Monday, December 27, 2010

More data points!

302 samples now. This set of draws were slightly better than the last set. Still, I've yet to see the winning ticket... or, for that matter, any prizes worth more than 20g :(

Cumulative stats summary:

Prize (gold) CountPercentage
0.1 16855.63%
0.5 6019.87%
1 4314.24%
5 258.28%
20 61.99%
50 00.00%
200 00.00%
1000 00.00%
5000 00.00%

Here's a plot of the distribution:

Total Samples: 302
Current total prize money won: 334.8 g
Current estimated fair price: 1 g 10 s 86 c

Here's the list of new data points:
DateTimeItem (linked)Prize (g)DateTimeItem (linked)Prize (g)
2010/12/2714:04:03[Fortune Card]12010/12/2714:20:56[Fortune Card]0.1
2010/12/2714:04:03[Fortune Card]52010/12/2714:20:57[Fortune Card]0.5
2010/12/2714:04:04[Fortune Card]0.12010/12/2714:20:57[Fortune Card]0.1
2010/12/2714:04:06[Fortune Card]202010/12/2714:20:58[Fortune Card]0.1
2010/12/2714:04:06[Fortune Card]0.52010/12/2714:20:59[Fortune Card]0.1
2010/12/2714:04:06[Fortune Card]0.12010/12/2714:25:43[Fortune Card]0.1
2010/12/2714:04:07[Fortune Card]0.12010/12/2714:25:43[Fortune Card]0.1
2010/12/2714:04:08[Fortune Card]0.52010/12/2714:25:44[Fortune Card]1
2010/12/2714:04:09[Fortune Card]0.52010/12/2714:25:45[Fortune Card]0.1
2010/12/2714:04:09[Fortune Card]0.52010/12/2714:25:45[Fortune Card]0.1
2010/12/2714:14:43[Fortune Card]0.12010/12/2714:25:45[Fortune Card]0.5
2010/12/2714:14:43[Fortune Card]0.12010/12/2714:25:46[Fortune Card]20
2010/12/2714:14:44[Fortune Card]52010/12/2714:25:47[Fortune Card]5
2010/12/2714:14:45[Fortune Card]0.52010/12/2714:25:47[Fortune Card]0.1
2010/12/2714:14:46[Fortune Card]0.12010/12/2714:25:49[Fortune Card]0.5
2010/12/2714:14:47[Fortune Card]0.12010/12/2714:32:25[Fortune Card]0.1
2010/12/2714:14:48[Fortune Card]0.12010/12/2714:32:26[Fortune Card]0.1
2010/12/2714:14:49[Fortune Card]52010/12/2714:32:27[Fortune Card]0.1
2010/12/2714:14:50[Fortune Card]0.12010/12/2714:32:27[Fortune Card]0.1
2010/12/2714:14:51[Fortune Card]0.12010/12/2714:32:29[Fortune Card]0.1
2010/12/2714:20:52[Fortune Card]0.12010/12/2714:32:29[Fortune Card]0.1
2010/12/2714:20:53[Fortune Card]0.52010/12/2714:32:31[Fortune Card]0.1
2010/12/2714:20:53[Fortune Card]0.12010/12/2714:32:31[Fortune Card]0.5
2010/12/2714:20:54[Fortune Card]0.12010/12/2714:32:32[Fortune Card]0.1
2010/12/2714:20:55[Fortune Card]12010/12/2714:32:33[Fortune Card]0.1


  1. Which tools do you use for collecting data? I'm interested in statistics for taht card too, so want to help :)

  2. It's all done manually on a spreadsheet at the moment. I would be nice to automate some parts of it though... got any ideas?

  3. Omg... And you are writing down each time and cost?
    My idea is obvious - create an addon to automate data collection :) And output through chat-logging (.../WoW/Logs/WowChatLog.txt)
    Think, I can careate such addon but seeing your data thought that you already have one :)

  4. I have these results from 205 cards:
    114 times 10s
    33 times 50s
    32 times 1g
    20 times 5g
    5 times 20g
    1 time 50g
    None of the 200, 1000 and 5000g cards.

  5. I just did another 100 cards:
    47 times 10s
    26 times 50s
    17 times 1g
    9 times 5g
    1 time 20g

    This is the total out of my 305 cards:
    10s: 161 -> 52.79%
    50s: 59 -> 19.34%
    1g: 49 -> 16.07%
    5g: 29 -> 9.51%
    20g: 6 -> 1.97%
    50g: 1 -> 0.33%

    Combining it with your data:
    10s: 329 -> 54.2%
    50s: 119 -> 19.6%
    1g: 92 -> 15.16%
    5g: 54 -> 8.9%
    20g: 12 -> 1.98%
    50g: 1 -> 0.16%

    I have a theory:
    There are 57 different cards worth 10s, 19 worth 50s, 16 worth 1g, 6 worth 5g, 2 worth 20g and one each for 50g, 200g, 1000g and 5000g. In total this are 104 cards and if they were uniformly distributed the percentages would look like this:
    10s: 54.81%
    50s: 18.27%
    1g: 15.38%
    5g: 5.77%
    20g: 1.92%
    50g-5000g: 0.96% each or 3.85% combined.

    Our data matches quite well for 10s, 50s, 1g, 20g cards but it is way off for the 5g card. It may still be within an unsuspicious interval because the sample size is not very big but consider this thought:
    There is no way that the 1000g card and the 5000g card are dropped with a 1/104 chance because Blizzard is not as stupid as many people like to claim. The 1000g card would rise the value of a single blackfallow ink to about 13g thus giving low-level herbs (Azshara's Veil, Cinderbloom, Heartblossom, Stormvine) a "vendor"-value of roughly 65g per stack (4 times milling @ 2.5 pigments average + burning ember as "trash") and the high-level herbs (Twilight Jasmine, Whiptail) a value of 80g per stack (4 times milling @ 3 pigments average). The 5000g card would result in an incredible 61g "vendor"-value per single ink!

    I think there are two possibilites:
    1)The big difference between the expected percentage of the 5g cards and our data (~2%) is intended. These 2% come from the 1000g and 5000g cards which have a very low drop chance (giving the 5000g card a 1/1000 drop rate would alone raise the "vendor"-value of an ink to 5g and herbs therefore to 20/25g!).
    2)The 1000g and 5000g have a very low drop chance and the deviation we see at the 5g card is still within normal limits because of the small sample size.

  6. Thanks for the interesting data and theory Patrick.

    I'm not sure if I follow your theory though. Where do you get those numbers (e.g., 57 10s, 19 50s, etc)? As far as I can see from and my own samples, there are only 58 unique cards in total. I've thought that they are uniformly distributed too... but that doesn't jibe with my samples so far.

  7. You are right about the 58 unique cards. I had a huge brain-lag and mixed it up with something else. I will have a look at it again tomorrow and see how this fits in my theory.
    But I guess the only reasonable thing to do is simply collect a huge amount of data.

  8. "But I guess the only reasonable thing to do is simply collect a huge amount of data." +1
    Will soon write a simple addon for easy data collection...
    BTW my stats:
    Total of 383 cards
    0.1 : 202 | 52.7%
    0.5 : 76 | 19.8%
    1 : 66 | 17.2%
    5 : 30 | 7.8%
    20 : 8 | 2.0%
    50 : -
    200 : -
    1000: 1 | 0.2%

  9. My Status on 77 cards:

    0.1 : 45
    0.5 : 8
    1 : 15
    5 : 7
    20 : 2
    50 : -
    200 : -
    1000: -

    I spent roughly 3k gold on AH cards. Gambling got me 98.5 gold. Or it returned to me roughly 3.28% of my original money.

  10. Thanks igounfazed. I'll update the user contributed data page soon.

    ... and not to rub it in, but the house always wins ;)