Results 1 to 8 of 8

Thread: Winning Probability between 2 clans in a war.

  1. #1
    Junior Member
    Join Date
    Oct 2020
    Posts
    71

    Winning Probability between 2 clans in a war.

    How about adding something new to the game which shows a winning probability between two clans in a particular war? Wouldn't it be interesting to see what it predicts/which clan has the higher chance for winning?

  2. #2
    rowman's Avatar
    Join Date
    Dec 2014
    Location
    On a lenghty sabbatical
    Posts
    11,930
    In order to calculate probability, you must first have data, and a valid model, to calculate that probable outcome upon. Exactly what data can be used to begin with, using what statistical model?

    And no, Vegas type odds are not probabilities. They are simply set via public opinion (how much is being bet on which side). Do they poll everyone to see whom they think will win prior the war? Of course not!

  3. #3
    Super Member LemonCream099's Avatar
    Join Date
    Aug 2020
    Location
    South Africa
    Posts
    784
    I agree with Rowman. This could discourage clans before a war

  4. #4
    Junior Member WhiteeWolf's Avatar
    Join Date
    Oct 2020
    Location
    At the snowing desert
    Posts
    98
    Quote Originally Posted by rowman View Post
    In order to calculate probability, you must first have data, and a valid model, to calculate that probable outcome upon. Exactly what data can be used to begin with, using what statistical model?

    And no, Vegas type odds are not probabilities. They are simply set via public opinion (how much is being bet on which side). Do they poll everyone to see whom they think will win prior the war? Of course not!
    Quote Originally Posted by LemonCream099 View Post
    I agree with Rowman. This could discourage clans before a war
    Dont be so strict, ok the idea has problems but the average townhall level of each team could be culculated.

  5. #5
    Senior Member
    Join Date
    Sep 2020
    Posts
    483
    If the matchmaking is based on similar weight, I'd suspect the probability would be pretty close to a 50/50 shot that either clan has the chance to win. You'd only get a more lopsided probability in CWL, but no one needs a statistic to know when you are completely outmatched! Therefore I don't see value in doing this since it's not precise, has no bearing on the outcome, and only messes with the psychology of the two clans (and probably moreso to the predicted underdog).

  6. #6
    rowman's Avatar
    Join Date
    Dec 2014
    Location
    On a lenghty sabbatical
    Posts
    11,930
    Quote Originally Posted by WhiteeWolf View Post
    Dont be so strict, ok the idea has problems but the average townhall level of each team could be culculated.
    I did not question if it could be performed, but rather how (by what data, and what model).

    If you think that there is a model, then please do try it, and let us know the results. In the next few clan wars your clan is in, compute the average TH level for each side. Repeat for 20 or so wars. Then ask, was there a correlation between that average and whom won, or was it just random? Standard statistical analysis process. Before TH average, or any other statistic is used as a outcome probability indicator, it first has to be shown that it is an indicator. Else, it is nothing more than a coin flip, a complete guess.

    You may be correct that it is a predictor, but to my own knowledge, no war outcome predictor has ever been proposed and then proven.

  7. #7
    Senior Member
    Join Date
    Sep 2020
    Posts
    483
    Quote Originally Posted by rowman View Post
    I did not question if it could be performed, but rather how (by what data, and what model).

    If you think that there is a model, then please do try it, and let us know the results. In the next few clan wars your clan is in, compute the average TH level for each side. Repeat for 20 or so wars. Then ask, was there a correlation between that average and whom won, or was it just random? Standard statistical analysis process. Before TH average, or any other statistic is used as a outcome probability indicator, it first has to be shown that it is an indicator. Else, it is nothing more than a coin flip, a complete guess.

    You may be correct that it is a predictor, but to my own knowledge, no war outcome predictor has ever been proposed and then proven.
    Agree with all of this. So then it becomes a cost/benefit discussion. SC has limited resources, as all companies do, so is this the best place for them to spend time on improvements? To do data analysis to come up with a predictor index vs other game improvements? Not IMO, I'd rather the SC energy spent elsewhere.

  8. #8
    rowman's Avatar
    Join Date
    Dec 2014
    Location
    On a lenghty sabbatical
    Posts
    11,930
    Quote Originally Posted by MandoCoC View Post
    Agree with all of this. So then it becomes a cost/benefit discussion. SC has limited resources, as all companies do, so is this the best place for them to spend time on improvements? To do data analysis to come up with a predictor index vs other game improvements? Not IMO, I'd rather the SC energy spent elsewhere.
    Will add that I suspect that they did, at one time, look for such a predictor, and even honed it! I suspect their goal has always been to match clans such that both clans have an equal chance once matched, hence always a coin flip (50%/50%). The best means they derived was their weighting algorithm that they use. Now, we all have seen gross mismatches, so we know that is not the ideal means, but merely the "best to date found" means.

    Honestly, I (and I am sure others) hope they do find an accurate predictor, then all wars could be matched with a 50/50 probability, no more gross mismatches! I want them, or someone else using data, to find an accurate predictor, not to show the odds (as suggested in this thread) but to improve matching.

    Cheers!

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •