League stats: week 1

[gentimouton]

I'm guessing this applies to a lot of people, but I've been being demoted/promoted between silver 1 (1200-1400 trophies) and silver 2 (1000-1200).
What may not apply to everyone is my 97 attacks won in one week! Which likely means between 120 and 140 attacks total (when I lose, it's 25-30 trophies, but when I win, it's (purposefully) rarely 3 star - I try to maintain around 1200 trophies). So that's 20 attacks per day on average.

League data!

But enough about me. What about the other people in my (current Silver1) league? I took all their numbers and plotted them together. Here's the distribution.


This is a cumulative distribution graph. (http://en.wikipedia.org/wiki/Cumulat...ution_function) What's interesting about it?
- The top 10% of attackers won more than 29 attacks in one week (this is the top horizontal blue line)
- Half of the players won less than 11 attacks (and half won more than 11 attacks) (this is the middle horizontal blue line), which means less than 2 per day.
- 13% of players won 0 attacks this week (this is the bottom-left corner). Does this mean 13% of players were not playing this week?
- I won most attacks (97 attacks, this is the top-right corner). OH YEAH

Not on the graph, but interesting nonetheless:
- The average player wins 14 attacks in one week. 2 per day.
- Only 93 players in my league, why not 100? Keeping room for demoted/promoted players from the gold3 and silver2 leagues?
- 2500 gold per attack won * 97 attacks won = 240k gold in one week. I hope Jorge managed to attack at least 2.4 times this week, otherwise he should drop down to Silver 1

Want your league plotted?

I'd be happy to plot your league data, or add your league data to the current graph. If you want to give me data about your league:
- Check that I'm not in your league (my in-game name is gentimouton), otherwise we would have redundant data, and that's bad. (I also know that two people could be in the same league and unknowingly submit the same data to me. Let's hope there are enough leagues and few forum users sending their data =) )
- Write down the number of days and hours left until the end of the season. This is important because it makes sure we are not comparing data from Monday to data from Friday (which should be very different). If there are 6 days and 21 hours left, write down on the first line: 6,21
- Number of attacks: let's say the first person in your league won 18 attacks, the second won 9, and so on. Then write down on the second line: 18,9,0,32,13,...
- Send this list of numbers + days/hours to me by private forum message (cick on your profile at the top right corner of the page, then look for "Send Private Message", and enter my name as a recipient). Please send a private message so as to not flood this thread with numbers!


If you want the script I used: http://pastebin.com/pW3qiN5L
This is a script for R, a free stats software: http://www.r-project.org/


Any thoughts about these numbers?

[mgd6912]

you give me hope for the younger generation... I love stats and what you have done...

[Junliang]

I like maths.. but don't like stats.
God, those mods *nightmare*

Anyway, regarding your data, sample size of 93 players.

Would be interesting if supercell would collate the data for all players.
Chances are, attacks won would be greater than defense won.
But there are factors like farmers, who place their TH outside, but would be a nice guage of whether "Offense" is greater than "Defense" in most cases.

Then supercell can tweak the balance accordingly.

But in recent updates, seems like they prefer to shift the balance between offense and defense..

But in this very last patch, seems like they strike a balance between offense and defense..
But yeah, numbers don't lie

[Ephilly]

Nice job, always nice to read some stats on this game...


Curious about your attack strategy... You attacked 97 times and stole about 240k loot? Were you simply raiding the first person you searched on? -- If not, would be curious to see your ratio for elixir expended per attack/elixir won per attack and suggest attacking villages w/ 100+ loot in each resource.

Edit: Believe you are referring to the bonus , now it makes sense... Nice work.

[Thecageyone]

118 here, with 10 defences won ... and still can't get to 1400

A lot of those were 5-6 trophy wins for 50%, whereas when I lose, it's on bigger, badder bases and I lose anything from 20 - 36 trophies on those. Takes a good few of the small wins to make back up for them

[gentimouton]

I like maths.. but don't like stats.
God, those mods *nightmare*

Anyway, regarding your data, sample size of 93 players.

Would be interesting if supercell would collate the data for all players.
Chances are, attacks won would be greater than defense won.
But there are factors like farmers, who place their TH outside, but would be a nice guage of whether "Offense" is greater than "Defense" in most cases.

Then supercell can tweak the balance accordingly.

But in recent updates, seems like they prefer to shift the balance between offense and defense..

But in this very last patch, seems like they strike a balance between offense and defense..
But yeah, numbers don't lie

Yes, I did not track the defenses won. In my opinion, it does not mean much when people are farming. It may make sense above 2000 trophies though.

[2fingaz2727]

Cool stats, thanks!

[BaMiao]

This is really cool. If I had time, I'd join in on data recording. I think the league system has a lot of potential to give information on the general game ecosystem.

I think you should track defenses won, too. With large samples, you can use it to gauge "battles lost", at least on a general scale. Perhaps you can see where the balance is between winning/losing. Things like farming/TH outside the walls will skew this, but it might be interesting nonetheless.

Also, there's the possibility of actually seeing what sort of bases are populating each trophy range. Right now, it would be too annoying, but if we could get supercell to restore the level badges, we could quickly record the base levels in every league.

Actually, now that I think about it, this could be the reason supercell removed the level badges in the first place. It would probably be too easy to find suitable ranges for farming.

[hb808]

Nice graph.

Now if you could only discern which ones were one, two, or three star wins; how many trophies were won, and check for confounding with people who put their TH outside the walls for quick shields.

[gentimouton]

I am surprised to not receive more feedback on these numbers:
- the average player wins 14 attacks per week. Do you win more than that?
- the top 10% of players win more than 29 attacks per week. Are you in the top 10%? Do you consider this number too high?