Dominion Strategy Forum

• January 16, 2021, 07:24:30 am
• Welcome, Guest

News:

DominionStrategy Wiki

Pages: [1] 2 3 ... 7  All

AuthorTopic: First player bias  (Read 38066 times)

0 Members and 1 Guest are viewing this topic.

RaVeNLoRD

• Herbalist
• Offline
• Posts: 7
• Respect: 0
First player bias
« on: April 30, 2012, 09:39:03 am »
0

It is well known fact that the 1st player has a slightly higher chance at winning.
However, on everage you should have 50% of your games as 1st player and 50% as secound player.
Right now I have 494 games as 1st player and 606 games as secound player.
The odds for this, assuming even chances, are 1 in 10940.
This most likely makes me the player with the worst luck on isotropic (as far as starting first goes).
Logged

rrenaud

• Offline
• Posts: 990
• Uncivilized Barbarian of Statistics
• Respect: +1184
Re: First player bias
« Reply #1 on: April 30, 2012, 09:40:58 am »
+3

The iso algorithm is more likely to start you in 2nd position if you won your last game, much like the rules suggest that the player who won the previous game goes last.
Logged

• Guest
Re: First player bias
« Reply #2 on: April 30, 2012, 10:21:46 am »
0

The iso algorithm is more likely to start you in 2nd position if you won your last game, much like the rules suggest that the player who won the previous game goes last.
C-C-C-COMBO BREAKER!
Logged

• Torturer
• Offline
• Posts: 1722
• Respect: +1752
Re: First player bias
« Reply #3 on: April 30, 2012, 10:37:44 am »
0

There might actually be something else happening to RaVeNLoRD here...

I checked my own stats, and even with a ~65% win rate, my split is only 880 games in first position, and 909 games in second.

A quick look at WW's stats shows a 4727/4755 split, but his win rate a bit lower at 60% since he matches mostly (exclusively?) against high level opponents.

RaVeNLoRD on the other hand, has a win rate pretty close to 50%, but is still playing from second position a disproportionately large amount of the time.

My expectation would be that the higher your win rate is, the more games you would have from second position.  It is possible there isn't anything going on, but the numbers make me wonder if RaVeNLoRD is somehow logging out after every game, putting himself back in the Winners pool even after a loss.  I vaguely remember a thread where another forum member had this issue.

RaVeNLoRD

• Herbalist
• Offline
• Posts: 7
• Respect: 0
Re: First player bias
« Reply #4 on: April 30, 2012, 11:52:50 am »
0

I log out often.
I also didn't know you have a higher chance at 1st seat after a lose.
Also I had a miss clicked and the odds are roughly 1 in 3,000 for 50%/50% split.
Logged

• Torturer
• Offline
• Posts: 1722
• Respect: +1752
Re: First player bias
« Reply #5 on: April 30, 2012, 12:35:49 pm »
+4

Yeah, so there are two pools.  The first is everyone who just won a game, and everyone who just logged in.  The second is everyone who just lost a game.  If you are matched against someone in the opposite pool, the player who just lost will always go first.  If you are matched against someone in the same pool, the match will be random.

Any time you log out after a loss, you lose the increased chance of being in first position that you should have had.  In the extreme case, where you log out after every match, you will end up being in first position for 50% of the games you play against other people in the winners/just logged in group.  Any games against the just lost group will be from second position.

ecq

• Conspirator
• Offline
• Posts: 224
• Respect: +158
Re: First player bias
« Reply #6 on: April 30, 2012, 01:59:11 pm »
0

Interesting.

I have a 2:1 win/loss ratio in 750 games, but I've only started from P2 28 more times than P1.

Does auto-match try to make people play others in the same pool to even things out?  Or perhaps most people only play 1-2 games after logging in, so the just-logged-in pool is big?
Logged

Fabian

• 2012 Swedish Champion
• Offline
• Posts: 666
• Respect: +541
Re: First player bias
« Reply #7 on: April 30, 2012, 02:36:18 pm »
0

WW,

Do you have an explanation for your 50/50 split despite the 60% winrate?
Logged

WanderingWinder

• Offline
• Posts: 5275
• ...doesn't really matter to me
• Respect: +4371
Re: First player bias
« Reply #8 on: April 30, 2012, 02:40:25 pm »
0

WW,

Do you have an explanation for your 50/50 split despite the 60% winrate?
Not really. I mean, I basically always win-quit, which is going to help, um... lots of very long sessions as opposed to really short ones, and I guess lots of other people play shorter, which means that 1 instance of having second-player bias just from entering the lobby is going to be a smaller percentage of my games in comparison to other people, I mostly always play other really high-level opponents, so they're more likely to be coming off a win too. Do those things all add up to make that much difference? I wouldn't think so, but I guess they do...

Fabian

• 2012 Swedish Champion
• Offline
• Posts: 666
• Respect: +541
Re: First player bias
« Reply #9 on: April 30, 2012, 02:45:44 pm »
0

WW,

Do you have an explanation for your 50/50 split despite the 60% winrate?
Not really. I mean, I basically always win-quit, which is going to help, um... lots of very long sessions as opposed to really short ones, and I guess lots of other people play shorter, which means that 1 instance of having second-player bias just from entering the lobby is going to be a smaller percentage of my games in comparison to other people, I mostly always play other really high-level opponents, so they're more likely to be coming off a win too. Do those things all add up to make that much difference? I wouldn't think so, but I guess they do...

This describes my playing habits pretty well (my long sessions are shorter though since you play more I guess), and it turns out my numbers aren't that far off. Seems like I go first in 47% (1830 - 2041) of games with a winrate 0.5% higher than yours. Would have guessed more games in second position before double checking, oh well.
Logged

rrenaud

• Offline
• Posts: 990
• Uncivilized Barbarian of Statistics
• Respect: +1184
Re: First player bias
« Reply #10 on: April 30, 2012, 02:51:52 pm »
+5

How to get on top of leaderboard on iso:

Make sure not to tell your opponents when they are clearly doing something suboptimal (playing banks first for \$1).
Make sure you leave after every win (but never after a loss), so that you can maximize chance of being first player.
Bias towards prosperity.

Anymore gems?
Logged

• Torturer
• Offline
• Posts: 1722
• Respect: +1752
Re: First player bias
« Reply #11 on: April 30, 2012, 02:54:41 pm »
0

Buy cards that you don't have enough cash for

(And unleash KC/Goons/Masq on unsuspecting players)

olneyce

• 2011 DS Champion
• Offline
• Posts: 245
• Respect: +210
Re: First player bias
« Reply #12 on: April 30, 2012, 02:55:31 pm »
+3

How to get on top of leaderboard on iso:

Make sure not to tell your opponents when they are clearly doing something suboptimal (playing banks first for \$1).
Make sure you leave after every win (but never after a loss), so that you can maximize chance of being first player.
Bias towards prosperity.

Anymore gems?
Get good at the game?
Logged

greatexpectations

• Saboteur
• Offline
• Posts: 1095
• Respect: +1063
Re: First player bias
« Reply #13 on: April 30, 2012, 03:15:48 pm »
0

i've seen this question a few times so i figured i'd throw it on the FAQ.  i also cleaned up the formatting so it should read a little easier.

Yeah, so there are two pools.  The first is everyone who just won a game, and everyone who just logged in.  The second is everyone who just lost a game.  If you are matched against someone in the opposite pool, the player who just lost will always go first.  If you are matched against someone in the same pool, the match will be random.

Any time you log out after a loss, you lose the increased chance of being in first position that you should have had.  In the extreme case, where you log out after every match, you will end up being in first position for 50% of the games you play against other people in the winners/just logged in group.  Any games against the just lost group will be from second position.

i know i have seen this explanation before but do you happen to have a link to where dougz (or someone else?) explained this method?
Logged
momomoto: ...I looked at the tableau and went "Mountebank? That's for jerks."
rrenaud: Jerks win.

WanderingWinder

• Offline
• Posts: 5275
• ...doesn't really matter to me
• Respect: +4371
Re: First player bias
« Reply #14 on: April 30, 2012, 03:34:43 pm »
0

How to get on top of leaderboard on iso:

Make sure not to tell your opponents when they are clearly doing something suboptimal (playing banks first for \$1).
Make sure you leave after every win (but never after a loss), so that you can maximize chance of being first player.
Bias towards prosperity.

Anymore gems?
If you are likely to be P1, use identical starting hands. If not, don't.

DG

• Governor
• Offline
• Posts: 4074
• Respect: +2621
Re: First player bias
« Reply #15 on: April 30, 2012, 03:48:57 pm »
0

Play lots of games. Hard to get to the top of the leader board when you have variance 15.997.
Logged

Obi Wan Bonogi

• Apprentice
• Offline
• Posts: 254
• Respect: +343
Re: First player bias
« Reply #16 on: April 30, 2012, 03:54:41 pm »
0

Anymore gems?

-Use classic proposal instead of auto-match with point counter turned off.   (Ben Warden)
-Propose these games against players that are higher level than they should be. (Ben Warden against me haha)
Logged

ecq

• Conspirator
• Offline
• Posts: 224
• Respect: +158
Re: First player bias
« Reply #17 on: April 30, 2012, 04:37:13 pm »
0

Anymore gems?

-Use classic proposal instead of auto-match with point counter turned off.   (Ben Warden)
-Propose these games against players that are higher level than they should be. (Ben Warden against me haha)

I was just about to put up a separate post on that.  It's annoying but ultimately fine when that happens.  Yesterday, though, I had an auto-match game proposed, declined, proposed again, declined, then a direct invite from the auto-match player with point counter turned off.  Again, the burden was probably still on me to notice that and not just click "yes" to every proposed match.  What got my goat is that the guy denied doing it when I asked him about it.  I assumed the best, then returned to the lobby and noticed that I indeed had point counter enabled so I have no better explanation.

I assume Ben Warden is above board with it.  He probably just likes to pick his opponent, rather than waiting for auto-match.  There's at least one other player, though, who uses it to gain a cheap advantage.  Annoying.

Anyhow, didn't mean to derail.  I'm still fuming just a little.
Logged

Axxle

• Torturer
• Offline
• Posts: 1664
• Most Valuable Serial Killer
• Respect: +1956
Re: First player bias
« Reply #18 on: April 30, 2012, 05:14:05 pm »
0

How to get on top of leaderboard on iso:

Make sure not to tell your opponents when they are clearly doing something suboptimal (playing banks first for \$1).
Make sure you leave after every win (but never after a loss), so that you can maximize chance of being first player.
Bias towards prosperity.

Anymore gems?
If you are likely to be P1, use identical starting hands. If not, don't.
I assume if you're likely to be P2 then you want to play veto mode?  So you can get rid of the cards that vastly favor P1?
Logged
We might be from all over the world, but "we all talk this one language  : +1 card + 1 action +1 buy , gain , discard, trash... " - RTT

rrenaud

• Offline
• Posts: 990
• Uncivilized Barbarian of Statistics
• Respect: +1184
Re: First player bias
« Reply #19 on: April 30, 2012, 05:18:48 pm »
0

It seems like if you are likely to be P2, you are doing it wrong.

Edit: The only wrong thing is me!
Logged

Captain_Frisk

• Saboteur
• Offline
• Posts: 1257
• Respect: +1261
Re: First player bias
« Reply #20 on: April 30, 2012, 05:58:49 pm »
0

It seems like if you are likely to be P2, you are doing it wrong.
If I understand it correctly, any time you're in the winner / neutral pool and playing random opponents - you are "likely" to be P2 - given that they are a mix of people in the Winner / Unknown Pool + the Loser Pool... and the losers will always go first.

Am I missing something?

Edit: Response now invalid because rrenaud was wrong!
« Last Edit: April 30, 2012, 09:15:45 pm by Captain_Frisk »
Logged
I support funsockets.... taking as much time as they need to get it right.

rrenaud

• Offline
• Posts: 990
• Uncivilized Barbarian of Statistics
• Respect: +1184
Re: First player bias
« Reply #21 on: April 30, 2012, 06:07:36 pm »
0

I am not sure how iso works.

I thought there were 3 states.  Won last game, lost last game, or unknown.  If you are laddering well, you should never be in the won last game state when playing.
Logged

Fabian

• 2012 Swedish Champion
• Offline
• Posts: 666
• Respect: +541
Re: First player bias
« Reply #22 on: April 30, 2012, 06:09:02 pm »
0

Thought we established many times over it doesn't work like that?

Edit: I suck!
« Last Edit: April 30, 2012, 06:50:07 pm by Fabian »
Logged

rrenaud

• Offline
• Posts: 990
• Uncivilized Barbarian of Statistics
• Respect: +1184
Re: First player bias
« Reply #23 on: April 30, 2012, 06:41:44 pm »
0

I never delete information, I just fix up spelling mistakes and grammar.  It's still there, just struck through!
Logged