TOPIC: Google Doc for 2016-2017 Season Treasure Pulls

Dave wrote:

jedibcg wrote:

bpsymington wrote:

Dave wrote: Looking at my totals again just gets me more depressed. It's hard to complain, because I had 2 really good pulls, but percentage wise it still really sucks compared to most. I think I am going to have a friend do my pulls at GHC. Something needs to be done to break out of this rut.

That's two more than I got!

He got 6 'other' it is just that two of them were good. I would argue that one of them really really good in that no one else has reported getting one (except me in almost 900 pulls).

It would be great to get more data, currently we are almost at double the data I originally had. With what we have so far I would guess 20% monster, 5% other, 37.5% rare, and 37.5% uncommon. But more data would help. Thanks all that have contributed.

No argument, one of my pulls was really, really, really good. It was about on my 70th pull and I was feeling really good after that. Then to go through another 100+ pulls with only a Minotaur Hide and Mystic Silk to show was disheartening. I still felt good until I saw other percentages. Who else was averaging 1 "other" for every 40 or more pulls? At the end of the day, that one really, really good item still makes me feel ok about my totals. And certainly it was much better than last year when all I got was 1 UR in 160+ pulls or the years prior to that (not counting the year with the stacked treasure box) with nothing special to show. Maybe it's an upward trend? Yeah, I think I'll go with that.

I'll trade you my 3x Charm of Health, Shirt of Baneful Prowess, Enhanced Creeper Amulet, Exalted Creeper Amulet, and the 5 or so random trade items I got for your 25,000 GP bar - if you wanted to get more specials that is .

I think the only reasonable way to evaluate luck here is to look at the average value of a pull. Selectively looking at area X and saying, yeah - I did above average there, but everywhere else I did below average is a recipe for dissatisfaction.

Let's see how you fare against this schedule:

Commons - 25 cents
Rares - 50 cents
Monster bits - 7.50
GP - 1.75 cents per GP
URs - 75
Fleece - 75
Trade items - 8
Enhanced - 5
Exalted - 45
Relic - 250
Legendary - 1100

This puts the value of Jedi's pulls at $5.51, Jeff's at $5.02, yours at $3.94, and mine at $3.90.

So - you seem to be at the bottom of the curve, but I got 20 out of 371 specials to your 6 out of 242 and ended up worse off per pull.

Yeah I know completions are tricky too because unless you know what they all are you'd think it's a regular rare. The only 2 I got were the WYC one though. Those certainly weren't accidental.

And the sample size just affects the margin of error I guess. But you can get good data with pretty small samples. Look at the presidential election polls. They poll a few hundred people and that's good enough for the whole state.

Matthew Hayward wrote:

Dave wrote:

jedibcg wrote:

bpsymington wrote:

Dave wrote: Looking at my totals again just gets me more depressed. It's hard to complain, because I had 2 really good pulls, but percentage wise it still really sucks compared to most. I think I am going to have a friend do my pulls at GHC. Something needs to be done to break out of this rut.

That's two more than I got!

He got 6 'other' it is just that two of them were good. I would argue that one of them really really good in that no one else has reported getting one (except me in almost 900 pulls).

It would be great to get more data, currently we are almost at double the data I originally had. With what we have so far I would guess 20% monster, 5% other, 37.5% rare, and 37.5% uncommon. But more data would help. Thanks all that have contributed.

No argument, one of my pulls was really, really, really good. It was about on my 70th pull and I was feeling really good after that. Then to go through another 100+ pulls with only a Minotaur Hide and Mystic Silk to show was disheartening. I still felt good until I saw other percentages. Who else was averaging 1 "other" for every 40 or more pulls? At the end of the day, that one really, really good item still makes me feel ok about my totals. And certainly it was much better than last year when all I got was 1 UR in 160+ pulls or the years prior to that (not counting the year with the stacked treasure box) with nothing special to show. Maybe it's an upward trend? Yeah, I think I'll go with that.

I'll trade you my 3x Charm of Health, Shirt of Baneful Prowess, Enhanced Creeper Amulet, Exalter Creeper Amulet, and the 5 or so random trade items I got for your 25,000 GP bar - if you wanted to get more specials that is .

You got Charms of Health from the boxes? Wow in 878 pulls I can say I got zero. All my enhanced were shirts or amulets.

Jeff321 wrote: Yeah I know completions are tricky too because unless you know what they all are you'd think it's a regular rare. The only 2 I got were the WYC one though. Those certainly weren't accidental.

And the sample size just affects the margin of error I guess. But you can get good data with pretty small samples. Look at the presidential election polls. They poll a few hundred people and that's good enough for the whole state.

WYC completion tokens I did count as rares.

Matthew Hayward wrote: Jeff's at $5.02

Thanks for doing this analysis! I was meaning to do so myself. One note though, my relic was Icecrag Hero's which is worth ~$150 rather than $250. It sounded like many of those were pulled.

Jeff321 wrote: And the sample size just affects the margin of error I guess. But you can get good data with pretty small samples. Look at the presidential election polls. They poll a few hundred people and that's good enough for the whole state.

Yeah, but the margin of error on those is usually +/- something like 4%. While that might be useful for polls like those, nobody's going to be impressed if you tell them the rate of "others" in the token bags is 5%, plus or minus 4%.

I just took the numbers we have so far, and plugged them into the first result via an online search for "sample size calculator". Using the sample size we have thus far (1761), and estimating the total token pool at 48,000, these are the 95% confidence intervals our data supports:

Uncommons: 35.26% to 39.70%
Rares: 35.54% to 39.98%
Bits: 17.66% to 21.30%
Other: 4.10% to 6.12%

In other words, there's a 95% chance that the actual numbers fall somewhere inside those ranges. More data points would allow us to narrow the ranges down.

Jeff321 wrote:

Matthew Hayward wrote: Jeff's at $5.02

Thanks for doing this analysis! I was meaning to do so myself. One note though, my relic was Icecrag Hero's which is worth ~$150 rather than $250. It sounded like many of those were pulled.

Yeah - I also got a Icegrag Hero's Earcuff as my only relic. At 250 I'm trying to balance between all the relics and vaguely assuming they are in there at the same density.

If it's 4 Icegrag Hero's Earcuff per current year transmute the value is too high.

Allen John wrote: Yeah, but the margin of error on those is usually +/- something like 4%. While that might be useful for polls like those, nobody's going to be impressed if you tell them the rate of "others" in the token bags is 5%, plus or minus 4%.

I just took the numbers we have so far, and plugged them into the first result via an online search for "sample size calculator". Using the sample size we have thus far (1761), and estimating the total token pool at 48,000, these are the 95% confidence intervals our data supports:

Uncommons: 35.26% to 39.70%
Rares: 35.54% to 39.98%
Bits: 17.66% to 21.30%
Other: 4.10% to 6.12%

In other words, there's a 95% chance that the actual numbers fall somewhere inside those ranges. More data points would allow us to narrow the ranges down.

Yeah I don't know that it is 48,000. That could be too low or too high, just the number someone threw out there. But that is nice to know that if it is that that we have to to the a good sampling size.

Matthew Hayward wrote:
Yeah - I also got a Icegrag Hero's Earcuff as my only relic. At 250 I'm trying to balance between all the relics and vaguely assuming they are in there at the same density.

If it's 4 Icegrag Hero's Earcuff per current year transmute the value is too high.

I doubt the density was the same, don't have any solid evidence, just the fact that it 'seems' there were a lot of Blessed Plates pulled vs Blessed Mails. I don't have evidence of this just people trying to trade the plate for mails and not the other way.

I also realized my numbers were off because I did pull two earcuffs that I had labelled incorrectly as trade items. So I have adjusted my numbers according like. My 4 Relics were 2 earcuffs and 2 plates.

The overall population doesn't make much difference. Dropping it from 48,000 to 24,000 (the theoretical minimum, since that's 3x the number of tickets) changes the U/R confidence intervals from +/- 2.22% to +/- 2.18%, and multiplying the 48k estimate by 10 only raises it from +/- 2.22% to +/- 2.26%. Bits and Other similarly only change a couple hundredths.

Matthew Hayward wrote:

Dave wrote:

jedibcg wrote:

bpsymington wrote:

Dave wrote: Looking at my totals again just gets me more depressed. It's hard to complain, because I had 2 really good pulls, but percentage wise it still really sucks compared to most. I think I am going to have a friend do my pulls at GHC. Something needs to be done to break out of this rut.

That's two more than I got!

He got 6 'other' it is just that two of them were good. I would argue that one of them really really good in that no one else has reported getting one (except me in almost 900 pulls).

It would be great to get more data, currently we are almost at double the data I originally had. With what we have so far I would guess 20% monster, 5% other, 37.5% rare, and 37.5% uncommon. But more data would help. Thanks all that have contributed.

No argument, one of my pulls was really, really, really good. It was about on my 70th pull and I was feeling really good after that. Then to go through another 100+ pulls with only a Minotaur Hide and Mystic Silk to show was disheartening. I still felt good until I saw other percentages. Who else was averaging 1 "other" for every 40 or more pulls? At the end of the day, that one really, really good item still makes me feel ok about my totals. And certainly it was much better than last year when all I got was 1 UR in 160+ pulls or the years prior to that (not counting the year with the stacked treasure box) with nothing special to show. Maybe it's an upward trend? Yeah, I think I'll go with that.

I'll trade you my 3x Charm of Health, Shirt of Baneful Prowess, Enhanced Creeper Amulet, Exalted Creeper Amulet, and the 5 or so random trade items I got for your 25,000 GP bar - if you wanted to get more specials that is .

I think the only reasonable way to evaluate luck here is to look at the average value of a pull. Selectively looking at area X and saying, yeah - I did above average there, but everywhere else I did below average is a recipe for dissatisfaction.

Let's see how you fare against this schedule:

Commons - 25 cents
Rares - 50 cents
Monster bits - 7.50
GP - 1.75 cents per GP
URs - 75
Fleece - 75
Trade items - 8
Enhanced - 5
Exalted - 45
Relic - 250
Legendary - 1100

This puts the value of Jedi's pulls at $5.51, Jeff's at $5.02, yours at $3.94, and mine at $3.90.

So - you seem to be at the bottom of the curve, but I got 20 out of 371 specials to your 6 out of 242 and ended up worse off per pull.

Trust me, I know that this is all perception and I realize that the one really good pull brings me back into average value range. if I had pulled 5 URs instead of that one item I'd be no better off, but probably would have felt better.

Still, if I decide to try and transmute a legendary item this year, that one pull will probably be the one that will allow me to do it. I'm sure I'll feel pretty good then.

Allen John wrote: The overall population doesn't make much difference. Dropping it from 48,000 to 24,000 (the theoretical minimum, since that's 3x the number of tickets) changes the U/R confidence intervals from +/- 2.22% to +/- 2.18%, and multiplying the 48k estimate by 10 only raises it from +/- 2.22% to +/- 2.26%. Bits and Other similarly only change a couple hundredths.

That is awesome I thought it would affect it more.

jedibcg wrote: I doubt the density was the same, don't have any solid evidence, just the fact that it 'seems' there were a lot of Blessed Plates pulled vs Blessed Mails. I don't have evidence of this just people trying to trade the plate for mails and not the other way.

I suspect that's just because the Mail is a better token.

Mail can be worn by any Plate wearing class with only a minor downside.

Mail is the best option for Ranger, which is going to push the desirability.

Basically if you got a Mail, and you're not a Wizard or Monk you're like: SWEET! I'M KEEPING THIS.

If you got plate and you're not one of 4 classes you're like: D'oh! I can't use this - maybe I can trade...