Newbie Data

[Faraday]

(obvious joke - I'd never pretend to be from Australia)

[Magister Ludi]

Whoa there mate

[mykonian]

Town are lynching scum D1 0.224489796 of the time while random D1 scum lynches would happen 0.222222222 of the time.

That's probably not good news.

if you compare with rolling a die and lynching that player, then the result seems bad.

If you compare with everybody rolling a die and voting that player, but scum not voting for each other, then the result already looks a bit better.

If you compare with scum preferring to vote together, but town rolling their die again, uhm: even cooler!

[Quilford]

(obvious joke - I'd never pretend to be from Australia)

I will quote hilariously embarrassing posts from you in the near past

[Psyche]

If you send me the spreadsheet or link me to it I can run a few tests to determine if there's actual statistical significance in the data you've collected Tierce.

There you go. I'm still working on it and there are a few labels missing, so if you have any difficulty understanding what a section means, let me know.

<<<333

Fantastic stuff.

[Isa]

I'd be willing to help in updating the spreadsheet with recent results if you want to, Tierce.

[Tierce]

Thanks, but I don't have any idea of continuing to do it at a frantic pace, especially since there are sections that always need manual updating (or better formulas that I can't be bothered to deal with). My idea was to wait until at least Newbie #1260 (and all previous games) is finished and then revamp that info.

I have 'game length in game days' info to post, but BBcode is annoying. Will do it soonish.

[Quilford]

Updating with the latest data. (Only updating Ia. though because it's the only interesting table in my opinion. >__>)

In this study were included all the finished 2of4 newbie games with available, complete data from Newbie 1102 (the start of 2of4 with Rolecop/Goon mafia team variant) to Newbie 1300, with a few exceptions: Newbie 1240, Newbie 1266, Newbie 1267, because I had no clue what the role distribution was; Newbie 1289, because player cheating influenced the gamestate; Newbie 1292, because the game is ongoing; whatever games Tierce discarded during her original data collection.

I. Overall Newbie Data per 2of4 Setup Variants

Ia. Wins per Variant

# Games	% Games	Setup	Mafia Wins	Mafia Win %	Town Wins	Town Win %
35	23.03	Cop/Doctor	13	37.14	22	62.86
22	14.47	Cop/Jailkeeper	9	40.91	13	59.09
28	18.42	Doctor/Jailkeeper	12	42.86	16	57.14
26	17.11	Cop/Townie	13	50.00	13	50.00
16	10.53	Doctor/Townie	12	75.00	4	25.00
25	16.45	Jailkeeper/Townie	11	44.00	14	56.00
152			70	46.05	82	53.95

Note:

If each of the six setups was present exactly 1/6 of the time, the above data can be corrected to:

• 48.32% mafia victory
  51.68% town victory

[Quilford]

For comparison:

# Games	% Games	Setup	Mafia Wins	Mafia Win %	Town Wins	Town Win %
22	22.45	Cop/Doctor	7	38.89	11	61.11
14	14.29	Cop/Jailkeeper	8	57.14	6	42.86
17	17.35	Doctor/Jailkeeper	5	29.41	12	70.59
16	16.33	Cop/Townie	6	37.5	10	62.5
11	11.22	Doctor/Townie	8	72.73	3	27.27
18	18.37	Jailkeeper/Townie	8	44.44	10	55.56
98			42	42.86	56	57.14

Note:

If each of the six setups was present exactly 1/6 of the time, the above data can be corrected to:

• 46.69% mafia victory
  53.31% town victory

--

Some big changes in the winrates of the Cop/JK, Doc/JK, and Cop/Townie setups.

[Mr. Flay]

Is singersigner assigning setups to mods? Because ugh those Doc/Townie percentages are off.

[Vi]

Is singersigner assigning setups to mods? Because ugh those Doc/Townie percentages are off.

To the point that it's statistically significant.

Ditto Cop/Doc.

[Cheery Dog]

Is singersigner assigning setups to mods? Because ugh those Doc/Townie percentages are off.

To the point that it's statistically significant.

Ditto Cop/Doc.

Looking at those two stats, it looks like there's been about 10 games given a cop they shouldn't have.

[Mr. Flay]

I'll ping her.

[Quilford]

Well, you know, I think now might just be the time to step away from 2of4 and pick a different setup(s) for Newbies.

[Xalxe]

Is singersigner assigning setups to mods? Because ugh those Doc/Townie percentages are off.

Nope.

[Mr. Flay]

Ugh, bet we lost that in the transition from VRK.

Quilford: Not a bad idea, but that's probably a different thread (I know we had one going for that a while ago, but it might have been eaten by the Rollback).

[Xalxe]

Ugh, bet we lost that in the transition from VRK.

She's supposed to? I never got an assignment from VRK IIRC. But I may not RC.

[Leafsnail]

Maybe some mods have been discarding it since it's clearly the least balanced setup.

[zoraster]

VRK never assigned setups did he?

[Mr. Flay]

VRK had started doing so after NG mods ignored the third or fourth warning from On High to randomize their setups properly (this has ALWAYS been a problem...)

Or at least he said he would do so. *shrug*

[Leafsnail]

The problem would probably go away if the setups were more equal. As it is if you run a doc/townie setup you're badly disadvantaging the town.

[Quilford]

Quadz and I have come up with the following refinement of 2of3.

Town gets 2 of {Cop, Jailkeeper, Vanilla Townie} + 5 VTs.
If Town has 1 PR, then Scum gets 1 of {Rolecop, 1-Shot Strongman}; if Town has both PRs, then Scum also gets both PRs.

This is pretty much the way to go for scum power (in that scum power should increase with town power).

I really think it couldn't hurt to at least modify the current 2of4 to give Mafia a One-Shot Strongman if there are two Town Power Roles in the setup.

[Cheery Dog]

But then scum will also know if they are able to fakeclaim the second power role if the one that has been had flips.

[Quilford]

And?

[Mr. Flay]

Setup Speculation, ahoy!
http://forum.mafiascum.net/viewtopic.php?f=5&t=24479