Hearthpwn

Can you please update that spreadsheet for upcoming druid nerves for melon and petal combo discounter? That should buff the deal a bit.

Nice work!   That WW bundle was pretty awesome.  Still got the 80 Boom one as well, though.

Quote from Dunscot >>

Quote from cassetto >>

 The dust/pack average is currently set on HS Wiki at 102.71, based on the opening of 32,697 packs for a total of 163,485 cards. If you know a better source for this data I'll gladly include that in the table (I'm not being ironic, or sarcastic)

I know that. I just don't like this figure since it's misleading.

I once took a closer look at the tables of pack opening statistics myself and then analysed roughly 1400 packs. I could have done more, but honestly, I did this all in one day and then I was a bit dried out, so bear with me that it's a not a huge number.

Anyway, what I did was to measure the average dust/pack value in blocks of 50 packs. My main problem with the numbers on the wiki was, that they all looked at very large numbers of pack openings, like people opening several hundred or thousand packs. Most players don't open thousands of packs. A more realistic number of packs seemed to me to be 50, since that's the amount you get from a pre-order.

As a result, many of these 50-blocks were under 100 dust per pack. And those that were at or above 100 ususally featured at least 4 or more legendaries or golden epics (400+ packs), which is alreay considered to be above average (1 in 20, 2.5 in 50). And of course, what really skews the results, are golden Legendaries since, as is said, one is already raising the average amount for 50 packs by over 30.

Almost every single block of 50 had 25 or more 40 dust packs. Only one had 24. To give a general idea (no exact numbers here yet), about 35 packs were under 100, and another 10-12 were between 100 and 180.So, it really came down to the best 4 or 5 packs in a block to determine whether it was above or below 100. From a statistical point of view, this is problematic.

If you would try to determine the average temperature in your local region, and found out that the average temperature is about 16°C across the year, but then you had two days where the temperature was 100000°C, while certainly unrealistic, you couldn't really say that the average temperature is far over 100°C across the year. The "average" number is misleading when it is heavily influenced by rare occurences that are far above normal. And that is exactly what happens with packs in Hearthstone. People who claim that the average amount of dust per pack is 100 have to assume, that you open enough packs to get lucky once in a while, preferably with a golden legendary, to end up with that number.

Most people open less packs, have less of a chance to get that lucky, and are more likely to end up with about 92 dust per pack, or even less. .

 While I could agree with your calculations made using a smaller pool of openings (50 blocks) and the calculations results, in the big picture and, generally, as a rule when we talk about statistics, the larger the data pool the more accurate the results. The average dust per pack is obviously different if you pick a different amount of pack pools. You used 50, I could use 5 and I'll get a different result. 

So what's the correct approach? The correct approach is to try and "absorb" the anomalies (3 legendaries per pack) increasing the pack pool. Why? Because let's say instead of analyzing 1400 cards you analyzed 50, and you opened 7 legendaries. Your dust per pack value would've skyrocketed. Let's say you open 100 packs and found 1 legendaries. Your value would be miserable.
Now let's say you opened up both pools, 150 packs. Your average would be different again, but generally more balanced. 
That's why it is always better to have the largest data pool possible.

Think about this other scenario. Think about we could, theoretically, analyze every single pack that has been and will be ever open. That would give us the EXACT dust per pack value that is set within Blizzard's algorithms. Of course, it would be different from your 50 packs pool analysis, but, what analysis will be applicable to the most of the people?

The fact that you say "nobody buys 35000 packs", well, obviously not, but if a data analysis wants to speak to the largest amount of people as possible it should include the largest amount of data as possible. Because yes, the 3 legandaries packs are the rarest, but what if you opened one in those 50 packs you just bought? YOUR value would skyrocket, while mine could be sinking. but for the average Joe, the most correct value is my packs + yours = average Joe openings.

Again, I don't disagree with your calculations, but I do believe that those are yours and yours alone, and the bigger the pool, the better the results.