• Welcome to Atlas Freeshard

    Register using Discord

Something Changed w Ranger bow miss rate?

robeoke

Member
Joined
4 Sep 2021
Messages
26
Reaction score
33
In PVE it seems like the miss rate for bow shots, both Critical Shot and regular, has increased. The math always LOOKS right when people say turn on combat stats. Yeah, look, a 2 was rolled and the miss chance was 10% so bang, miss. Except that it's happening a lot more than 10% of the time, and I don't know why. I feel like this is one of those statistics issues where CHANCE does not equal RESULT bc it is CHANCE on every shot. It's like flipping a two sided coin 100 times does not mean 50 heads and 50 tails. Nor does flipping it 1000 times mean 500 heads and 500 tails, etc. I read THE DECK post about RNG but it doesn't match what I tracked.

Is anyone else playing archers noticing this? I was having a fun time levelling rangers over and over through the beta. Until these last 3. The first one I thought I was just tired and grumpy and making the miss #s out to be greater than they actually were. The second I started really noticing the misses. Now this third one I did my own tracking, thousands of bowshots, hundreds of mobs, and found yeah, the actual miss rate IS higher than listed. And now, dismayingly so, I find myself... reluctant to level up another ranger. The number of critical shot opening misses is maddening. The number of multiple misses on single targets where I only get 3 or 4 shots off is maddening.

So what changed? Is this WAI?

Is the answer to reduce the chance to miss rates for bow shots to bring expected actual miss rates back to what they were earlier this year? Idk if that is even possible. How about setting Critical Shot to a standardized 2% flat rate? I could certainly handle 2 missed shots of 4 on an incoming target if my CS landed first.
 

naezgul

Member
Joined
1 Aug 2021
Messages
553
Reaction score
204
If your teaching hundreds or thousands of shots what’s the data ?

Or is the problem on SOME mobs that you have 2 or 3 misses?

How does the deck of cards work ?
At what point going through do you get a new deck?
 

robeoke

Member
Joined
4 Sep 2021
Messages
26
Reaction score
33
If your teaching hundreds or thousands of shots what’s the data ?

Or is the problem on SOME mobs that you have 2 or 3 misses?

How does the deck of cards work ?
At what point going through do you get a new deck?
The last set of data was from Friday and Saturday, 258 mobs w listed miss rate 11%/10%/9%, 974 shots, 154 misses (15.8%), Critical Shot misses 41 (26.6%). No Rapid Fire, 197 mobs 4 shots fired, 61 mobs 3 shots fired.

35 of the 258 mobs pulled had 2 misses. 1 mob had 3 of 4 shots missed.

As I understand it, the Deck RNG generates a set of numbers from 1 to whatever # (let's say 100 to make it easy) and each time a random number is generated, that number is removed from the deck, until there are no more numbers left at which point a new deck is generated and the process repeats. This is supposed to ensure that a 10% chance to miss is held to only 10 times in 100 (100 being this example of the highest number generated in our example) as once numbers 1 through 10 are removed, every remaining number generated would be higher than 10 and therefor not a miss. But this system cannot possibly be WAI if 10% miss chances are generating 16% actual misses.
 

robeoke

Member
Joined
4 Sep 2021
Messages
26
Reaction score
33
Fen DECK RNG

This is the post where Fen explains the RNG and DECK system.
 

Dong Johnson

Member
Joined
31 Dec 2021
Messages
74
Reaction score
37
Realm
Albion
The last set of data was from Friday and Saturday, 258 mobs w listed miss rate 11%/10%/9%, 974 shots, 154 misses (15.8%), Critical Shot misses 41 (26.6%). No Rapid Fire, 197 mobs 4 shots fired, 61 mobs 3 shots fired.

35 of the 258 mobs pulled had 2 misses. 1 mob had 3 of 4 shots missed.

As I understand it, the Deck RNG generates a set of numbers from 1 to whatever # (let's say 100 to make it easy) and each time a random number is generated, that number is removed from the deck, until there are no more numbers left at which point a new deck is generated and the process repeats. This is supposed to ensure that a 10% chance to miss is held to only 10 times in 100 (100 being this example of the highest number generated in our example) as once numbers 1 through 10 are removed, every remaining number generated would be higher than 10 and therefor not a miss. But this system cannot possibly be WAI if 10% miss chances are generating 16% actual misses.
Forgive me if you already accounted for this or if it’s mentioned in Fen’s post you linked (I only skimmed it) but if the deck size is significantly bigger than your sample size, wouldn’t that be a super simple explanation for why you’re seeing higher miss rates with your analysis?

For example, if your largest sample size is 250, but the deck size is 1,000, then you’d potentially have 100 misses with a 10% miss rate before the deck system starts “working” to “correct” any bad luck RNG, right?
 

Wreckanishu

Member
Joined
20 Jul 2021
Messages
292
Reaction score
178
Did a test today, 171 misses over 1200 fired arrows, so 14.125% on Hunter with a newly made RR1 toon and a 99% qua rog bow. Admittedly, 1200 shots is not ideal, but 14.125% still ends up showing a 41.25% increase when compared to the displayed 10% chance to miss over that sample size.

Attached the log file for those interested to look at potential streaks.
 

Attachments

  • chat1.log
    265.8 KB · Views: 23
Last edited:

robeoke

Member
Joined
4 Sep 2021
Messages
26
Reaction score
33
Forgive me if you already accounted for this or if it’s mentioned in Fen’s post you linked (I only skimmed it) but if the deck size is significantly bigger than your sample size, wouldn’t that be a super simple explanation for why you’re seeing higher miss rates with your analysis?

For example, if your largest sample size is 250, but the deck size is 1,000, then you’d potentially have 100 misses with a 10% miss rate before the deck system starts “working” to “correct” any bad luck RNG, right?
This is very true and it would require more expansive testing, but it would be irresponsible to make a deck with such a large range as to skew desired results. If it IS this, then reducing that deck range would be a wonderfully simple way to fix it.

The recent sample size is almost 1k shots, I should hope the deck would be significantly less than that. I mean, truly why even bother with a deck larger than 10 if you want to achieve an actual 10% result. Pull the numbers 1 thru 10 as they are selected and Bob's your uncle. If they want variance, I'm ok with that but to a point.

There is mention of "streaks" in this system and holy cow there sure are massive streaks in every selection I tracked. One streak had 1 miss in 50 shots... That was followed by 9 misses in 15 shots. There was a streak of 5 misses in 12 shots with 4 of those 5 being the damn critical shot openers.
 

Wreckanishu

Member
Joined
20 Jul 2021
Messages
292
Reaction score
178
We had a bug with the Deck of Cards system that caused around a 10% increase to miss rates. This was fixed in Atlas Patch v1.7.9b1. Would be interesting to see your results/tests now.
119 misses out of 1800 arrows fired for an observed 6.61% miss rate (RR1 hunter with 40+11 bow, 99%qua 16.5bow, yellow bc buff)
Log attached below (don't pay any mind to damage, it's on leather but without resists, because i am terrible at remembering atlas /commands)
 

Attachments

  • chat5.log
    391.1 KB · Views: 16
Last edited:

flame

Atlas Team
Joined
23 Apr 2022
Messages
6
Reaction score
13
119 misses out of 1800 arrows fired for an observed 6.61% miss rate (RR1 hunter with 40+11 bow, 99%qua 16.5bow, yellow bc buff)
Log attached below (don't pay any mind to damage, it's on leather but without resists, because i am terrible at remembering atlas /commands)
and 10% is the expected miss rate?
 

Wreckanishu

Member
Joined
20 Jul 2021
Messages
292
Reaction score
178
and 10% is the expected miss rate?
10% is the % miss rate that shows up in the log whenever an arrow shot misses it's mark yeah.
I am not entirely sure if i am understanding the deck system 100%, but i had observed around 4% too high miss rate in my previous test, now i am around 4% under, but maybe it's still sample size variance.
 

Wreckanishu

Member
Joined
20 Jul 2021
Messages
292
Reaction score
178
I also did a puppydog test, on leather, this time with proper resists hiked to 26%, and counted 312 misses out of 1297 attacks, for an observed 24.06% miss rate. I ran similar test on the pupper yesterday and had observed 25.39% on a 697attacks run and 24.54% on a 705 attacks run, so i am guessing it didn't change and the patch correction was just for players :D No idea if that missrate is intended or no for a lvl41 pet. Seems a touch high?

It's off-topic as well, but the hunter pet dmg is under-performing compared to live, and i will run BD tests but i think it's also underperforming compared to the BD commander pet too. source: http://web.archive.org/web/20120211142334/http://vnboards.ign.com/daoc_galahad/b20663/53619594/p1 edit: it is underperforming compared to the BD pet :D

Log attached below.
 

Attachments

  • chat6.log
    112.1 KB · Views: 17
Last edited:

flame

Atlas Team
Joined
23 Apr 2022
Messages
6
Reaction score
13
Yeah the Deck of Cards is only used for player attacks.
 

robeoke

Member
Joined
4 Sep 2021
Messages
26
Reaction score
33
I started another test yesterday but got busy IRL so it's kind of short:

107 mobs w listed miss rate 13%, 345 shots 67 misses (19.4%), Critical Shot misses 19 (28.4%). No Rapid Fire, 43 mobs 4 shots fired, 40 mobs 3 shots fired, 24 mobs 2 shots fired. Bonus stat: at least one miss on 55 of the 107 mobs.

12 of the 107 mobs pulled had 2 misses. 0 mobs had 3 shots missed.

I hope this was before the bug was found bc it lines up with the previous results: (+49.2% increased miss rate above listed 13% to 19.4% versus earlier test +58% increased miss rate above listed 10% to 15.8%).
 

robeoke

Member
Joined
4 Sep 2021
Messages
26
Reaction score
33
So far miss rate is down SIGNIFICANTLY. 378 shots on a 10% miss mob and 14% miss mob but generated a 7.4% actual miss rate. Much better.
 

Wreckanishu

Member
Joined
20 Jul 2021
Messages
292
Reaction score
178
I know it's not going to be too popular, but my last test showed a 6% missrate instead of a 10% one on 1800 fired arrows, isn't that a bit low, given the intricacies of the deck of card thing? or is it still sample size at this point
 

Fenyn

Atlas Team
Joined
29 Jul 2021
Messages
65
Reaction score
336
Our deck is admittedly slightly stacked in favor of the player. Observing 6-8% miss rates on an expected 10% is likely a direct result of that.

We had to re-adjust how heavily the deck was weighted after discovering the bug with the deck, as the bug was adding roughly 10% additional miss rates, so it may shift a bit closer to 10% but you can still expect it to be slightly more friendly than raw RNG.
 

Similar threads

Top