Dominion Strategy Forum

Please login or register.

Login with username, password and session length
Pages: [1]

Author Topic: Popular Buys Analysis  (Read 4140 times)

0 Members and 1 Guest are viewing this topic.

rod-

  • Conspirator
  • ****
  • Offline Offline
  • Posts: 213
  • Respect: +49
    • View Profile
Popular Buys Analysis
« on: June 16, 2011, 06:03:06 pm »
0

Excluding prizes, I have a terrible "effect with" rate with the most powerful cards: Why?
Idea 1:  These cards are powerful 5$s and sometimes shuffle luck will bite you and you will lose.  This is a somewhat boring hypothesis, but could be confirmed by a systematic negative effect among top players: (Yes, i'm being somewhat presumptuous putting myself in this company, but I started by looking at myself because i'm trying to find places to improve my game)

                      Rod|CFrisk|Theory|Rrenaud|Boloni
Mountebank:     -1.2|-0.7|-1.4|+0.4|+0.6
Witch :             -0.7|+0.4|-0.6|-2.9|-0.7
Familiar:          -0.5|+1.5|-2.1|+0.1|-0.2
Ghost Ship:      -1.3|+0.5|-0.9|-0.7|+0.2
Ambassador:    -1.1|-0.5|+0.6|+1.8|-0.3
(Control: pawn) -0.2|+0.4|+0.7|+0.5|-0.0

While each player other than me has 1 of the 4 power 5$ cards they are strong with, and ambassador probably does not belong in the list, it does seem to me that each of these cards has a pretty strong negative impact overall.

My conclusion: Look at how i play ambassador & improve that.  Worry about the 5$ attacks later on.
« Last Edit: June 16, 2011, 06:10:32 pm by rod- »
Logged

david707

  • Baron
  • ****
  • Offline Offline
  • Posts: 51
  • Respect: +17
    • View Profile
Re: Popular Buys Analysis
« Reply #1 on: June 16, 2011, 06:30:42 pm »
0

My hypothesis (may be very wrong):
I think in most games with (insert powerful $5 card here), both player's will buy it which skews the win rate towards 50%, overall meaning top players like theory have a lower win rate than usual if a powerful card is in the setup, which hurts the effect with. I'm not easily going to accept an explanation of  "some of the top player's need to work on how they play powerful $5 attacks".

Edit:
My stats: (I'm level 19 mainly due to high uncertainty due to not playing many games on isotropic, that's my excuse anyway :) ;) )

Mountebank: -0.10
Witch: +1.39
Familiar: -0.03
Ghost Ship: -0.88
Ambassador: +0.35
Pawn: -0.74

If your to go just on that, I'm far better at playing Witch then all the player's you mentioned, yet far worse at playing pawn.

« Last Edit: June 16, 2011, 06:38:51 pm by david707 »
Logged

fp

  • Thief
  • ****
  • Offline Offline
  • Posts: 94
  • Respect: +5
    • View Profile
Re: Popular Buys Analysis
« Reply #2 on: June 16, 2011, 06:37:35 pm »
0

My hypothesis (may be very wrong):
I think in most games with (insert powerful $5 card here), both player's will buy it which skews the win rate towards 50%, overall meaning top players like theory have a lower win rate than usual if a powerful card is in the setup, which hurts the effect with. I'm not easily going to accept an explanation of "some of the top player's need to work on how they play powerful $5 attacks".

 8) +1
Logged

david707

  • Baron
  • ****
  • Offline Offline
  • Posts: 51
  • Respect: +17
    • View Profile
Re: Popular Buys Analysis
« Reply #3 on: June 16, 2011, 06:42:49 pm »
0

Witch: +1.39

If your to go just on that, I'm far better at playing Witch then all the player's you mentioned...

It seems my high Witch stat is because I won all 6 games I've played with Witch involved according to councilroom.
Logged

rrenaud

  • Administrator
  • *****
  • Offline Offline
  • Posts: 990
  • Uncivilized Barbarian of Statistics
  • Respect: +1184
    • View Profile
    • CouncilRoom
Re: Popular Buys Analysis
« Reply #4 on: June 16, 2011, 07:17:02 pm »
+1

You are mis-intepretting the stat.

All it says is that david_with_witch - david_overall > other_player_with_witch - other_player_overall.

Logged

david707

  • Baron
  • ****
  • Offline Offline
  • Posts: 51
  • Respect: +17
    • View Profile
Re: Popular Buys Analysis
« Reply #5 on: June 16, 2011, 07:43:11 pm »
0

You are mis-intepretting the stat.

All it says is that david_with_witch - david_overall > other_player_with_witch - other_player_overall.



Okay that's how it works, seems to confirm my hypothesis. So effectively, theory_with_mountebank (MB) - theory_overall (OV) < other_player_with_MB - other_player_OV. This is because theory_with_MB - theory_OV is negative, since theory overall is higher than theory with MB, possibly due to the whole both player's buying one thing. Also, other_player_with_MB - other_player_OV is positive, since for the average player's win rate increases by buying MB, which makes sense since if you buy MB and your opponent doesn't that's a likely win.

I think the main things influencing the stat are:
-How good are you
-How good are the people you are plating against.

So rrenaud, stop playing newbs and using MB to win :);)
Logged

theory

  • Administrator
  • *****
  • Offline Offline
  • Posts: 3594
  • Respect: +6044
    • View Profile
    • Dominion Strategy
Re: Popular Buys Analysis
« Reply #6 on: June 16, 2011, 11:39:45 pm »
0

You can still always look at "Win Rate With" if you want a more direct number than "Effect With", which is much more nuanced.

But you're right: KC and Mountebank tend to be a little more chaotic, and it's difficult to get win rates far above 1.00.
Logged

boloni

  • Steward
  • ***
  • Offline Offline
  • Posts: 25
  • Respect: +8
    • View Profile
Re: Popular Buys Analysis
« Reply #7 on: June 17, 2011, 12:52:44 am »
0

I feel very honored to be listed among the top players.
I don't really know what to make out of the popular buy statistics. I know that I'm often successful in building decks where I can play a ghost ship every turn. And I try too often to give out curses with my ambassadors instead of buying victory cards myself. That could explain why I'm better with ghost ship than with ambassador. No idea why I'm good with mountebank though.

My power card seems to be pearl diver. For me it has the highest Win Rate given available(1.56), the second highest Win Rate with(1.74) and the highest Effect with(3.93)
Logged

rod-

  • Conspirator
  • ****
  • Offline Offline
  • Posts: 213
  • Respect: +49
    • View Profile
Re: Popular Buys Analysis
« Reply #8 on: June 17, 2011, 01:26:19 am »
0

Actually had an interesting thought about trying to correlate position to win rate given the availability of this card.  Particularly:

Given that mountebank is available, does the p1 vs p2 win rate shift relative to the average?  If so, it is possible that top players will have a lower 'effect with' card X simply by virtue of being in the p2 position more often than average.

I'd enjoy playing with the data a bit, but i'm not quite ready to learn python yet...although perhaps i should just do it.

Alternatively, I could get the big tarball and just use what i know...where's that located at again?
« Last Edit: June 17, 2011, 01:29:37 am by rod- »
Logged

rrenaud

  • Administrator
  • *****
  • Offline Offline
  • Posts: 990
  • Uncivilized Barbarian of Statistics
  • Respect: +1184
    • View Profile
    • CouncilRoom
Re: Popular Buys Analysis
« Reply #9 on: June 17, 2011, 09:51:00 am »
0

Here is the big tarball.

http://councilroom.com/margin.txt.gz
Logged

rod-

  • Conspirator
  • ****
  • Offline Offline
  • Posts: 213
  • Respect: +49
    • View Profile
Re: Popular Buys Analysis
« Reply #10 on: June 17, 2011, 01:30:15 pm »
0

Just to verify that I'm analyzing the data correctly:
 for the first column in the margin ; I assume it is point difference, player A minus player B ?  And player A is 1st player, listed to the left?  And the 0.5 point indicates that the 2nd player missed a turn?

Logged

rod-

  • Conspirator
  • ****
  • Offline Offline
  • Posts: 213
  • Respect: +49
    • View Profile
Re: Popular Buys Analysis
« Reply #11 on: June 17, 2011, 02:10:46 pm »
0

Anyway, assuming that I read the margin data correctly, my "idea" at 1:30am was without any merit whatsoever:

Regardless of the particular cards present, the win rate for player 1, the ratio of games with even/uneven # of turns, and the winrate given either an even or uneven number of turns were constant. 
That is to say:  Playing in seat #2 is equally bad in the presence + absence of all individual cards.

Data:
% of Games ended with an uneven number of turns: P1 Win % if even #turns ;  P1 Win% if uneven # turns
All Games
.554.532.635
Witch
.555.535.630
.552.530.633
.551.537.619
.553.529.624
.552.528.634
.556.533.630
(Minion Familiar Ghost ship Pawn Mountebank, for the previous 5 rows)
« Last Edit: June 17, 2011, 03:24:17 pm by rod- »
Logged

rrenaud

  • Administrator
  • *****
  • Offline Offline
  • Posts: 990
  • Uncivilized Barbarian of Statistics
  • Respect: +1184
    • View Profile
    • CouncilRoom
Re: Popular Buys Analysis
« Reply #12 on: June 17, 2011, 02:58:04 pm »
0

Yeah, you have parsed the data correctly.
Logged

Superdad

  • Moneylender
  • ****
  • Offline Offline
  • Posts: 161
  • Respect: +2
    • View Profile
Re: Popular Buys Analysis
« Reply #13 on: June 23, 2011, 04:43:23 pm »
0

The effect in play here is card-strength, and how that ADVERSELY (relative to weaker boards) affects top player win-rates.

Strong players have a BIGGER edge on non-obvious boards. When there are harder choices to make, the better players will make the correct call more often than weaker players. However, for boards with Montebank or Witch, it's almost always (baring something like masquerade or ambassador) correct to rush for the powerful $5 attack. This is obvious, even to a lesser skilled opponent. On boards like this, the better player just lost some of his "edge".

So this lesser skilled opponent is quite lucky when he opens a kingdom with some of these game-defining cards. Their tough decision on the intial buy is essentially made for them, and this is one less area that they might get outplayed this game. This is GOOD for the worse player of the two, and BAD for the better player.

The top players would get immensely more "edge" on non-obvious boards, where their higher knowledge/experience would lead them to buy correctly more often than a non-elite player.


TL:DR... this stat really only tells me that in games with powerful cards, the powerful cards dominate. In really whacky boards, the top players would outplay a slightly worse opponent more-often. I.e. if I play against theory, I should be happy if I open an obvious witch-dominated game.. because half my decisions are pre-made for me, and I have less room to make errors against such a strong opponent.

« Last Edit: June 23, 2011, 04:49:06 pm by Superdad »
Logged

Amaranth

  • Ambassador
  • ***
  • Offline Offline
  • Posts: 33
  • Respect: 0
    • View Profile
Re: Popular Buys Analysis
« Reply #14 on: June 24, 2011, 06:02:50 am »
0

Hmm...my stats are:

Mountebank: 0.03
Witch: -0.29
Familiar: -0.49
Ghost Ship: 1.35
Ambassador: -0.02
Pawn: -0.45

I will note that I hardly ever buy Ghost Ship, so when I do buy it it must be for a good reason. Interestingly, I have a positive Effect Without for Familiar, which isn't true of any of the others, where I always benefit by buying (yes I'm aware that's not how it works).

The card I do worst with is Treasure Map. I have a -3.03 Effect With and a -1.59 Effect Without. My best Effect With is Treasury at 2.19 (and -1.76 Effect Without). Royal Seal is the card with the strongest positive effect on my record regardless of whether I buy it.

Incidentally, I am apparently more likely to win games where I don't buy Provinces, Gold, Platinum, Colonies, Duchies and Silver. I'm not sure what all this says about me, but I like long action chains, and the cursing attacks disrupt that kind of playstyle.
« Last Edit: June 24, 2011, 06:07:37 am by Amaranth »
Logged

Death to Sea Hags

  • Thief
  • ****
  • Offline Offline
  • Posts: 90
  • Respect: +20
    • View Profile
Re: Popular Buys Analysis
« Reply #15 on: June 24, 2011, 08:35:38 am »
0

Interestingly, I have a positive Effect Without for Familiar, which isn't true of any of the others, where I always benefit by buying (yes I'm aware that's not how it works).

You must be very good at seeing when you can skip it.

Incidentally, I am apparently more likely to win games where I don't buy Provinces, Gold, Platinum, Colonies, Duchies and Silver. I'm not sure what all this says about me, but I like long action chains, and the cursing attacks disrupt that kind of playstyle.

You REALLY like to end on piles, with VP or, with Gardens/Vineyards/Fairgrounds?
Logged

Amaranth

  • Ambassador
  • ***
  • Offline Offline
  • Posts: 33
  • Respect: 0
    • View Profile
Re: Popular Buys Analysis
« Reply #16 on: June 24, 2011, 08:33:07 pm »
0

Interestingly, I have a positive Effect Without for Familiar, which isn't true of any of the others, where I always benefit by buying (yes I'm aware that's not how it works).

You must be very good at seeing when you can skip it.
I think I only skip it when there are other cursing alternatives and no other strong cards requiring Potion.

Quote from: Death to Sea Hags
Incidentally, I am apparently more likely to win games where I don't buy Provinces, Gold, Platinum, Colonies, Duchies and Silver. I'm not sure what all this says about me, but I like long action chains, and the cursing attacks disrupt that kind of playstyle.

You REALLY like to end on piles, with VP or, with Gardens/Vineyards/Fairgrounds?
Well, I think there's a loser's buy effect on Duchy, in that it's the kind of card you buy when your opponent is ahead on Provinces. I rarely go for it as a positive strategy. I have a positive Effect With for Duke, Vineyards, Fairgrounds, Island and Harem, and negatives for Nobles and Great Hall. Mind you, my Effect Without is more positive for Duke and Vineyards than my Effect With. I also have a positive Effect With for Bishop and Goons, but negative for Monuments. I do tend to go for pile endings when I see them, though one of the main times I go for them is when there are cursing cards out.
Logged

drg

  • Spy
  • ****
  • Offline Offline
  • Posts: 83
  • Respect: +1
    • View Profile
Re: Popular Buys Analysis
« Reply #17 on: June 26, 2011, 07:23:51 pm »
0

Hmm...my stats are:

Mountebank: 0.03
Witch: -0.29
Familiar: -0.49
Ghost Ship: 1.35
Ambassador: -0.02
Pawn: -0.45

I will note that I hardly ever buy Ghost Ship, so when I do buy it it must be for a good reason. Interestingly, I have a positive Effect Without for Familiar, which isn't true of any of the others, where I always benefit by buying (yes I'm aware that's not how it works).

The card I do worst with is Treasure Map. I have a -3.03 Effect With and a -1.59 Effect Without. My best Effect With is Treasury at 2.19 (and -1.76 Effect Without). Royal Seal is the card with the strongest positive effect on my record regardless of whether I buy it.

Incidentally, I am apparently more likely to win games where I don't buy Provinces, Gold, Platinum, Colonies, Duchies and Silver. I'm not sure what all this says about me, but I like long action chains, and the cursing attacks disrupt that kind of playstyle.

If you just simply neglected to buy those cards at all times, you would find your win rate would drop immensely.  The reason your win rate with them is high, is that you choose not to buy them in specific games where there are other strategies (goons, gardens, yinyard, etc for VP cards, action chains worth money for treasure cards) and your opponent either doesn't or doesn't use it effectively as you do.  I'm sure your percent of not buying them when available is quite low, except for the lesser green cards in colony games, where it may well be to your advantage to not buy them.
Logged
Pages: [1]
 

Page created in 0.085 seconds with 21 queries.