PVP Drop Rates

GrimSkald

Hello all! A bunch of us have been discussing the drop rates of PVP rewards in this particular topic. This subject has been of interest to me - I started a PVP reward tracker waaaaay back when they started dropping specific characters instead of standard tokens, and gone through several iterations of it. My most recent one tracks data on different lines, so I can parse some of the data differently (e.g. are Lightning Rounds different than other PVPs?) This is a relatively recent development - so there's not a lot of data there. Regardless, I thought I'd post this. I'd love to see if anyone else has been collecting this data, maybe we can get a good model.

Anyway, the above topic started when people noticed a lower 2* drop rate on the lightning rounds. I had not been tracking anything separately until then, so I started. I quickly noticed a distinct difference between the Heroic Lightning Rounds (every third one,) and the Villainous Lightning Rounds. I also separated out the Season Sim for the most recent season, so I had the following hypotheses:

- All Featured Character PVP has roughly the same drop rates - if this isn't so it's going to be hard to test as the sample size is going to be really small.

- the Season Simulator has the same drop rate as Featured Character PVP

- Heroic Lightning Rounds have a higher chance of 2*s

- Villainous Lightning Rounds are similar to other PVPs.

So my theory is that the Heroic LRs are the outlier, everything else is pretty much the same. I don't think the data is particularly conclusive, but so far it doesn't quite support that theory. Here's my breakdown:

Featured Character PVP
Reward ---- Count --- %
70 ISO-8 --- 154 --- 34.8%
140 Iso-8 --- 100 --- 22.6%
250 Iso-8 --- 30 --- 6.8%
1* Cover --- 133 --- 30.1%
2* Cover --- 25 --- 5.7%
Total --- 442


Season Sim
Reward ---- Count --- %
70 ISO-8 --- 11 --- 39.3%
140 Iso-8 --- 4 --- 14.3%
250 Iso-8 --- 3 --- 10.7%
1* Cover --- 10 --- 35.7%
2* Cover --- 0 --- 0.0%
Total --- 28

Villians LR
Reward ---- Count --- %
70 ISO-8 --- 246 --- 46.2%
140 Iso-8 --- 143 --- 26.9%
250 Iso-8 --- 23 --- 4.3%
1* Cover --- 88 --- 16.5%
2* Cover --- 32 --- 6.0%
Total --- 532

Heroes LR
Reward ---- Count --- %
70 ISO-8 --- 140 --- 36.1%
140 Iso-8 --- 73 --- 18.8%
250 Iso-8 --- 12 --- 3.1%
1* Cover --- 51 --- 13.1%
2* Cover --- 112 --- 28.9%
Total --- 388

With only 28, I think we can discard the Season Sim as statistically insignificant. Here's with that number folded in to the Featured Character PVP:

Normal + Season Sim
Reward ---- Count --- %
70 ISO-8 --- 165 --- 35.1%
140 Iso-8 --- 104 --- 22.1%
250 Iso-8 --- 33 --- 7.0%
1* Cover --- 143 --- 30.4%
2* Cover --- 25 --- 5.3%
Total --- 470

What I wasn't counting on was the difference between that and the Villains LR to be so sharp. Nearly twice as many 1*s drop in "normal" PVP compared to the Villains LR. The difference is made up mostly in 70 Iso, but we also have a slightly higher amount in 140 as well. As expected, the Heroic LRs are very different than others - the 70 Iso is about the same as "normal," 140 Iso slightly lower, 250 Iso noticeably lower (but the drop rate on that is so low it's difficult to be really accurate with it,) the 1* drop rate is less than half, but the 2* drop rate is through the roof - something like 5x normal.

Of course, with only 1000 points of data it's hard to be absolutely certain, but it does look like we have some very significant trends. While I definitely have less Heroic LRs than other forms of data, I tracked four week's worth, and all of them showed pretty much this trend, here are the %s (in order, 70 Iso, 140 Iso, 250 Iso, 1*, 2*)

32.1% --- 15.4% --- 2.6% --- 7.7% --- 42.3%

31.4% --- 13.7% --- 2.0% --- 17.6% --- 35.3%

40.4% --- 19.2% --- 4.1% --- 13.7% --- 22.6%

35.4% --- 23.0% --- 2.7% --- 14.2% --- 24.8%

So even my worst week had quite a lot more than normal.

tizian2015

The important thing to say is: It´s lowered since the patch with the higher iso-sell-rates for full-leveled champions.

And the data of yours show this. In the other thread your first note was this statistic (mostly data BEFORE the questioned patch):

70 ISO-8 590 32.7%
140 Iso-8 290 16.1%
250 Iso-8 51 2.8%
1* Cover 429 23.8%
2* Cover 443 24.6%
Total 1803

The overall data of your first post in this thread ( data PAST the patch) is:

70 ISO-8 551 39,6%
140 Iso-8 320 23,0%
250 Iso-8 68 4,9%
1* Cover 282 20,3%
2* Cover 169 12,2% (!)
Total 1390

Your 2* droprate was halfed by the patch.

As I wrote in the other thread, its a nerf because we discussed if rechampioning 2*s would be a good thing or not, D3 knows it is and so they nerf the overall 2* droprate with the patch. The pattern of 2 weeks villains and one week champs in lightningrounds have the effect that surplus 2* covers from a champs week for not rechampable chars (like a red for moonstone which has red on 5 covers) WILL probably expire (if not getting enough covers from tokens in this time) in the two bad villainweeks and you have to wait for the next champs week. So with this (because of expiring 2* covers) rechamping is extended by another factor.

Another side effect is, that your earned iso has dropped by about 10% (excluding skip taxes or the max 30 iso bonus). Before patch you earn 135,7 iso per match, after it 122,9 iso. Sounds not much, but for the 1390 matches you earned 170820 iso and with the old droprates by complete selling the covers you could have earned 188623 iso.

And we have enough data, let me quote user simonsez from the other thread:

´And before anyone says, "The sample size is too small, we need 8 million LRs in order to really see what the drop rate is", just take a moment and learn how to do a significance test between two binomial distributions, or ask someone who does. And use that information to compute the probability that hundreds of pulls at 1% and hundreds of pulls at 20%, were drawn from the same population. And then stop insisting we need more data.´

DFiPL

I'm an outlier apparently because I have 10 2* drops from 20 LR plays today in a Villain round, and while that's higher than what I usually see, even my 'normal' Villain drop rate is above what's been reported. Conversely, I get bupkis in Heroic LR's.

FaustianDeal

I hope I statistical outlier. In the previous 3 weeks I counted the 2* cover drops I received from Lightning Rounds:

(2* covers dropped / total matches during the 24 rounds)
Rounds between 8/30-9/2 - 2/148
Rounds between 9/6-9/8 - 0/163
Rounds between 913-9/15 - 11/165 -- 1 round dropped 8 covers. Without that 1 round this week would have been abysmal also.
So far this week: 0/35.

Across the window I've counted so far I am 13/511. That's a 2* prize rate of roughly 2.5%

I am convinced there is something wrong with the 2* reward drop rate in LRs.

FaustianDeal

Continuing to hope I'm an outlier:

(2* covers dropped / total matches during the 24 rounds)
Rounds between 8/30-9/2 - 2/148
Rounds between 9/6-9/8 - 0/163
Rounds between 9/13-9/15 - 11/165 -- 1 round was 8/22. Without that 1 round this week would have been abysmal also.
Rounds between 9/20-9/22 - 1/111
Rounds between 9/27-9/29 - 2/104
Rounds between 10/4-10/6 - 1/102

Over 6 weeks:
17/793 -- if you leave aside the 1 anomalous round its 9/771.

Is 2% the intended drop rate?
I get how random distributions work, but it seems like poor design to not have a safeguard in place for when the RNG just has to make a point

DFiPL

DFiPL wrote:

I'm an outlier apparently because I have 10 2* drops from 20 LR plays today in a Villain round, and while that's higher than what I usually see, even my 'normal' Villain drop rate is above what's been reported. Conversely, I get bupkis in Heroic LR's.

I haven't gotten many seeds in the two LR's I've played so far today, but of the fights I've had, five 2* covers from 7 seeds.

yay villain week