Random Race Test

[Titan]

So I've been running the Starcraft random algorithm for choosing a race (when a user selects "Random" in the game lobby). After about 1 1/2 hours I have some interesting results.

Race            Times Picked        Probability
--------        ------------        -----------
Zerg:           10,941,256          0.333217035
Terran:         10,941,658          0.333229278
Protoss:        10,952,310          0.333553686
--------        ------------        -----------
Total:          32,835,224

It looks like Protoss is in the lead right now. I'm going to run the test all night and see what happens. No, there is no grand purpose to this. I was just looking for something interesting to do. It will probably just continue to converge to 1/3 each, but we'll see.

After 331,975,848 tests, Protoss lost a lot of ground, but the probabilities are still converging to 1/3.

Race            Times Picked        Probability
--------        ------------        -----------
Zerg:           110,668,607         0.333363429
Terran:         110,669,002         0.333364619
Protoss:        110,638,239         0.333271952
--------        ------------        -----------
Total:          331,975,848

[hure]

its more interessting if you have more randoms.

if you play racewar with 6 players, it will never have more than 3 players of the same type.


did you also analyze streak lengths?

[Titan]

its more interessting if you have more randoms.

if you play racewar with 6 players, it will never have more than 3 players of the same type.


did you also analyze streak lengths?

This is just testing a scenario where one player goes random. I haven't done any deep analysis... I was just bored last night.

[John]

I would be bored to doing that. What's the point?

[Otaku]

Curiosity?

[John]

Hell yes!

[Polie13]

That's pretty neat, any chance of running it against multiple race selections?

[Titan]

Here is something interesting if you weren't amazed by those results. It seems that...

If you play a game with 8 people, 7 of them choose Protoss as their race, and you go random... you will still have a 1/3 chance of getting Protoss.

Multiple randoms is a different story though.

[John]

Dude that owns!

[BD-ICY]

I've played a game where all 8 of us went random, and 4 of us were terran, 3 were toss and only 1 zerg(he got pwned by the toss lol)

[MiCrOz]

I would be bored to doing that. What's the point?

He's not doing it physically.

The results are interesting, but expected.

[dt]

This gets interesting in case of multiple random players. It's not entirely random any more, the following rules apply:

2 players: the probability of a mirror matchup (same race twice) is only 1/8, as opposed to 1/3 as you would expect from a random distribution. All other matchups are equally likely.

More players: All matchups are equally likely, however only the following ratios are possible:
3: 1-1-1 or 2-1-0
4: 2-1-1
5: 2-2-1
6: 2-2-2 or 3-2-1
7: 3-2-2
8: 3-3-2