Some MP maps ARE more likely than others!

[[IPC] timor ultima]

So a while back I was reading a thread and it was being discussed whether or not certain maps are more likely to get rolled than others. No one seemed to have any data on the subject, so I set to secure some. I collected 114 map observations and I came up with some interesting results.

Methodology: In the pre-game lobby I made note of which map was being offered as "Next." In addition, if "Random" received the most votes I made note of which random map the game gave the lobby. To be clear this is not a study of which maps get played the most, because that isn't an interesting question, it is a study of the maps that the system offers the players.

All-in-all I collected 114 total map offering observations while playing Core Team Deathmatch. 86 of these were offered as "Next" and 28 were offered as "Random" (once Random received the most votes in the pre-game lobby)

UPDATES:
01/27/11 - Some users have been reporting they see "uncommon" maps rolled more often in their own experiences. I have added a poll to see if these experiences are unique (which is why I would be hearing about them) or if there may be something else going on.

Summary of Findings - Last Updated 01/25/11
System's Top 5 Favorite Maps to Offer
1. Summit (14.0% of offerings)
2. Array (9.6%)
3. Grid (8.8%)
4. Jungle (8.8%)
5. Firing Range (7.9%)

System's Top 5 LEAST Favorite Maps to Offer
14. Hanoi (2.6% of offerings)
13. WMD (4.4%)
12. Nuketown (4.4%)
11. Havana (5.3%)
10. Cracked (5.3%)

Now I am not going to go into the statics of the matter because that's boring, but the jist of the data I collected is the following:

The system does prefer certain maps over others. This preference does not appear to be random. There is a statistically significant chance that this variation in map preference is built into the system and is not merely an observance of a chance event.

Below are bar charts detailing the prevalence of the various (pre-DLC) maps. The image may be resized down based on your browser or screen resolution, causing the map names to not be legible. Click the image to see the full-sized version.



In addition, I did a study of how the system responds to ties between the "Next," "Previous," and "Random" map selections in the pre-game. Now, admittedly, ties between categories are somewhat uncommon so my data set is limited. I will post again when I have even more date for category ties. Currently the resulting "implied odds" look like this:




Again, the data I have on each scenario is limited at this point so I will repost once I have a greater set of data for each instance. "Odds of X '(type)' wins with Above Odds" was determined using corresponding probabilities and a cumulative binomial probability model.

If anyone has any questions regarding methodology I would be happy to discuss. Beyond that I hope everyone finds this information helpful. I will continue posting updates as I get more data.

[idkmanidk]

whoah....how long did that take you to figure out?

[[IPC] timor ultima]

It took maybe 80 or 90 games/pre-game lobbies to get the data. Beyond that getting it organized and analyzed took prolly an hour. Then making it look pretty enough to post took less than an hour. All for the cause! Hope it was helpful!

[[DOGS] PitifulGrunt]

Wow, impressive! Are you working on your thesis? One of my buddies is a stats guy in pharma. Crazy stuff.

So, just to throw this out there... maybe the developers or whoever is controlling things is steering things towards maps that are more "server friendly" than others. Just a thought.

Thanks for the analysis. Now we know which maps to practice on.

[[IPC] timor ultima]

No no thesis for me, just curiousity. Your theory on 'server-friendly" maps is interesting. I don't know a ton about how the servers for this game work, so I don't know what would really make a map more "server friendly." The only thing I can think of is that smaller is better, though if that were the case I am surprised to see Array near the top of the list since it is one of biggest if the not the actual largest MP map in the game. Practice up on those system favorites, odds are you'll be playing on them quite a bit!

I am going to start a new data set once DLC drops. I figure Treyarch is going to reweight the games once the new maps are live. I will have exclusive stats just pertaining to the DLC maps as well, so stay tuned!

[[DOGS] PitifulGrunt]

Originally Posted by recquiescatpace

I am going to start a new data set once DLC drops. I figure Treyarch is going to reweight the games once the new maps are live. I will have exclusive stats just pertaining to the DLC maps as well, so stay tuned!

Ugh! That sounds like a beeyotch! You'll have to take into account people that don't purchase the DLC. Maybe I'm wrong but that sounds HARD. Thanks for the efforts.

[[IPC] NuckFuggets]

Wow nice stats! Sharing this.

I do think you should keep compiling data because 100 or so offerings isn't enough to say for sure. If you had 1,000 or so and this was the result it would be more conclusive i think.

[[DOGS] PitifulGrunt]

Originally Posted by NuckFuggets

Wow nice stats! Sharing this.

I do think you should keep compiling data because 100 or so offerings isn't enough to say for sure. If you had 1,000 or so and this was the result it would be more conclusive i think.

Actually 100 is more than enough. If he did 1000 then he would just have 10 samples of 100 and the signifigant trends in each sample would be the same. He would only have 10 of them. The confidence might sharpen a tad but that's all. the same trends would be apparent, probably after even only 30. This is pretty reliable stuff actually.

[[413] chronicoffender]

wow nice work man im a little surprised that firing range and nuketown wernt picked more it seems like there always in the next box

[Blak Smyth]

The wierd thing that doesn't show in your data is that whenever I experience a tie vote the winner always seems to go to the top option. Previous over next, next over random. I have been watching this regularly and it has held true everytime except once this last week, and me and my friends where like what happened?

It may be different for different user as well. I know I almost never see nuketown as an option anymore, but some people still do. I also think I see cracked, Hanoi, and Crisis far more than summit. My opinion is speculative since I have not documented any of this. Maybe 100 games is not enough to determine a pattern here? I am impressed with the time and energy you have put into this though, hats off. I just believe I am experiencing different results. Possible variables: Rank?, connection?, First time selecting next compared two consecutive next selections, what else? Does party size play a factor?

I have also seen a map with less votes win a selection. Does anybody ever see it not count your vote sometimes?