2018 in stats.

[awan]

End of Year Dev team post.
Now that 2018 is over I feel like making an end of the year post.
I have shared this post with the dev team so they can add stuff.
But the original draft and most of the post has been done by me. (Awan)
You might wonder what we are going to talk about in this post.
It will be a recap on mostly last year and we will compare the stats we have gathered this year to the stats last year and share some (interesting) observations.

2018 in stats compared to earlier years.
So let me start off by saying that we do not share a lot of the data we collect.
The data we talk about will be mostly the win rate and player amounts.
I will compare 2018 to both 2017 and sometimes 2016.

Winrate
There is a bit of a catch with the win ratio.
Until December this year the official position of CM has always been a 75% Xeno favored win rate.
I think it comes as no surprise when I say that that figure is a fake.
Something also quite funny is that our win rate goes up and down a lot.
The graphs have a lot of spikes and in general do not go up slowly or consistently instead they spike a lot.
As much as 30% between weeks.

Player amounts
In general we track the amount of players at the end of a round.
This is because sudden spikes are not of that much interest to the dev team we want consistent numbers over multiple rounds.
In general player amounts is a lot less volatile than win rate. But it is not a clear line instead it is also best represented as a wave.
The three stats we track during a week seem to follow each other most of the time.
The only thing that really influences this is special events like the release of the tank and new xeno castes.

2016
In 2016 there were 4 weeks with the winrate being 75% +/- 5%.
In 2016 there were 8 weeks with a winrate being 75% +/- 10% (The 4 mentioned above included.)
There were 16 weeks were the marine win rate was above the Xeno win rate.
The total win rate for Xeno's was 55.27.
The total win rate for Marines was 44.73.
The best week to be a Xeno was 17.
The best week to be a marine was 21.
The highest measured end of the round pop was 136 in week 44.
We only started tracking it in week 43 though.
The highest measured end of the round low pop is 33 in week 49.
The average amount of en of round pop was 71.40.


2017
In 2017 there were 7 weeks with the winrate being 75% +/- 5%.
In 2017 there were 18 weeks with a winrate being 75% +/- 10% (The 7 mentioned above included.)
There were 5 weeks were the marine win rate was above the Xeno win rate.
The total win rate for Xeno's was 62.77.
THe total win rate for Marines was 37.23.
The best week to be a Xeno was 34 or 24 if you count draws as a xeno win.
The best week to be a marine was 50.
The highest measured end of the round pop was 213 in week 47.
The highest measured end of the round low pop is 80 in week 53.
The average amount of en of round pop was 101.78.


2018
In 2018 there were 9 weeks with the winrate being 75% +/- 5%.
In 2018 there were 14 weeks with the winrate being 75% +/- 10% (The 9 above are counted).
In 2018 there were 12 weeks with the winrate being 65% +/- 5%.
In 2018 there were 30 weeks with the winrate being 65% +/- 10%. (The 12 above are counted)
There were 11 weeks with a marine win rate above the Xeno win rate.
The total win rate for Xeno's was 60.00.
THe total win rate for Marines was 40.00.
The best week to be a Xeno was 23.
The best week to be a marine was 10.
The highest measured end of the round pop was 258 in week 22.
The highest measured end of the round low pop is 96 in week 32.
The average amount of en of round pop was 127.62.

Last but not least if there are any questions you have or if there is data you want to know feel free to ask here.

[KingKire]

Is there a way to have the data in raw format? So we could look for correlation?

Possibly with a few sample data points:
- Time start of round (Time zone/ Hour/ day/week)
- Time end of round
- Player count of round (Beginning of round being the more prevalent data point, but end of round count might also tell something)



I think we can use that data to make a nice graph, with the change logs being the points of interest of what is actually happening. (Plus maybe adding in a few other extraneous dates, like start/end of vacations, Big OOC CM13 drama points, etc)


That way we can average out the win rates against similar time-zones/ population sizes, as well as average it out among the dates of changelogs.

[awan]

Is there a way to have the data in raw format? So we could look for correlation?

Possibly with a few sample data points:
- Time start of round (Time zone/ Hour/ day/week)
- Time end of round
- Player count of round (Beginning of round being the more prevalent data point, but end of round count might also tell something)



I think we can use that data to make a nice graph, with the change logs being the points of interest of what is actually happening. (Plus maybe adding in a few other extraneous dates, like start/end of vacations, Big OOC CM13 drama points, etc)


That way we can average out the win rates against similar time-zones/ population sizes, as well as average it out among the dates of changelogs.

In general we have the start time of all rounds, their duration and thus also the end time.
I specificly do not want to give out the raw data.
The reason for this is that:
Wrong conclusions are quite easy to make.

Example:
There is a new xeno caste.
The win rate suddenly goes more xeno side.
Now it does not have to be the caste causing it.
It could be that because a lot of people wanna see the new caste and suddenly play xeno.
Or even worse it could quite well just be part of a natural spike. (As stated above the winrate fluctuates quite a bit.)

In general I do not have much to gain to show data over time and I feel it leads to endless discussions because of people using the data wrong in arguments.
Correlation is not causation but that is one of the easiest traps that people fall for.

[spookydonut]

As a correction, after I became head dev there was no official winrate position up until you guys made a new position on it. My position was 'winrates are dumb, fun is more important'

Even before I was head dev the 75% winrate was refuted by bmc.

[Heraclitus]

I have always wanted to get a peek at the CM statistics, so thank you very much for posting what you have so far. I think you can release statistics like these without negatively influencing the game in anyway, it can only inform players and that's always a positive (as long as it doesn't lead to any powergaming).

Being able to say that there has never been a week with a 90% winrate is nice too.

With that in mind, can you give a break down of round lengths, low, high and average pops, and winrates for each map?

[awan]

As a correction, after I became head dev there was no official winrate position up until you guys made a new position on it. My position was 'winrates are dumb, fun is more important'

Even before I was head dev the 75% winrate was refuted by bmc.

Maybe yes, the 75% winrate comes from a post made by apop.
https://cm-ss13.com/old/viewtopic.ph...=7902&p=83937&
Bmc can refute that and you can also say that was not your personal opinion.
I did not really see comments by you on the win rate when I did my quick search.
In my quick search I did not even see you public post anything about the winrate on the forum and this was unknown to me.

So far almost always when I saw it mentioned it was mentioned as 75% xeno favored.
And that goal was never what we had even during high pop in 2016 when the post was made.

Our 65% +/- 5% did not take into account the current winrate of that time and is a compromise.
It is also a goal but not a hard number. It has a philosophy just like you have one behind yours.
So apologies for the misinformation.

[awan]

I have always wanted to get a peek at the CM statistics, so thank you very much for posting what you have so far. I think you can release statistics like these without negatively influencing the game in anyway, it can only inform players and that's always a positive (as long as it doesn't lead to any powergaming).

Being able to say that there has never been a week with a 90% winrate is nice too.

With that in mind, can you give a break down of round lengths, low, high and average pops, and winrates for each map?

There has never been a week with a 90%+ winrate.
There has been 9 weeks in the last 3 years with a 75% or above winrate.
4 in 2018 4 in 2017 and 1 in 2016 If we count draws as xeno wins without that it is 5.
4 of these were above 80% If we count draws as xeno wins. If we do not do that it is 1.
The highest ever was 85.7% in a week. (That is with draws counted as a xeno win.)
Without counting a draw as a xeno win it is 85.2%

[Vispainius]

Really interesting data, thanks for showing us some of it Awan.