Defining a Classic war mismatch

[Munitor]

Interesting. Will look at the logs later.

Have you considered my suggestion of including the higher troops for better explanatory power of matches?

Also that last one looks weird. How can there be a weight difference of almost a max TH10? What's causing that?

[joshsgrandad]

View the log here: https://cocp.it/clan/UL0Y9G9Y

vs Red Waters, we were 2/1/5/1/1 they were 1/2/5/0/1/1. Heroes: we held 4.3% advantage at queen and a 60% advantage at RC, while they held a 3.6% advantage at king and a 1% advantage at Warden. We had a 5.24% weight advantage.
End Result - we had a decent advantage overall, and beat them by 5 stars.

vs beast towers, we were 3/1/3/3/3/1/1, they were 3/1/4/3/2/2. Heroes: we held queen, warden and RC advantage by 4.4%, 1% and 30% respectively, they held king advantage by 3.1%. They had a 3% weight advantage.
End Result - we again had a significant RC advantage, but I would call this a more even war than the one vs Red Waters. We won 44-42.

vs Ataraxis II, we were 2/1/3/3/4/1/0/0/0/1 and they were 2/1/4/2/6. Heroes: we held GW and RC advantage by 4.8% and 62.5% respectively. They held king and queen advantage by 9.3% and 3.2% respectively. They also held a 7.1% weight advantage.
End Result - we won 45-43. This was a weird one as their top bases did not match well with ours at all. At the top 2, we held a 10K weight advantage, even though overall they held an 89K advantage. I assume the MM felt that the offset of the defense weight lower in the lineup made up for this. I believe we had a distinct advantage in this match, we just had to do a little more dipping than normal to get everything covered.


Only thing I can glean from this so far is that I believe the RC is underweighted comparatively speaking. Near as I can tell everyone should be maxing RC ASAP at 13.

Or offensive weight is adding a lot of war weight to the match too..
Camp space, spell slots, troop levels, spell levels, cc capacity, as well as heroes make up part of the matchmaking procedure..
If one cant make sense of matches with the things we can count, then it those hidden figures and the amount of distortion the algorithm is adding that would add up.

Go see Sam, he can give some figures that add up "under the hood".

[Thegreatpuma]

Interesting. Will look at the logs later.

Have you considered my suggestion of including the higher troops for better explanatory power of matches?

Also that last one looks weird. How can there be a weight difference of almost a max TH10? What's causing that?

I have, I just need to come up with a better way to input the troop levels...I.e. I need to learn the API. Doing it by hand is error prone and I lost interest after the third war.

Or offensive weight is adding a lot of war weight to the match too..

Camp space, spell slots, troop levels, spell levels, cc capacity, as well as heroes make up part of the matchmaking procedure..
If one cant make sense of matches with the things we can count, then it those hidden figures and the amount of distortion the algorithm is adding that would add up.

Go see Sam, he can give some figures that add up "under the hood".

Completely forgot about camp space and CC. Ugh this is so much work...not sure how those guys do it to be honest.

[Munitor]

I have, I just need to come up with a better way to input the troop levels...I.e. I need to learn the API. Doing it by hand is error prone and I lost interest after the third war.

Haha, bet you're looking forward to harvesting CWL data then!

On topic though, I have seriously considered doing a 10-20 war project, where I use all of the offensive levels, as well as the gold weights, to try to estimate the offensive weights (the difference in gold weights must necessarily be explainable by a difference in offense). But yea, we'll see if I can find the time. Will let you know if I do.

[samratulangi]

Or offensive weight is adding a lot of war weight to the match too..
Camp space, spell slots, troop levels, spell levels, cc capacity, as well as heroes make up part of the matchmaking procedure..
If one cant make sense of matches with the things we can count, then it those hidden figures and the amount of distortion the algorithm is adding that would add up.

Go see Sam, he can give some figures that add up "under the hood".

I have, I just need to come up with a better way to input the troop levels...I.e. I need to learn the API. Doing it by hand is error prone and I lost interest after the third war.



Completely forgot about camp space and CC. Ugh this is so much work...not sure how those guys do it to be honest.

Long time no post, I think the biggest problem here is u only look at Hero and def weight, while camps, troops lvl(unlocking troops gain more weight then leveling up) and spells, so I believe u need a lot more data to make these anecdotal matches give more proper view.

Imagine you have to select a base with low lvl heroes but max troops and max lvl heroes but low lvl troops to attack a mid th13, which one will give you better result without using power potion?

[Thegreatpuma]

Long time no post, I think the biggest problem here is u only look at Hero and def weight, while camps, troops lvl(unlocking troops gain more weight then leveling up) and spells, so I believe u need a lot more data to make these anecdotal matches give more proper view.

Imagine you have to select a base with low lvl heroes but max troops and max lvl heroes but low lvl troops to attack a mid th13, which one will give you better result without using power potion?

for the first two wars while I was doing this I totaled all troop levels, and that told me nothing. If I get to the point where I am pulling all this from the API, I could certainly total the number of accounts that have unlocked specific troops to add to the comparison. Like I said though I need a better way to get this data as hand typing it in is just not going to happen.

[samratulangi]

for the first two wars while I was doing this I totaled all troop levels, and that told me nothing. If I get to the point where I am pulling all this from the API, I could certainly total the number of accounts that have unlocked specific troops to add to the comparison. Like I said though I need a better way to get this data as hand typing it in is just not going to happen.

certainly u can get all of the offense u can see from player profile using CoC API, but not on def weight and camps lvl

[Thegreatpuma]

certainly u can get all of the offense u can see from player profile using CoC API, but not on def weight and camps lvl

Yeah, I can get this manually - it's tedious, but when added to all the other stuff it's just too much. getting troop levels manually from clash of stats takes the longest of any part of this.

[samratulangi]

U can make a code in PHP or some other form and spend a few hours of coding and all will be done to export the war data to a CSV files automatically every wars, and the only data u need to add manually is camp space and def weight, then u should be able to make a better analysis to your wars, by doing proper calculation with more complete data, using just some data like heroes lvl and def weight while omitting other significant parameters certainly will make your analysis go wrong way