So ReddBoiler's thread prompted me to wonder whether we could try to measure the best ww players.

Obviously the checkmywwstats database has some results for wins/losses by role, but the sample size is pretty small for many players.

This is reminscent of the number of goals a football team scores a season being small. One suggested solution there is to count shots/shots-weighted-by-area-from=which-taken.

An analogue for ww might be 'correct'-lynches for the player's side - or even 'correct' votes for the player's side. Both of theses to be counted only whilst the player is alive, or, perhaps better, with a 50%/day decay rate after death. (Or maybe some decay rate depending on the day???)

I suppose turbos would be easiest to look at because the format repeats and there are fewer roles. But I know nothing about turbos.

I imagine at this point that there are somewhat decent stats on, eg, winning %s when the seer is lynched d1 versus when they aren't lynched d1 and so on. But that may be fantasy.

What sort of weightings would be appropriate? (Views welcome from turbo players, those who've played with the turbos-db or anyone else with an opinion. The aim would be to do/extend to longer games.)

If I were a Stern Judge I would weigh the stats only 35%~ or so. More of the decision need to come through subjective analysis.

Long games only, greater than 15 games played in both wolf and village roles

Gentlemen, the GOAT:

I have contemplated the concept of some sort of Win Probability Added thing for vanilla games (similar to baseball stats), since we should in fact have a decently large sample of "win % given a game state of X villagers/Y wolves/Z seers".

It's kinda messy though.

In my humble - and probably wrong - opinion, a villager's most important stat is the ratio at which they're mis-lynched vs night killed.

Someone with lots of time on their hands should compile that for me.

Yes. I'ld be interested in that kind of thing too. My nebulous idea, though, was to try to evaluate players based on their impact on the game whilst they are alive (and/or and shortly after death), rather than the proxy (for people who are habitually half-way-ruled) of what happens at the end of the game.

Chips hasnt been lynched as a villager in nearly sixty turbos i think.

Every game you've modded the wolves have won. So you can just go and sit in the corner with CDL for now.

Not true, village won my south park mash.

Rank	Name	Lynched as Villager %	NK %	NK/lynch
1	Chips Ahoy	0%	68%	undefined
2	traz	0%	69%	undefined
3	JimHalpert	3%	43%	14.79
4	rebonkulous	4%	57%	13.14
5	metsandfinsfan	10%	55%	5.55
6	Monkey Banana	11%	43%	4.01
7	wdcbooks	12%	46%	3.76
8	MDoranD	14%	50%	3.68
9	derm	11%	37%	3.50
10	Luckbox Inc	16%	55%	3.50
11	Swiitch	14%	48%	3.33
12	karnelchik	15%	45%	3.00
13	well named	16%	46%	2.92
14	atakdog	16%	42%	2.60
15	vixticator	19%	49%	2.58
16	captain binkles	22%	56%	2.57
17	soah	22%	56%	2.55
18	Zurvan	19%	48%	2.51
19	Montecore	13%	33%	2.50
20	OnThInIcE911	17%	42%	2.50
21	Larry Legend	15%	35%	2.33
22	CPHoya	20%	46%	2.33
23	bsball8806	16%	37%	2.33
24	shortline99	25%	56%	2.25
25	MrGatito	17%	38%	2.25
26	LKJ	24%	52%	2.20
27	aao	22%	46%	2.13
28	XXsooted	19%	40%	2.11
29	Andynan	17%	33%	1.99
30	Sun Tzu	20%	38%	1.94
31	VarianceMinefield	21%	41%	1.89
32	Felix the Cat	24%	43%	1.80
33	VoraciousReader	24%	43%	1.80
34	FCBLComish	28%	50%	1.80
35	Crossnerd	18%	32%	1.75
36	HerbieGRD	26%	44%	1.71
37	Aksdal	29%	48%	1.67
38	dmk	22%	35%	1.63
39	DonkDonkDonkDonk	27%	42%	1.57
40	derwipok	25%	39%	1.57
41	Nez477	24%	35%	1.50
42	ScottHoward	30%	44%	1.50
43	chuckleslovakian	23%	35%	1.50
44	Nicholasp27	28%	41%	1.47
45	younguns87	16%	21%	1.34
46	Forsythio	25%	33%	1.33
47	bhuber2010	18%	24%	1.33
48	fnord_too	29%	35%	1.20
49	ShipItUp	31%	38%	1.20
50	magic_gazz	26%	30%	1.16
51	UAW710	25%	28%	1.16
52	globetrotter	32%	35%	1.08
53	Birdman10687	37%	40%	1.08
54	hardcoreUFO	30%	30%	1.00
55	ZomgHax	26%	26%	1.00
56	McAvoy	33%	33%	1.00
57	BitchiBee	38%	38%	1.00
58	chrisrjdk	32%	32%	1.00
59	LuckayLuck	36%	35%	0.95
60	wahoopride	32%	27%	0.86
61	Anarchist	38%	31%	0.81
62	CalledDownLight	25%	20%	0.80
63	gtpitch	31%	24%	0.78
64	ElliotR	44%	32%	0.73
65	Duckburg	50%	36%	0.73
66	TimeLady	33%	24%	0.71
67	kyro	33%	24%	0.71
68	xxaceshigh77xx	30%	22%	0.71
69	Toganim	50%	35%	0.70
70	sixfour	39%	26%	0.67
71	J.D.	39%	25%	0.64
72	tappokone	44%	26%	0.60
73	Telcontar	33%	19%	0.58
74	Systolic	46%	27%	0.58
75	HiFi	40%	20%	0.50
76	Xkf	32%	16%	0.50
77	PlzBeALevel	42%	17%	0.40
78	Nofear3838	39%	15%	0.40
79	Titanrulez	47%	18%	0.38
80	Perdition	64%	23%	0.36
81	KruZe	42%	13%	0.30
82	chrja	63%	17%	0.27
83	clowntable	71%	14%	0.20

Ok or someone with 15 minutes on their hands.

Ty Aksdal

Pretty damn strong top 20.

if aksdal's list included me then yea pretty accurate

jk - but took it from here: http://checkmywwstats.com/pog/index....layerid=132620

(One thing with the db is it doesn't seem to say by whom people were night/day killed.)

Objectively measuring players based on those statistics (specifically) is still a fairly worthless indication of their strength as a player (even if you look at it from a polar perspective (i.e. just the strength of their villager game)).

For example,

A 0% lynch rate - while good for a villager to have, doesnt mean he's actually a good villager.

It doesnt take into account 1) that they're wolf game is generally poor 2) their mis-lynch percentage doesnt represent the accuracy of their reads/doesnt mean they're a good player at all.

A terrible player can clear themselves every game/never get lynched, and have a very high NK% because the wolves are essentially forced to kill them.

There a literally tons of players like this.

So yeah, they can clear themselves every single game/never get lynched...but does that mean they're any good at the game? Not really...they probably just have a really bad wolf game.

Take tappo for example...a decently high mis-lynch rate...and a relatively low NK%...both indicators of his above average wolf game showing through in his villager statistics (IN his case...this is not always the case for someone with those numbers).

His villager game is probably in the top 5% (and wolf in the top 1%).

Measuring player strength on objective variables alone is worthless without context.

Hence perception and subjective variables.

i'm pretty sure this is way off

i remember him being mislynched quite a bit the last 20 or so games he played or at least >2 times iirc

he was rightfully rustled i think

count me as one of those who thinks stats are pretty worthless. i think most of us know who the good players are when we play with them. it's somewhat difficult to be a really quiet and "sneakily good" player in this game because a big part is getting other people to follow you so if you don't really do that then your amazing reads don't matter that much.

He meant to say vanilla long games I think

boobs makes good points - which is why my suggested measures are of who gets voted (or maybe lynched) as a proxy for the player's effect on the game (whilst they are alive). Eg if you're a good wolf more villagers are likely to get voted for and/or lynched while you are alive.

oh yeah that would make more sense

pretty impossible not to get mislycnehd at some point in turbos

been lynched 0 out of 23 village games, and killed by the wolves in 16 out of the 23 (70%).

not counting the two times i was lynched in a neutral hunt, incidentally both times led by LKJ.

that's vanilla's only

having a low lynched as villager % is analogous to having a high win % at showdown. not necessarily indicative of high skill. dost thou seest why?

Like boober pointed out, mislynch rate isn't even close to a definitive metric for villager ability, let alone overall ability.

A ww bot that correctly claimed wolf/seer/villager and then voted at random every day would rock the mislynch vs nk stats but would only be as skilled/useful as the average player.

I mean there's obviously exceptions. But you can't deny that players that are mis-lynched rarely and nk'd often are usually assets to the village.

being easy to read doesn't say anything about your village game

it just makes you a bad wolf