gives you a length of the game. This also does not need to wait until the end of a game, you could use the current time as the end. Take the length of game and divide that by the number of moves, gives you the number of moves per time unit. You can then add up this number from several games, say however many games you have played over the past 4 weeks, and divide by the number of those games, giving you a nice average of moves per whatever length of time you used to begin with.

    Obviously this may be skewed by some extreme games, so you could drop off the games with the highest average and the lowest average.
  10. 29 Jul '10 05:35
    Originally posted by adramforall
    Lets say you made 600 moves in the last 30 days.

    At the start you have 30 games in progress. Of these 27 were in closing stages and have now finished.

    So your stats would show

    600 moves / 3 games in progress = 200

    If your games hadn't finished until the following day your stats would show

    600 moves / 30 games in progress = 20

    So the figures are pretty meaningless
    Meaningless? No, I don't think so.

    He makes 20 moves in each game on average in the last 30 days. Crisp, I think.
    If your game with him lasts 40 moves, then it will take statistically two days to compleate a game.

    I never play a game for myself, I play it with another person. Some opponents I play resign when he loses his first pawn, others play every game out tot hte last move with mate. Only when you have that stat too, you will know how long a game will last, still statistically.

    In my proposition, you can find out how active he is. How many moves he makes, in relation to how many games he has.

    My point is that the numbre of moves made since the 1st of the month doesn't say much in the beginning of the month. Number of moves made the last 30 days does. This with the number of games currently played (as is already today) gives how active, statistically, he will be in our games.
  11. 29 Jul '10 12:16
    Originally posted by micahchu
    Well of course they are meaningless if you calculate them that way. I hope you don't play chess the way you reason

    Try looking at the problem more than one way. Each game has a start time and a end time, which gives you a length of the game. This also does not need to wait until the end of a game, you could use the current time as the end. Take t ...[text shortened]... extreme games, so you could drop off the games with the highest average and the lowest average.
    Originally posted by FabianFnas
    What about number of moves during the last 30 days, divided by number of games in progress?

    Now we have number of moves since the 1st this month. Rather meaningless in teh beginning of the month. The last 30 days would be much better.




    The calculations were based on the above.
  12. 29 Jul '10 12:25
    Originally posted by FabianFnas
    [bIn my proposition, you can find out how active he is. How many moves he makes, in relation to how many games he has.
    [/b]
    Only so long as he maintains an average number of games.

    However, a sudden influx of games, tournaments starting etc, which will skew the averages.

    You could have someone averaging 20 moves per game per day but suddenly get a further 20 games because they have advanced in a tournament or joined a new one.

    400 moves over 20 games = 20 moves average per game

    Further 20 games added gives

    400 moves over 40 games = 10 moves average per game

    Obviously with an increased gameload it is likely they will play less moves unless they allocate more time to the site to make more moves.

    Statistics, lies all damned lies.
  13. 29 Jul '10 13:05
    Originally posted by adramforall
    Only so long as he maintains an average number of games.

    However, a sudden influx of games, tournaments starting etc, which will skew the averages.

    You could have someone averaging 20 moves per game per day but suddenly get a further 20 games because they have advanced in a tournament or joined a new one.

    400 moves over 20 games = 20 moves av ...[text shortened]... ess they allocate more time to the site to make more moves.

    Statistics, lies all damned lies.
    Right you are. Skip the division with number of game in progress. It confuses things.

    I'd be happy with number of moves the last 30 days (and not the since the 1st in the current month).

    (I log my own current games with times in a Excel sheet. There I can find out about my gameload. With thoo much gameload I don't start any more games for a while.)
  14. Standard member randolph
    the walrus
    30 Jul '10 03:31
    Originally posted by adramforall
    Only so long as he maintains an average number of games.

    However, a sudden influx of games, tournaments starting etc, which will skew the averages.

    You could have someone averaging 20 moves per game per day but suddenly get a further 20 games because they have advanced in a tournament or joined a new one.

    400 moves over 20 games = 20 moves av ...[text shortened]... ess they allocate more time to the site to make more moves.

    Statistics, lies all damned lies.
    The solution is to do [(Moves)/(Average games in progress)]/Month
    With "Average games in progress" being a weighted average of the number of games that were going when each individual move was made.