Small pack of spring rewards

Solandri - Heavens Tear

Here's the pwdatabase entry on the results for opening a small pack of spring. I wrote down my results for 1000 packs to verify (I have a spreadsheet which calculates my profit margin, and the only numbers I was unsure of were the chances of getting each reward result).

pwdb % - my measured % - Item

12% - 13.1% - Oracle I
10% - 9.9% - Oracle II
10% - 10.1% - Oracle III
15% - 14.4% - Scarlet Fruit
10% - 10.4% - Multiflavor Jiaozi
10% - 9.1% - Sesame Yuanxiao
2% - 2.4% - Mold: Crazed Lupine Ring
2% - 1.6% - Mold: Ring of Blazing Fury
5% - 5.1% - Training Esoterica
8% - 9.5% - Beautiful Garnet Shard
8% - 7.8% - Beautiful Sapphire Shard
8% - 6.6% - Beautiful Citrine Shard

Most of the results are within a percent of what pwdatabase says. The three exceptions are the citrines (1.4% lower than expected), garnets (1.5% higher than expected), and Oracle I (1.1% higher than expected). The Sesame comes close (0.9% lower than expected).

It looks like PWI shaved 1.5% off your chance to get a citrine, and added it to the garnets. And possibly took 1% off your chance to get Sesame and gave it to Oracle I. But this sample size is too small to really say for sure. So I was wondering if anyone else was collecting similar data and if they noticed the same thing.

Calliope - Heavens Tear

I haven't been keeping careful track, but I've gotten just about as many sets of Oracle IIs as IIIs, and almost identical numbers of jiaozi and sesame. I've gotten a few more sapphires than the other types, though.

1000 packs isn't a large enough sample size for the kind of accuracy you want.

BratFury - Heavens Tear

I agree with Calliope. For a more accurate database you'd need to open at least a million. And I doubt that you'd want to do that. You could probably get pretty close with opening around 10,000 of them. But if you think about it, if you only opened 10 and took the percentage off that they would be way off too. I'm not sure how they calculate the percentage, but either way it was very close to be too upset about it with the test you did.

Kudos to having the money and patience to do this and then submit it!

Solandri - Heavens Tear

BratFury - Heavens Tear wrote: »

I agree with Calliope. For a more accurate database you'd need to open at least a million. And I doubt that you'd want to do that. You could probably get pretty close with opening around 10,000 of them.

Yes, that's why I'm asking if anyone else has done this.

Probability-wise, the sample of 1000 gives 1.96*sqrt((0.08*0.92)/1000) = 1.68% margin of error with a 95% confidence interval. So I'm very close to being very certain about this. I just need someone else's data to push my sample size over the edge. A 1.5% margin of error (difference between the 8% listed and 9.5% measured margin for garnet shards) at a 95% confidence interval requires a sample size of 2800.

Of course with 12 different prizes even with a 95% confidence interval there's about a 50% chance of getting one prize falling outside that 95% range. But one step at a time.

Dominatiger - Sanctuary

opened 1000 packs? ... I just opened 93

Solandri - Heavens Tear

Dominatiger - Sanctuary wrote: »

opened 1000 packs? ... I just opened 93

I refer you to what I said here:
http://pwi-forum.perfectworld.com/showthread.php?p=3620131#post3620131

Setting your price high does not maximize profit. Generally, profit is maximized at the lowest price which is still profitable. You make less profit per item, but you sell a whole lot more items. My cat shop is undercutting my competitors by about 1k on nearly every Jolly Jones item. But I am making a fortune because people are buying from me instead of from them.