An amusing alternative. Hmm.

Right you are Ken VMf!

Wolf pt 3:

Previous rankings:
http://forumserver.twoplustwo.com/sh...&postcount=963
http://forumserver.twoplustwo.com/sh...&postcount=965

Here is the last and least correllated with wolf success (I believe since hunting the seer is sometimes more art than science and is also highly dependent on opposing vanilla villagers.). First I'll do the average day in the game the seer was night killed in games that the seer was night killed before the player was alive. This uses atak's (deathday - 1)/(n-1/2) formula, thanks again.

Player	Average day seer night killed when alive * % games killed seer before dying
UAW710	0.077777777700
chrisrjdk	0.130062530030
gtpitch	0.156250000000
Nofear3838	0.162250320466
Montecore	0.167550043086
sixfour	0.172857340327
Sun Tzu	0.175900513135
Anarchist	0.182186255397
Telcontar	0.187805901357
metsandfinsfan	0.195547395176
well named	0.199200517263
Hitch	0.212548100009
atakdog	0.214766615024
VarianceMinefield	0.224357484276
TimeLady	0.224471050080
captain binkles	0.234838109048
McAvoy	0.235998089815
Systolic	0.256594158236
bsball8806	0.259032753055
Titanrulez	0.261715395993
LuckayLuck	0.267313265552
JimHalpert	0.267489614042
XXsooted	0.271430911392
CPHoya	0.280156128119
norseman1066	0.280586572721
PlzBeALevel	0.304545454545
ElliotR	0.304892126106
Nez477	0.305008043048
OnThInIcE911	0.310207423803
chuckleslovakian	0.312469723549

Next since games where they didn't find the seer before they died matter as well I now take the percentage of wolf games they played where they didn't find the seer, multiplied by how long they lasted that game (aka if they died lynched d1, the multiplication is 0 as opposed to living with the seer to f3, which would be .8).

Player	Games didn't find seer * Average day died
Hitch	0.027777777750
UAW710	0.062169312107
gtpitch	0.089707074586
chrisrjdk	0.114257513735
soah	0.123523002028
Titanrulez	0.126171529582
captain binkles	0.147245185429
Anarchist	0.166119479905
CPHoya	0.166531782982
Montecore	0.175545439142
Nez477	0.182040644476
PlzBeALevel	0.186601337259
Nofear3838	0.199129129093
metsandfinsfan	0.206640872841
ZomgHax	0.213415966133
bsball8806	0.220677623384
Telcontar	0.233610615690
JimHalpert	0.245771622336
HiFi	0.251669380800
OnThInIcE911	0.255652487963
well named	0.264977808694
McAvoy	0.275102245392
Sun Tzu	0.301224085498
ElliotR	0.308681790013
VarianceMinefield	0.310206006559
atakdog	0.346181494485
LuckayLuck	0.350543260161
norseman1066	0.389661835749
TimeLady	0.402392267346
CalledDownLight	0.415535199601

I now take the first table and add it to the second table to get this:
average day seer killed before dying* %games seer killed before dying + average day died as wolf * % games seer still was living when dying

Player	Combined Ranking
UAW710	0.139947089807
Hitch	0.240325877759
chrisrjdk	0.244320043765
gtpitch	0.245957074586
Montecore	0.343095482228
Anarchist	0.348305735302
Nofear3838	0.361379449559
captain binkles	0.382083294477
Titanrulez	0.387886925575
metsandfinsfan	0.402188268017
Telcontar	0.421416517047
CPHoya	0.446687911101
well named	0.464178325957
Sun Tzu	0.477124598632
bsball8806	0.479710376438
Nez477	0.487048687523
PlzBeALevel	0.491146791805
McAvoy	0.511100335206
JimHalpert	0.513261236378
soah	0.534353893832
VarianceMinefield	0.534563490835
ZomgHax	0.541808271292
atakdog	0.560948109509
OnThInIcE911	0.565859911766
HiFi	0.573365351496
ElliotR	0.613573916119
LuckayLuck	0.617856525713
sixfour	0.623224654964
TimeLady	0.626863317426
norseman1066	0.670248408470

Let me know feedback so I can edit this around if anything seems off. One thing I notice right away, is that there is a brutal punishment for not nk'ing the seer before you die.

Iversonian's alternative offers too much of a leveling game for me.

Beyond that, I don't see why any village would ever accept a draw at f4. Because no village would accept a draw, if there are no clear villagers, it's in the wolf's best interest to no-kill and force the villagers to lynch at a 25% chance instead of a 33% chance.

I'm not on any of these tables. Checks out.

Why am I missing from the first table, but on the summary?

You have only 5 wolf vanilla/slow games. I cut off at 8.

I will have full rankings for on the site later - you'll be able to sort and cut off to your heart's content.

wrong cutoff - I just fixed it. Thanks!

The second table favors wolves who die early. I get that things are mostly out of the player's hands once he's dead, but you might want to look for a different way to normalize it. Off the top of my head I got nothin', but I'll think about it.

Boy, I've always said I suck at seerhunting. And I've been right.

yeah I caught that. I couldn't think of a better way of doing it either. There has to be some way though.

1 vanilla wolf game. Found the seer n1.

I AM THE GREATEST

still don't understand what those tables mean hurpiesdurpies..

One way would be to ignore such games entirely. Another would be to give wolves "credit" for seers caught after they're dead. It's possible that neither is as bad as it sounds.

In vanillas, yes -- in V+ and mishmashes there could be peculiar but valid reasons not to (tripwires, alternate-night wolf tracker/watcher, other weird stuff that normally doesn't apply but I'm sure you can think of it).

I'd say requiring seer peek (/randing among everyone who hasn't previously been peeked) and requiring NK (/randing among non-wolves) in vanillas is a good idea. Once you get past straight vanilla it probably depends on the roleset but I don't think it's wise to have a default of "you must".

Yeah, just tell me when the statistics look good for me, so I can care

and tell me when they look bad, so I can say statistics are stupid.

Yah we can ignore those games. It's just there won't be a penalty for living until d7, getting lynched and the seer dying n7.

atak:

Here is the fixed version of the original table (I edited to ignore partial seers and forgot to account for multiple seers in the original version):

Player	Average day seer night killed when alive * % games killed seer before dying
UAW710	0.004895105
sixfour	0.005647426
Nofear3838	0.006265664
Hitch	0.006308019
TimeLady	0.007328605
chrisrjdk	0.008497434
gtpitch	0.009469697
Sun Tzu	0.00947638
Telcontar	0.009483889
Montecore	0.009783642
Anarchist	0.009996158
ElliotR	0.010074701
well named	0.010370624
atakdog	0.01075526
metsandfinsfan	0.010888691
XXsooted	0.010917345
bsball8806	0.011372154
HiFi	0.011771356
norseman1066	0.011866852
captain binkles	0.012236631
chuckleslovakian	0.012296705
VarianceMinefield	0.012520003
Systolic	0.012931057
McAvoy	0.013011415
PlzBeALevel	0.013015436
JimHalpert	0.013096951
LuckayLuck	0.013146193
Nez477	0.013588722
Titanrulez	0.014142704
Nicholasp27	0.015181724
FCBLComish	0.015319729
OnThInIcE911	0.015521324
CalledDownLight	0.015945438
CPHoya	0.016296906
VoraciousReader	0.018216568
ZomgHax	0.018457405
wahoopride	0.019382279
soah	0.021492111
dmk	0.022130755
HerbieGRD	0.029876957

This table looks like it has a stronggggg association with dying early. If you die early every time the one time you bink the seer n1 you have a great score!

Ex. Krayz gets lynched d1 every game for 7 games. On his 8th game he gets lynched d2 and binks the seer n1. He now has a perfect score as a seer hunter!

you are gonna really need to help us interpret this data better...

like I know I am a a bad seer hunter, so not sure what that is doing here

It's just rankings, final numbers have no specific relevance. The key is looking at the table headers to see the formula used.

the higher you are to the top, the better you are then by the data, on every table?

Correct. Follow the formulas and it should be clear whether lower or higher is better!

I like the tables where I'm near the top. Those are obviously the stats that matter most

This is pretty much my thoughts on the matter.

---

I will almost always rand if a seer hasn't submitted a peek - once. Any more than that and the skeptical face appears and PMs go down. But it's really rare that a PR doesn't submit, especially a seer.

Other than that, I allow people not to submit (though it is always nice to get told that they're choosing not to use their action so I don't freak out and sift through 100 PMs to make sure I'm not missing something). Imo, roles like village roleblocker are better to not use in the early days of a mish mash, so forcing them to submit is stupid.

That's sort of how it works on epic mafia except the village is eventually forced to lunch or lose