Notes on new team difficulty settings


Abigail Frost

 

Posted

I forgot to mention I got a costume recipe.. dunno if that's significant since we believe it is a separate pool.


 

Posted

Quote:
Originally Posted by Chaos Creator View Post
My very small testing indicates what I believe to be expected drop rates. ~ half of liberate 8 spawn no boss, -1, I got 7 commons and an uncommon recipe
I also did a tiny bit of very informal testing and took my AR/EM blaster to the wall in Cim. I was only able to spend a few minutes at it, but the drop rate 'felt' right to me, based on many, many previous runs live. I didn't come close to filling up on salvage and recipes, but I'll go back and spend some more time at it and see how fast the two fill up. I've got a pretty good feel for how fast both salvage and recipe storage should max out there.


 

Posted

If there are certain maps that do this, it might be the ghost of a bug.

Remember the cargo ship map, which used a not-so-random number generator? It might be a good idea to count which mobs are dropping recipes. The demon ship map had only minions, which made it easy to count. When the spawn was minions and Lt's, the count got all messed up, but overall patterns remained, they were just harder to detect because defeat order mattered, and it's possible the order in which things were spawned mattered as well.

I think what's bothering me is that these maps seem to have a set number of drops given, within a small margin, and regardless of how many the maps are actually spawned for. Might be worth checking out the demon ship map and doing the counts again.


119088 - Outcasts Overcharged. Heroic.

 

Posted

I reran liberate entirely 8 spawn -1 no bosses;

17 recipes: 14 common 3 uncommon
52 salvage
5 enhancements
879,280 infamy


 

Posted

What we need for figuring out if there's a change is that data plus how many of each rank of mob was defeated.

We can't tell from the aggregate view if there were non-proportional changes among the different mob ranks, but we can tell if the aggregate is different from expected values. Alternatively, with data from live we can tell if the "expected values" we're using are actually valid for live.


Blue
American Steele: 50 BS/Inv
Nightfall: 50 DDD
Sable Slayer: 50 DM/Rgn
Fortune's Shadow: 50 Dark/Psi
WinterStrike: 47 Ice/Dev
Quantum Well: 43 Inv/EM
Twilit Destiny: 43 MA/DA
Red
Shadowslip: 50 DDC
Final Rest: 50 MA/Rgn
Abyssal Frost: 50 Ice/Dark
Golden Ember: 50 SM/FA

 

Posted

Hmm I don't think I'm likely to find 6 fillers redside.


 

Posted

Yep. That's why I've been asking the devs to just tell us if this is expected or not, and/or if live drop rates have changed since I9. That way we can avoid the pain of gathering data we don't have to that I know would be really painful to collect. At least I know it would be painful for me.


Blue
American Steele: 50 BS/Inv
Nightfall: 50 DDD
Sable Slayer: 50 DM/Rgn
Fortune's Shadow: 50 Dark/Psi
WinterStrike: 47 Ice/Dev
Quantum Well: 43 Inv/EM
Twilit Destiny: 43 MA/DA
Red
Shadowslip: 50 DDC
Final Rest: 50 MA/Rgn
Abyssal Frost: 50 Ice/Dark
Golden Ember: 50 SM/FA

 

Posted

Some of you were asking for hard data on current drop rates. I believe the rate is about 1 recipe per 40 defeats. 80% of these recipes are whites (commons).

The rest of this post just gives the facts behind this figure.

TopDoc's data gives about 29 defeats per recipe. (18,116 defeats and 623 recipes.) He did his runs in the "Liberate TV" mission redside in March this year. TD estimated that 30% of his defeats were lieutenants.

I've recently (August) been working the wall in Cim. (Also redside) I've received 124 recipes in 4,471 for a rate of 36 defeats per recipe. About 10% of my defeats were LTs.

According to the Wiki, LTs drop twice as often as minions. Adjusting the kill rates to account for this gives 38 and 40 minions equivalents respectively. This is a good match to the Wiki rate of 2.33% which gives 1 recipe per 43 minions.

Hope this helps.


 

Posted

It does.

Sadly, it really does suggest that something on test is lowering drop rates compared to live, because it's working out to lower drops than the Wiki probabilities predict, by a pretty wide margin.


Blue
American Steele: 50 BS/Inv
Nightfall: 50 DDD
Sable Slayer: 50 DM/Rgn
Fortune's Shadow: 50 Dark/Psi
WinterStrike: 47 Ice/Dev
Quantum Well: 43 Inv/EM
Twilit Destiny: 43 MA/DA
Red
Shadowslip: 50 DDC
Final Rest: 50 MA/Rgn
Abyssal Frost: 50 Ice/Dark
Golden Ember: 50 SM/FA

 

Posted

This sounds like a job for Arcanaville.


 

Posted

FWIW, I just played with some numbers to try to work out how many defeats are needed in order to establish whether the drop rate has changed or not.

If we assume that the old drop rate is 1 in 40 (2.33%) and the new rate is EarthWyrm's rate of 1 in 100 (1.00%) then we need more than 1500 defeats to tell them apart with 95% confidence. If the drop rate has changed less dramatically then we need more defeats to tell the difference.

In other words, EarthWyrm's runs of around 2000 defeats are significant. (In the statistical sense.)

The Maths
This is based on treating drops as a normal distribution and looking 2 standard deviations above and below the expected number of drops. If my maths and my assumptions are right (I wouldn't bet on that!) then 95% of the time, we can expect 1500 minion defeats to give 23 to 47 drops at the 2.33% rate and 7 to 23 at 1%.

I'm no statistician, so take this with a big pinch of salt. (Or if you are good at statistics then tell us whether this seems reasonable or not.)


 

Posted

In other words we need to summon Arcanaville.

Arcanaville, Arcanaville, Arcanaville...


 

Posted

Quote:
Originally Posted by Chaos Creator View Post
In other words we need to summon Arcanaville.

Arcanaville, Arcanaville, Arcanaville...
Do you have to turn off the lights and look in a mirror to do that right, or is that only for Hickman?


Freedom: Blazing Larb, Fiery Fulcrum, Sardan Reborn, Arctic-Frenzy, Wasabi Sam, Mr Smashtastic.

 

Posted

Quote:
Originally Posted by Sardan View Post
Do you have to turn off the lights and look in a mirror to do that right, or is that only for Hickman?

Are you implying that I have to turn off the lights when I look in the mirror BECAUSE OF MY ACCURSED UGLINESS?


Troy Hickman - So proud to have contributed to and played in this wonderful CoH universe

 

Posted

Quote:
Originally Posted by Sardan View Post
Do you have to turn off the lights and look in a mirror to do that right, or is that only for Hickman?
I dunno, I've never tried to summon her.


 

Posted

Quote:
Originally Posted by Archie Gremlin View Post
FWIW, I just played with some numbers to try to work out how many defeats are needed in order to establish whether the drop rate has changed or not.

If we assume that the old drop rate is 1 in 40 (2.33%) and the new rate is EarthWyrm's rate of 1 in 100 (1.00%) then we need more than 1500 defeats to tell them apart with 95% confidence. If the drop rate has changed less dramatically then we need more defeats to tell the difference.

In other words, EarthWyrm's runs of around 2000 defeats are significant. (In the statistical sense.)
Precise numbers from the time I was tracking minions and lieutenants separately. (I had another 2000 defeats where this was not the case, so we'll ignore those data, despite their being consistent with what I'll present below.)

1996 minions
488 lt’s

=2484 kills

Salvage: 192
Recipes: 25

Expected recipes: 1996*.026667 + 488*.05333 = 53.23 + 26.03 = 79.29
Expected salvage: 1996*.08 + 488*.1064 = 159.68+51.92 = 211.6

Overall % for recipes = 25/79.29 = .32
Overall % for salvage = 192/211.6 = .91

Using the "quick and dirty" method of computing margins of error

1.96*(sqrt((p*(1-p))/n))

(forgive me if I have one too many or too few parentheses above; I did the calculations a step at a time by hand/calculator, so there's no chance of my having mis-programmed an excel spreadsheet if I goofed on what I typed above)

If I'd had 2500 dudes who were just minions get whacked and we start from the 2.6667% for their expected drops, with 95% probabilities used I think the margin of error is +/- 0.6%. That is, anything from around 2.07% up to 3.27% would be within what we'd expect and wouldn't be cause for concern. Now, it's been a while since I did margins of error, so I may be misinterpreting here, but when I said before that my observations were outside the margin of error, I was being as precise as I could without throwing out all the numbers.

Also, note that I'm starting from a lower expected drop probability than what exists in reality, since almost 20% of my observations were lieutenants, whom ParagonWiki says drop at the 5.3333% rate. So my numbers are even further outside what would be expected. (I think the right expected drop rate for my sample here would be around 3.2%.)

Like UberGuy, I find this testing pretty painful. If all I managed to do was find a map that's borked, that's fine. Last I checked, people aren't big on farming Council warehouses anyway.

Quote:
I'm no statistician, so take this with a big pinch of salt. (Or if you are good at statistics then tell us whether this seems reasonable or not.)
Seemed reasonable to me. When Arcanaville shows up to point and laugh at my maths, I guess we'll get another opinion.


My postings to this forum are not to be used as data in any research study without my express written consent.

 

Posted

One of the reasons I'm running the demon ship over and over is that it spawns only minions (Spawning for x1 and x2 team size- beyond that, it makes LTs that make things harder). So far, run after run, I'm seeing that inspiration and salvage drops seem to spawn according to the virtual team size: double the mobs, double the inspiration and salvage drops.

I can't say that I'm seeing the same for recipes, but I need a LOT more defeats to be able to tell for sure.


119088 - Outcasts Overcharged. Heroic.

 

Posted

QR to EarthWyrm.

It looks like you're using the same analysis that I am.

The difference between your 25 drops and the expected value of 79 is over 6 standard deviations. In other words, there isn't any realistic chance that this is an unlucky roll of the dice.


 

Posted

Synapse just posted to the CB forums (about 20 minutes ago) to say that there have been no adjustments to recipe or salvage drop rates. He also said he's going to run around on the test server and test them out in-play.

So while my data may be six-sigmas out, they're obviously not impossible.


My postings to this forum are not to be used as data in any research study without my express written consent.

 

Posted

Quote:
Originally Posted by EarthWyrm View Post
So while my data may be six-sigmas out, they're obviously not impossible.
Definitely not. But when they're that far off of predicted values, that's a pretty strong hint that it's not supposed to work like that.

I am very relieved to see that there's no intentional change, not just because I don't want it to change (!) but because this allows us to focus testing on the test server and trying to identify conditions that may lead to the difference(s).

I'm looking forward to outdoor, non-instanced results.


Blue
American Steele: 50 BS/Inv
Nightfall: 50 DDD
Sable Slayer: 50 DM/Rgn
Fortune's Shadow: 50 Dark/Psi
WinterStrike: 47 Ice/Dev
Quantum Well: 43 Inv/EM
Twilit Destiny: 43 MA/DA
Red
Shadowslip: 50 DDC
Final Rest: 50 MA/Rgn
Abyssal Frost: 50 Ice/Dark
Golden Ember: 50 SM/FA

 

Posted

Quote:
Originally Posted by UberGuy View Post
Definitely not. But when they're that far off of predicted values, that's a pretty strong hint that it's not supposed to work like that.

I am very relieved to see that there's no intentional change, not just because I don't want it to change (!) but because this allows us to focus testing on the test server and trying to identify conditions that may lead to the difference(s).

I'm looking forward to outdoor, non-instanced results.
I'm relieved as well.

I took a different character to a different map and got... different results.

Defeat Freaks in TV map

308 lieutenants (lots of downgraded bosses, same as in the council map)
309 minions

I missed probably 5 spawns at the beginning before I realized I wasn't tracking right. That's another 80-ish enemies, none of whom dropped recipes. Of the 617 I tracked, I got:

58 salvage
17 recipes, including a purple

Expecteds are around 57.5 and 24.67. So while I didn't get as many recipes as paragonwiki might suggest, it's (a) much higher than what I was getting on the other map and (b) probably within statistical tolerance.

I guess next I'll try going back to the other map with a different character and see if I replicate the low rates, which would indicate it's map-based.


My postings to this forum are not to be used as data in any research study without my express written consent.

 

Posted

Quote:
Originally Posted by EarthWyrm View Post
Expecteds are around 57.5 and 24.67. So while I didn't get as many recipes as paragonwiki might suggest, it's (a) much higher than what I was getting on the other map and (b) probably within statistical tolerance.
I get different numbers here. I'd say the expected values are between 34 and 15. This makes your 17 recipes just within the expected range. However it's also consistent with a drop rate of about 1.2%.


 

Posted

Quote:
Originally Posted by EarthWyrm View Post
Synapse just posted to the CB forums (about 20 minutes ago) to say that there have been no adjustments to recipe or salvage drop rates. He also said he's going to run around on the test server and test them out in-play.
Thanks Synapse.

I wonder if they can analyse the aggregate drop rate for all players from the server logs. This would give a much bigger sample size.


 

Posted

Somebody should try a proper analysis on a rikti map and count vanguard merits too. I ran the RWZ arc mish where you have to kill the leader (Sal'tar ?) on the burning forest map set for 8 with no bosses. I don't run herostats, but killed a number of rikti significantly in excess of 100 and got 1 vanguard merit.


It's true. This game is NOT rocket surgery. - BillZBubba

 

Posted

Quote:
Originally Posted by Minotaur View Post
Somebody should try a proper analysis on a rikti map and count vanguard merits too. I ran the RWZ arc mish where you have to kill the leader (Sal'tar ?) on the burning forest map set for 8 with no bosses. I don't run herostats, but killed a number of rikti significantly in excess of 100 and got 1 vanguard merit.
During CB, a number of people were complaining about reduced AE tickets. I did no testing of that, but I wonder if (assuming it's true, it was quite anecdotal) if it could be related to what we're seeing. That might have some credence if Vanguard Merits are really being reduced.


Blue
American Steele: 50 BS/Inv
Nightfall: 50 DDD
Sable Slayer: 50 DM/Rgn
Fortune's Shadow: 50 Dark/Psi
WinterStrike: 47 Ice/Dev
Quantum Well: 43 Inv/EM
Twilit Destiny: 43 MA/DA
Red
Shadowslip: 50 DDC
Final Rest: 50 MA/Rgn
Abyssal Frost: 50 Ice/Dark
Golden Ember: 50 SM/FA