More Advanced Statistics.

[ch0rn]

Concise description:
In addition to having an overall winning percentage, have a separate winning percentage for how many people are involved in the game. And possibly for different game types. Your winning percentage really is meaningless if one person only plays 2 person matches, while another person only plays matches against 8 people.

Specifics/Details:
ie..)
Overall Winning percentage :25%
2 players : 50% 9 games played
3 players : 30% 12 games played
4 players : 90% 1 game played
5 players : 0%
etc..

This of course can go much deeper, I am just suggesting a simple upgrade, but this can further more be broken down by spoils, maps , games including specific opponents etc. Would be nice to see how you do against people in your "friends" group. I play several private games against friends, it would be nice to have a breakdown of stats for those games.

How this will benefit the site and/or other comments:

It's interesting to see how you fare against other players comparing statistics.

[DoomYoshi]

How about Win Differential where win % is presented as a difference from expected wins? i.e. a win is recorded as a win per # of players, and every loss is recorded as one loss. What you are proposing will cause a lot more server strain.

[GoranZ]

How about Win Differential where win % is presented as a difference from expected wins? i.e. a win is recorded as a win per # of players, and every loss is recorded as one loss. What you are proposing will cause a lot more server strain.

Buhahahahahaha... Don't put your head in things you know nothing about. You have no idea how horrible you sound.

His proposal will not create any problems for the servers, it will be done on DB level and it will be super fast and easy to be implemented.

[DoomYoshi]

How about Win Differential where win % is presented as a difference from expected wins? i.e. a win is recorded as a win per # of players, and every loss is recorded as one loss. What you are proposing will cause a lot more server strain.

Buhahahahahaha... Don't put your head in things you know nothing about. You have no idea how horrible you sound.

His proposal will not create any problems for the servers, it will be done on DB level and it will be super fast and easy to be implemented.

The monthly challenge scoreboards cause problems for the servers, this is why we only show top 100. Every individual datapoint is one extra request.

[ch0rn]

How about Win Differential where win % is presented as a difference from expected wins? i.e. a win is recorded as a win per # of players, and every loss is recorded as one loss. What you are proposing will cause a lot more server strain.

I am all for any improvements, I do not know the inner workings of the database here etc. I just simply... if possible... would like to see more stats. They don't even have to be live, those stats can update on a schedule, if the server load is too high then once per day, or even once per week. A lot of the data can be converted over to static tables or files since once the games are over they don't ever change again. Stats don't necessarily have to be *live* to be *better*

I suppose that if the servers are already having a hard time creating statistics for simply top 100 of the month it will likely be out of the question to see statistics based on games with just a specific set of users. I play tons of public games, but also play a few private games with the same people , would be really cool to see stats just based on my private games with friends. In a perfect world we could somehow create a group and have our own internal score within our group. (But now I may be asking for crazy stuff! )

[molespe]

I personnally would find more useful having detailed statistics by map, sunny/fog, trench/no, spoils than over dice (certainly helpful when we become sore losers but much less priority)

[Arama86n]

I'm in favour of pretty much any change to the totally useless win ratio that is displayed on our profiles today.

Great initiative

[GoranZ]

How about Win Differential where win % is presented as a difference from expected wins? i.e. a win is recorded as a win per # of players, and every loss is recorded as one loss. What you are proposing will cause a lot more server strain.

Buhahahahahaha... Don't put your head in things you know nothing about. You have no idea how horrible you sound.

His proposal will not create any problems for the servers, it will be done on DB level and it will be super fast and easy to be implemented.

The monthly challenge scoreboards cause problems for the servers, this is why we only show top 100. Every individual datapoint is one extra request.

Thats stats he is proposing are easy retrievable, no extra requests, only the one that is being in use now... The difference would be that it will give more data.

I work with databases(MS SQL Server tho) ever day.

[Lord Arioch]

I dont know shit about large databases... but i would love more statistics. If wordfeud can manage a graph over your winnings over time why cant we:)

im supporting!

[ch0rn]

I have to imagine adding more stats won't take up that much server load if it's done on a schedule with cached or static data. I realize how much data there is, but it doesn't have to pull a query every page load, it can just updated on a weekly basis with only the new data added somehow. There's a lot of data, but it's at least all integers and simple data to create stats.

But either way, it would be a nice enhancement.

[GoranZ]

I have to imagine adding more stats won't take up that much server load if it's done on a schedule with cached or static data. I realize how much data there is, but it doesn't have to pull a query every page load, it can just updated on a weekly basis with only the new data added somehow. There's a lot of data, but it's at least all integers and simple data to create stats.

But either way, it would be a nice enhancement.

its super easy and its not server demanding... no need for updates

[ch0rn]

I have taken a swing at this , was not sure if I should post my write up here or in the tool development forum.

viewtopic.php?f=529&t=213911