Issues with critical hits. - Phoenix Freeshard

I've been running MoP7-9 and noticing extremely long periods of combat with ZERO crits. I understand RNG but is it realistic to not see ONE crit with 49% crit chance on 52 landed hits. Unfortunately releasing ate some of my combat log but this shit is absurd. What am I spending 34 points on?

For reference this was in DF around 0130 EST on Sunday with MoP9.

https://imgur.com/a/jntEA0s

Edit: and after release I go and hit the dummy outside of svasud and an immediately see crits on at least every other swing like you'd expect. wtf.

Double Edit: To add insult to injury, archer guards no longer have draw times? 4 shots by the time CD goes off with 1.7sec swing delay? Did these terminator guards really need an upgrade??

https://imgur.com/a/JlGZreL

Mon 2 Sep 2019 9:05 AM by gotwqqd

Mavella wrote: ↑
Mon 2 Sep 2019 5:52 AM
I've been running MoP7-9 and noticing extremely long periods of combat with ZERO crits. I understand RNG but is it realistic to not see ONE crit with 49% crit chance on 52 landed hits. Unfortunately releasing ate some of my combat log but this shit is absurd. What am I spending 34 points on?

For reference this was in DF around 0130 EST on Sunday with MoP9.

https://imgur.com/a/jntEA0s

Edit: and after release I go and hit the dummy outside of svasud and an immediately see crits on at least every other swing like you'd expect. wtf.

Double Edit: To add insult to injury, archer guards no longer have draw times? 4 shots by the time CD goes off with 1.7sec swing delay? Did these terminator guards really need an upgrade??

https://imgur.com/a/JlGZreL

Odds of that happening are around 1 in 4 quadrillion

You are reading something wrong or there is a glitch

And from what I understand they even have an algorithm to smooth out the rng results. i.e. if you are on the wrong end of rng the chance it happens again are skewed in your favor

Mon 2 Sep 2019 9:37 AM by

gruenesschaf

Crits use PRD as do procs, the failure streak counter is per weapon hand so that you don't "consume" mh crits with OH.

https://dota2.gamepedia.com/Random_distribution

As for the archer guards: they now draw when an enemy is in range and hold indefinitely until someone gets back into los.

Mon 2 Sep 2019 12:27 PM by Sepplord

gruenesschaf wrote: ↑
Mon 2 Sep 2019 9:37 AM
As for the archer guards: they now draw when an enemy is in range and hold indefinitely until someone gets back into los.

Indefinitely sounds unreasonable.
Is it a technical issue (aka letting the lowering their bows after a time takes additional work) or a conscious decision?

Imo they should at least drop their bows when the keep isn't under attack or after 5minutes or something, otherwise they will just be similar to instant-shooters unless killed

Mon 2 Sep 2019 1:36 PM by

gruenesschaf

They start drawing when they have los and someone is in range and either shoot as soon as someone goes into los or stop holding once nobody is in range anymore, but it's indefinite while some enemy is still in range.

There were basically 3 choices for their behavior:
1) Like in OF (previous behavior): ignore los, normal aggro list -> shoots the same target that eventually will be some healer as long as that one is in range, basically do nothing once that healer is out of range
2) Like in NF: respect los, normal aggro list -> Rarely shoots, might as well not exist
3) Like it is now: respect los, ignore aggro, randomly shoot something you can see

Mon 2 Sep 2019 1:55 PM by Mavella

gruenesschaf wrote: ↑
Mon 2 Sep 2019 9:37 AM
Crits use PRD as do procs, the failure streak counter is per weapon hand so that you don't "consume" mh crits with OH.

https://dota2.gamepedia.com/Random_distribution

As for the archer guards: they now draw when an enemy is in range and hold indefinitely until someone gets back into los.

Shouldn't PRD all but guarantee a streak like the one I posted is impossible. I fought a rr9 inf a few nights ago and neither of us crit once and this was also in DF. I know I had MoP7 at the time obviously I can't speak for what they had but I imagine at least MoP5. I didn't notice the lack of crits until we fought but usually ill go through the combat log and evaluate what happened when I lose a 1v1. It's a surprise if two high RR sins fight and never crit each other once.

The only common thread I can see between these events is being logged off in DF, log in the game and I don't have a crit chance until you zone out.

Mon 2 Sep 2019 4:34 PM by Warlay

Mavella wrote: ↑
Mon 2 Sep 2019 1:55 PM
gruenesschaf wrote: ↑
Mon 2 Sep 2019 9:37 AM
Crits use PRD as do procs, the failure streak counter is per weapon hand so that you don't "consume" mh crits with OH.

https://dota2.gamepedia.com/Random_distribution

As for the archer guards: they now draw when an enemy is in range and hold indefinitely until someone gets back into los.

Shouldn't PRD all but guarantee a streak like the one I posted is impossible. I fought a rr9 inf a few nights ago and neither of us crit once and this was also in DF. I know I had MoP7 at the time obviously I can't speak for what they had but I imagine at least MoP5. I didn't notice the lack of crits until we fought but usually ill go through the combat log and evaluate what happened when I lose a 1v1. It's a surprise if two high RR sins fight and never crit each other once.

The only common thread I can see between these events is being logged off in DF, log in the game and I don't have a crit chance until you zone out.

this strange things happens quite often serveral months back and also happens with misses, evade, fumble and more. several ppl adressed this problem but got smashed down by hype ppl or devs. it is because they use a custom rng code which does strange things. one of the main things why i quitted, besides NF :p

Mon 2 Sep 2019 5:03 PM by

gruenesschaf

Warlay wrote: ↑
Mon 2 Sep 2019 4:34 PM
this strange things happens quite often serveral months back and also happens with misses, evade, fumble and more. several ppl adressed this problem but got smashed down by hype ppl or devs. it is because they use a custom rng code which does strange things. one of the main things why i quitted, besides NF :p

Good thing that misses, evade and fumble - anything other than procs and crits really - use an entirely different rng mechanism.

Mon 2 Sep 2019 5:13 PM by Warlay

gruenesschaf wrote: ↑
Mon 2 Sep 2019 5:03 PM
Warlay wrote: ↑
Mon 2 Sep 2019 4:34 PM
this strange things happens quite often serveral months back and also happens with misses, evade, fumble and more. several ppl adressed this problem but got smashed down by hype ppl or devs. it is because they use a custom rng code which does strange things. one of the main things why i quitted, besides NF :p

Good thing that misses, evade and fumble - anything other than procs and crits really - use an entirely different rng mechanism.

i could write good excuse but its senseless to start a different discussion then im sry that i wrote it i go back to lurking this server

Mon 2 Sep 2019 5:32 PM by

gruenesschaf

Warlay wrote: ↑
Mon 2 Sep 2019 5:13 PM
i could write good excuse but its senseless to start a different discussion then im sry that i wrote it i go back to lurking this server

What I mainly wanted to point out is that the rng for crits and the defensive rolls are not related at all, even if there was an issue with one it wouldn't affect the other, PRD for example at the end still uses the default rng implementation of the .net framework while the defensive rolls don't use rng at all but are fully deterministic.

The rng for all defensive combat rolls was changed somewhat recentlyish: at first in alpha it was normal rng after the expected and typical complaints at the start of the beta it was changed to our rng bag system (numbers 1 - 1000 inside, each time draw one, fill and restart when empty) but with 1 bag per roll which lead to weird corner cases when you have multiple higher chances. Later, near the end of beta, it was changed to use actual rng for parry, block and evade but kept this bag stuff for misses / full resists and sometime recentlyish it was changed to combine all those separate defensive rolls into a single roll, keeping the statistical properties / order of the relevant chances and this one roll uses the bag stuff. The bag is filled in such a way that the numbers avoid unlikely streaks.

Mon 2 Sep 2019 6:01 PM by Killaloth

gruenesschaf wrote: ↑
Mon 2 Sep 2019 9:37 AM
Crits use PRD as do procs, the failure streak counter is per weapon hand so that you don't "consume" mh crits with OH.

https://dota2.gamepedia.com/Random_distribution

As for the archer guards: they now draw when an enemy is in range and hold indefinitely until someone gets back into los.

I have a couple of questions about PRD on weapon procs.

1) Does weapon speed affect proc chance?

I run solo with a 2.7mh and 2.3oh, both rogs. It seems to me that when switching to Broodmother fang the dot proc happens more often than the drainlife on the 2.7mh. This puzzles me as I'm swinging slower with a 4.3mh and overtime I would expect less procs.

2) Do weapons have individual proc rates?
I don't think so but this would explain question 1)

Thanks

Mon 2 Sep 2019 8:10 PM by

gruenesschaf

Killaloth wrote: ↑
Mon 2 Sep 2019 6:01 PM
I have a couple of questions about PRD on weapon procs.

1) Does weapon speed affect proc chance?

I run solo with a 2.7mh and 2.3oh, both rogs. It seems to me that when switching to Broodmother fang the dot proc happens more often than the drainlife on the 2.7mh. This puzzles me as I'm swinging slower with a 4.3mh and overtime I would expect less procs.

2) Do weapons have individual proc rates?
I don't think so but this would explain question 1)

Thanks

Yes to both, player crafted procs and rogs are set to 15%, epic dungeon / feather item procs to 18% at 3.5 weapon speed, slower weapons have a higher chance per hit and faster weapons a lower one, literally chance * weapon speed / 3.5 .

Mon 2 Sep 2019 8:19 PM by gotwqqd

This “bag of tiles” to smooth out rng...does it ever reset? Or is it persistent until emptied then “new bag”

Mon 2 Sep 2019 9:04 PM by Cadebrennus

gruenesschaf wrote: ↑
Mon 2 Sep 2019 9:37 AM
Crits use PRD as do procs, the failure streak counter is per weapon hand so that you don't "consume" mh crits with OH.

https://dota2.gamepedia.com/Random_distribution

As for the archer guards: they now draw when an enemy is in range and hold indefinitely until someone gets back into los.

Holy shit that is brutal! (Referring to the indefinite draw time.) Even players that have their bow drawn will get the message "you are too tired to hold your shot any longer!" In addition it's super easy for players to lose a target via target out of clipping range, accidentally clicking another player or a structure, or even by being hit by another player or mob. In addition players will not have the instantaneous target acquisition that an AI will have. Nor will players have the ability to shoot through structures and ignore LOS. That last sentence is why clearing Archers from keeps/towers has become a huge priority on this server. If mobs have bows they should at least follow Archery rules which include the numerous drawbacks.

Mon 2 Sep 2019 10:08 PM by

gruenesschaf

gotwqqd wrote: ↑
Mon 2 Sep 2019 8:19 PM
This “bag of tiles” to smooth out rng...does it ever reset? Or is it persistent until emptied then “new bag”

Persistent and then restarting

Mon 2 Sep 2019 10:35 PM by Mavella

Can I least get some conformation the screens I posted are clearly displaying some sort of glitch/bug? With 49% crit chance something is obviously messed up to not see 1 crit with that many attacks.

Tue 3 Sep 2019 12:01 AM by Killaloth

gruenesschaf wrote: ↑
Mon 2 Sep 2019 8:10 PM
Killaloth wrote: ↑
Mon 2 Sep 2019 6:01 PM
I have a couple of questions about PRD on weapon procs.

1) Does weapon speed affect proc chance?

I run solo with a 2.7mh and 2.3oh, both rogs. It seems to me that when switching to Broodmother fang the dot proc happens more often than the drainlife on the 2.7mh. This puzzles me as I'm swinging slower with a 4.3mh and overtime I would expect less procs.

2) Do weapons have individual proc rates?
I don't think so but this would explain question 1)

Thanks

Yes to both, player crafted procs and rogs are set to 15%, epic dungeon / feather item procs to 18% at 3.5 weapon speed, slower weapons have a higher chance per hit and faster weapons a lower one, literally chance * weapon speed / 3.5 .

Thanks, this explains

0.15*2.7/3.5=11.57% fast ROG
VS
0.18*4.3/3.5=22.11% Broodmother fang

One more question: it seems to me that the PRD applies to the mainhand/offhand slots and not to the weapon itself.

So let's say I swing 3 times with a fast mainhand with no proc, then I switch to Broodmother, the chance to proc would be higher than 22.11% as the previous 3 mainhand swings had no proc?

Or switching weapon resets the PRD to 1st attempt?

Thx again!

Tue 3 Sep 2019 12:48 AM by

gruenesschaf

Mavella wrote: ↑
Mon 2 Sep 2019 10:35 PM
Can I least get some conformation the screens I posted are clearly displaying some sort of glitch/bug? With 49% crit chance something is obviously messed up to not see 1 crit with that many attacks.

If your fight against Seolan did not get messed up by you zoning (doesn't look like it as the damage done to him totals 1900+) it looks like something was wrong there. If you really had MoP9 at that time and hence 49% crit, you would have a guaranteed crit per weapon hand on the 4th hit (~30% on the first, 60% on the second, 90% on the third and 100% on the fourth).

The zone itself has no influence, the only things affecting the melee crit rate are Mastery of Pain, Dual Threat, Berserker, Triple Wield and Crit shot. Is there any chance you respecced RAs between logging in and before that fight?

Tue 3 Sep 2019 12:50 AM by

gruenesschaf

Killaloth wrote: ↑
Tue 3 Sep 2019 12:01 AM
One more question: it seems to me that the PRD applies to the mainhand/offhand slots and not to the weapon itself.

So let's say I swing 3 times with a fast mainhand with no proc, then I switch to Broodmother, the chance to proc would be higher than 22.11% as the previous 3 mainhand swings had no proc?

Or switching weapon resets the PRD to 1st attempt?

Thx again!

Switching does not affect it, it's also not directly bound to the slot but the weapon hand used for the actual hit. In practice distance, mh and 2h slot use the same failure counter and the offhand has its own.

Tue 3 Sep 2019 1:26 AM by Mavella

gruenesschaf wrote: ↑
Tue 3 Sep 2019 12:48 AM
Mavella wrote: ↑
Mon 2 Sep 2019 10:35 PM
Can I least get some conformation the screens I posted are clearly displaying some sort of glitch/bug? With 49% crit chance something is obviously messed up to not see 1 crit with that many attacks.

If your fight against Seolan did not get messed up by you zoning (doesn't look like it as the damage done to him totals 1900+) it looks like something was wrong there. If you really had MoP9 at that time and hence 49% crit, you would have a guaranteed crit per weapon hand on the 4th hit (~30% on the first, 60% on the second, 90% on the third and 100% on the fourth).

The zone itself has no influence, the only things affecting the melee crit rate are Mastery of Pain, Dual Threat, Berserker, Triple Wield and Crit shot. Is there any chance you respecced RAs between logging in and before that fight?

No respec. I was logged at Mid entrance. So I logged in, used the hastener, ran down buffed up near mid 4 way and then stealthed over to hib side. Killed the few people included in the screenshots and then engaged that ranger. I of course died in a fight I didn't really expect to and scrolled up my combat log to find all that. I then went to the dummies outside svasud and immediately had multiple crits within the first handful of swings as expected.

I also had a similar scenario the other night against RR9 inf Honeybadger. Log in buff up stealth over to alb side and killed a few people before engaging him. I noticed neither of us crit either other the entire fight which I found to be extremely odd but didn't scroll back to see if I had crit anyone else on any of my kills before that. I know I was MoP7 at the time and that's part of what prompted me to respec to 9 because I found it insane I wasn't getting any crits.

Tue 3 Sep 2019 1:56 AM by

gruenesschaf

Mavella wrote: ↑
Tue 3 Sep 2019 1:26 AM
No respec. I was logged at Mid entrance. So I logged in, used the hastener, ran down buffed up near mid 4 way and then stealthed over to hib side. Killed the few people included in the screenshots and then engaged that ranger. I of course died in a fight I didn't really expect to and scrolled up my combat log to find all that. I then went to the dummies outside svasud and immediately had multiple crits within the first handful of swings as expected.

I also had a similar scenario the other night against RR9 inf Honeybadger. Log in buff up stealth over to alb side and killed a few people before engaging him. I noticed neither of us crit either other the entire fight which I found to be extremely odd but didn't scroll back to see if I had crit anyone else on any of my kills before that. I know I was MoP7 at the time and that's part of what prompted me to respec to 9 because I found it insane I wasn't getting any crits.

Sounds quite odd, try to post a longer screenshot if it happens again and maybe the next time you log into DF take a look at your RAs if they are set properly. It would be even better if you could just enable combat log once you log in and turn it off if you see the expected crit rate / zone.

The only things that come to mind are kind of impossible:
1) Somehow getting a negative failure counter and thereby producing a negative chance -> impossible for 2 reasons: there is never a subtraction only an increment or setting to zero and an overflow would be incredibly unlikely, also impossible due to it working fine again after zoning, zoning does not touch the failure counter at all.

2) Not properly loading RAs or stats in general on login, basically impossible as otherwise a lot more people would report this issue as many people just log into NF and never see a loading screen during their session.

I somehow still suspect that it is number 2: under some rare circumstances you have 0 (or the base 10%, who knows) crit after a login until you zone or force a recalculation of the crit rate (ie by speccing mopain or use anything that affects the crit rate) but nothing in the code suggests that this is possible. The skills / abilities are loaded only and as soon as you click login, everything about your char is recalculated whenever you enter a region which happens whenever the loading screen is done, once on login and whenever you zone, it can never not happen as that would mean you wouldn't even be able to attack.

Tue 3 Sep 2019 2:00 AM by gotwqqd

Even if it was the base chance of 10%
.90^52 is still statistically insane 1 in 266
Not to mention it’s both players exhibiting this
266*266=70,842

Tue 3 Sep 2019 2:18 AM by

gruenesschaf

gotwqqd wrote: ↑
Tue 3 Sep 2019 2:00 AM
Even if it was the base chance of 10%
.90^52 is still statistically insane
Not to mention it’s both players exhibiting this

.90^52 is irrelevant, mh and oh have to be separated, meaning it would be something like 26 or a 6.5% chance of it happening: https://www.wolframalpha.com/input/?i=streak+of+26+successes+in+26+trials+with+p%3D0.9 - still quite unlikely but a lot more plausible.

Anyways, with PRD and a 10% nominal chance your first hit has a 1.5% actual chance and every failure increases the actual chance by 1.5% for the next hit meaning even if incredibly unlikely, you're only guaranteed to have a crit on the 67th try.

Tue 3 Sep 2019 2:32 AM by gotwqqd

gruenesschaf wrote: ↑
Tue 3 Sep 2019 2:18 AM
gotwqqd wrote: ↑
Tue 3 Sep 2019 2:00 AM
Even if it was the base chance of 10%
.90^52 is still statistically insane
Not to mention it’s both players exhibiting this

.90^52 is irrelevant, mh and oh have to be separated, meaning it would be something like 26 or a 6.5% chance of it happening: https://www.wolframalpha.com/input/?i=streak+of+26+successes+in+26+trials+with+p%3D0.9 - still quite unlikely but a lot more plausible.

Anyways, with PRD and a 10% nominal chance your first hit has a 1.5% actual chance and every failure increases the actual chance by 1.5% for the next hit meaning even if incredibly unlikely, you're only guaranteed to have a crit on the 67th try.

It doesn’t matter if they are separated
.90^26 * .90^26 still are the odds that neither crit if there is 10% chance for either to crit
Now if the odds are different for mh/oh it’s a different story

But it sounds like this PRD is different than having a bag with separate chances of incidents. Because if there was a bag with 10 crits, 10 misses, xx fumbles etc, and (100-20-xx-whatever else can happen) there still would be a 10% initial chance of the crit happening.

Edit...well it doesn’t even matter because my argument is flawed using statistics from a true rng and comparing it to a system where those things are impossible

Tue 3 Sep 2019 2:43 AM by Mavella

I'll start being mindful to turn on the chatlog when I log in and keep an eye out for any more funny business.

Thank you.

Tue 3 Sep 2019 2:46 AM by

gruenesschaf

gotwqqd wrote: ↑
Tue 3 Sep 2019 2:32 AM
But it sounds like this PRD is different than having a bag with separate chances of incidents. Because if there was a bag with 10 crits, 10 misses, xx fumbles etc, and (100-20-xx-whatever else can happen) there still would be a 10% initial chance of the crit happening.

PRD is used for crits and procs. Defensive rolls (evade, guard, block, parry, miss etc) use the bag stuff. In my first response I linked an explanation for the used PRD.

Tue 3 Sep 2019 2:55 AM by gotwqqd

gruenesschaf wrote: ↑
Tue 3 Sep 2019 2:46 AM
gotwqqd wrote: ↑
Tue 3 Sep 2019 2:32 AM
But it sounds like this PRD is different than having a bag with separate chances of incidents. Because if there was a bag with 10 crits, 10 misses, xx fumbles etc, and (100-20-xx-whatever else can happen) there still would be a 10% initial chance of the crit happening.

PRD is used for crits and procs. Defensive rolls (evade, guard, block, parry, miss etc) use the bag stuff. In my first response I linked an explanation for the used PRD.

Yea

I just went back and read that stuff

Tue 3 Sep 2019 5:44 AM by easytoremember

If stuff like crits has a "minimum" does it also have a max on it preventing 100% for a series of hits?

Tue 3 Sep 2019 6:10 AM by Sepplord

gruenesschaf wrote: ↑
Mon 2 Sep 2019 1:36 PM
They start drawing when they have los and someone is in range and either shoot as soon as someone goes into los or stop holding once nobody is in range anymore, but it's indefinite while some enemy is still in range.

There were basically 3 choices for their behavior:
1) Like in OF (previous behavior): ignore los, normal aggro list -> shoots the same target that eventually will be some healer as long as that one is in range, basically do nothing once that healer is out of range
2) Like in NF: respect los, normal aggro list -> Rarely shoots, might as well not exist
3) Like it is now: respect los, ignore aggro, randomly shoot something you can see

Ah okay, thanks for clarification. I assumed they would always hold indefintely. If they stop holding when no enemies are around that makes sense.

Tue 3 Sep 2019 6:26 PM by Horus

Killaloth wrote: ↑
Tue 3 Sep 2019 12:01 AM
gruenesschaf wrote: ↑
Mon 2 Sep 2019 8:10 PM
Killaloth wrote: ↑
Mon 2 Sep 2019 6:01 PM
I have a couple of questions about PRD on weapon procs.

1) Does weapon speed affect proc chance?

I run solo with a 2.7mh and 2.3oh, both rogs. It seems to me that when switching to Broodmother fang the dot proc happens more often than the drainlife on the 2.7mh. This puzzles me as I'm swinging slower with a 4.3mh and overtime I would expect less procs.

2) Do weapons have individual proc rates?
I don't think so but this would explain question 1)

Thanks

Yes to both, player crafted procs and rogs are set to 15%, epic dungeon / feather item procs to 18% at 3.5 weapon speed, slower weapons have a higher chance per hit and faster weapons a lower one, literally chance * weapon speed / 3.5 .

Thanks, this explains

0.15*2.7/3.5=11.57% fast ROG
VS
0.18*4.3/3.5=22.11% Broodmother fang

One more question: it seems to me that the PRD applies to the mainhand/offhand slots and not to the weapon itself.

So let's say I swing 3 times with a fast mainhand with no proc, then I switch to Broodmother, the chance to proc would be higher than 22.11% as the previous 3 mainhand swings had no proc?

Or switching weapon resets the PRD to 1st attempt?

Thx again!

Sorry. I'm still waking up but is this more or less saying there in no proc benefit to using faster weapons, especially RoGs? Slower weapons make up for the number of swings of faster weaps by procing more often and at higher dmg?

Tue 3 Sep 2019 6:46 PM by easytoremember

Horus wrote: ↑
Tue 3 Sep 2019 6:26 PM
Sorry. I'm still waking up but is this more or less saying there in no proc benefit to using faster weapons, especially RoGs? Slower weapons make up for the number of swings of faster weaps by procing more often and at higher dmg?

It depends on your haste bonuses and if you're being hit with debuffs. If you're swinging at cap speed the slower weapon is always superior, but once you dip below that you're losing swings, and less swings reduces likelihood both of a proc and of landing a hit to proc in the first place

While less swings makes more [weapon] damage in a given swing it also means the loss brought on by a miss is huge, and the overall dps is less (cap speed vs uncapped for same weapon).
The proc damage doesn't change

Tue 3 Sep 2019 9:32 PM by Cadebrennus

Horus wrote: ↑
Tue 3 Sep 2019 6:26 PM
Killaloth wrote: ↑
Tue 3 Sep 2019 12:01 AM
gruenesschaf wrote: ↑
Mon 2 Sep 2019 8:10 PM
Yes to both, player crafted procs and rogs are set to 15%, epic dungeon / feather item procs to 18% at 3.5 weapon speed, slower weapons have a higher chance per hit and faster weapons a lower one, literally chance * weapon speed / 3.5 .

Thanks, this explains

0.15*2.7/3.5=11.57% fast ROG
VS
0.18*4.3/3.5=22.11% Broodmother fang

One more question: it seems to me that the PRD applies to the mainhand/offhand slots and not to the weapon itself.

So let's say I swing 3 times with a fast mainhand with no proc, then I switch to Broodmother, the chance to proc would be higher than 22.11% as the previous 3 mainhand swings had no proc?

Or switching weapon resets the PRD to 1st attempt?

Thx again!

Sorry. I'm still waking up but is this more or less saying there in no proc benefit to using faster weapons, especially RoGs? Slower weapons make up for the number of swings of faster weaps by procing more often and at higher dmg?

Indeed. I'm not sure when this change went into effect but it has been around for a very long time. If I had to guess then I'm guessing for at least as long as sometime between SI and TOA.

Thu 5 Sep 2019 12:20 AM by Killaloth

gruenesschaf wrote: ↑
Tue 3 Sep 2019 12:48 AM
Mavella wrote: ↑
Mon 2 Sep 2019 10:35 PM
Can I least get some conformation the screens I posted are clearly displaying some sort of glitch/bug? With 49% crit chance something is obviously messed up to not see 1 crit with that many attacks.

If your fight against Seolan did not get messed up by you zoning (doesn't look like it as the damage done to him totals 1900+) it looks like something was wrong there. If you really had MoP9 at that time and hence 49% crit, you would have a guaranteed crit per weapon hand on the 4th hit (~30% on the first, 60% on the second, 90% on the third and 100% on the fourth).

The zone itself has no influence, the only things affecting the melee crit rate are Mastery of Pain, Dual Threat, Berserker, Triple Wield and Crit shot. Is there any chance you respecced RAs between logging in and before that fight?

With MoP9 on 4th swing you get 116.62% to be precise, with 16.62 percentage pts wasted.

Once the crit occurs, the counter is reset.

Therefore ~92% crit chance on 4th hit with MoP8 suggests that it's perhaps worth going for MoP8 and use the 7rps left to increase swing speed.

But also:

Crit on 3rd hit: 87.47% MoP9 VS 68.97% MoP8

So one could argue that being 87.47% very likely to reset the counter at the 3rd hit with MoP9, going pain9 is better as you need 4 hits with MoP8 to achieve just ~3.5% more chance to reset the counter (91.96%). While it's ~5hits with MoP7 to get to 87.45%

For the curious: crit on 2nd hit: 58.31% MoP9 VS 45.98% MoP8

Also note of the C value below for different MoP levels. Interestingly MoP8 gets you almost the same amount of C points per rp spent than MoP9 so one could argue that it's a better RPs investment than MoP9.

MoPlvl-crit%-Cvalue-RPs-Cpts per rps*100

MoP9-49%- 0.291552 - 34 - 8575
MoP8-43%- 0.229899 - 27 - 8514
MoP7-37%- 0.174909 - 20 - 8745
MoP6-32%- 0.134001 - 15 - 8933
MoP5-27%- 0.097826 - 10 - 9782
MoP4-23%- 0.072488 - 7 - 10355
MoP3 go rank your toon you pheasant

This makes me want to compare the total damage dealt with 1000 swings including procs and crits with:

MoP9 MoA4 = 34+7 = 41pts
VS
MoP8 MoA5 Qui3 = 27+10+4 = 41pts

On two scenarios, 4 tests in total:

1) slow mh / fast oh
2) medium speed mh/oh

It's just a pity that SUM of weapon speed for 1) is never equal to SUM speed for case 2). The sum is always off by 0.1 or 0.2 total speed and you cannot get a perfect comparison.

There are also other interesting considerations to be made about the value 3.5 used to calculate proc chance (weapon proc% x weapon speed / 3.5). I.e. about where it stands in the scale fastest to slowest speed for mainhand (exactly between 4.3 and 2.7) and offhand (biased towards slow weapons) and its effect on actual mh/oh proc chance for both feather weapons and rogs.

Also, going for both slow mh/oh to maximise proc chance it's too simplistic as it doesn't take into account the big loss of dps in case of block/miss/fumble/parry/evade.

Also, for grp setup feather weapons are always superior to RoGs in terms of proc chance (18% VS 15%) however you cannot buy with feathers a 95 value lifetap that drains 60%=57hps and hits like a DD, only available to RoGs.

Therefore if you have time (yeah, sure) you could run again tests 1) and 2) to find out if a 95RoG lifetap with 15% chance and 57 drainlife is more beneficial than a 18% chance feather weapon wirh 100% returned value 60 lifedrain. Thank god all these various speed drainlife feather weapons just don't exist.

Also, this fucking 6 hours bus from Puno to Arequipa with barely no internet connection and lots of rain has turned me into a sociopath nerd. I hope the bus driver doesn't kills us all but the motherfucker is definitely making an effort.

Tue 10 Mar 2020 12:34 PM by Killaloth

gruenesschaf wrote: ↑
Mon 2 Sep 2019 9:37 AM
Crits use PRD as do procs, the failure streak counter is per weapon hand so that you don't "consume" mh crits with OH.

https://dota2.gamepedia.com/Random_distribution

Hi Gruen,

I have noticed a series of 3 casts in a row with no crit on my chanter. I am WP8 = 43% crit chance.

It's a bag system and I thought that I would always have a crit at the 3rd cast. But then I remembered your post above and went to look at the C value for 43%, which is 0.229899 found here (scroll down to answered Nov 9 '16 at 15:02).

0.229899 x 4 = 0.919596 meaning that it is indeed possible to have 4 casts in a row with no crits

However, I have been testing hundreds of casts and I never managed to land 4 casts in a row with no crit.

Coincidence? Am I doing the math right? If you confirm whether a missed cast is still included in the crit count I can run a test with 1000 casts to confirm that 4 casts in a row are not possible despite the 0.229899 C value.

Thanks!

Tue 10 Mar 2020 2:53 PM by

gruenesschaf

Killaloth wrote: ↑
Tue 10 Mar 2020 12:34 PM
Coincidence? Am I doing the math right? If you confirm whether a missed cast is still included in the crit count I can run a test with 1000 casts to confirm that 4 casts in a row are not possible despite the 0.229899 C value.

Thanks!

Resists do not count, the resist check happens first and only if it would be a hit does a crit check (and consequently increase of the failure streak in case of no crit) happen.

Wed 11 Mar 2020 8:54 AM by Killaloth

gruenesschaf wrote: ↑
Tue 10 Mar 2020 2:53 PM
Killaloth wrote: ↑
Tue 10 Mar 2020 12:34 PM
Coincidence? Am I doing the math right? If you confirm whether a missed cast is still included in the crit count I can run a test with 1000 casts to confirm that 4 casts in a row are not possible despite the 0.229899 C value.

Thanks!

Resists do not count, the resist check happens first and only if it would be a hit does a crit check (and consequently increase of the failure streak in case of no crit) happen.

Thank you, can you confirm that for a 43% crit chance it should be possible to cast 4 times in a row with no crit?

0.229899 x 4 = 0.919596

I will test more but so far I haven't seen it happening and I'd like to know if I'm looking at the correct figures. Not urgent at all tho, just curiosity.

Wed 11 Mar 2020 1:47 PM by gotwqqd

Killaloth wrote: ↑
Wed 11 Mar 2020 8:54 AM
gruenesschaf wrote: ↑
Tue 10 Mar 2020 2:53 PM
Killaloth wrote: ↑
Tue 10 Mar 2020 12:34 PM
Coincidence? Am I doing the math right? If you confirm whether a missed cast is still included in the crit count I can run a test with 1000 casts to confirm that 4 casts in a row are not possible despite the 0.229899 C value.

Thanks!

Resists do not count, the resist check happens first and only if it would be a hit does a crit check (and consequently increase of the failure streak in case of no crit) happen.

Thank you, can you confirm that for a 43% crit chance it should be possible to cast 4 times in a row with no crit?

0.229899 x 4 = 0.919596

I will test more but so far I haven't seen it happening and I'd like to know if I'm looking at the correct figures. Not urgent at all tho, just curiosity.

.57^4

Wed 11 Mar 2020 3:37 PM by Sepplord

gotwqqd wrote: ↑
Wed 11 Mar 2020 1:47 PM
.57^4

That's the theoretical chance...they aren't talking about theory, but how it practically works ingame, where different methods are used for different things to simulate RNG while avoiding long streaks that are possible in theory, but don't fit humans perception of randomness

Fri 13 Mar 2020 11:44 AM by Killaloth

gruenesschaf wrote: ↑
Tue 10 Mar 2020 2:53 PM
Killaloth wrote: ↑
Tue 10 Mar 2020 12:34 PM
Coincidence? Am I doing the math right? If you confirm whether a missed cast is still included in the crit count I can run a test with 1000 casts to confirm that 4 casts in a row are not possible despite the 0.229899 C value.

Thanks!

Resists do not count, the resist check happens first and only if it would be a hit does a crit check (and consequently increase of the failure streak in case of no crit) happen.

Hi Gruen,

I have been keeping an eye in rvr on my Enchanter crits streaks and done a lot of testing with some friends too.

I understand the concept of pseudo random distribution and something doesn't add up.

The C value for a 43% crit chance is 0.229899 but I never get 4 casts in a row w/o crits. Even by rounding up to an approximate C of 0.23 the crit chance after 4 casts should be 0.92 and although rarely, we should see 4 casts in a row w/o crits.

I just want to understand what is the real C value for a 43% (WP8) and 49% (WP9) nominal chance on Phoenix.

MAYBE with WP8 the C value on Phoenix is 0.1 (base nominal chance) + C value of 33% nominal (0.141805) = 0.241805, perhaps rounded down to 0.24 ??

Plz shed some light when u have a sec, thanks!

Sat 14 Mar 2020 6:54 PM by

gruenesschaf

Killaloth wrote: ↑
Fri 13 Mar 2020 11:44 AM
Hi Gruen,

I have been keeping an eye in rvr on my Enchanter crits streaks and done a lot of testing with some friends too.

I understand the concept of pseudo random distribution and something doesn't add up.

The C value for a 43% crit chance is 0.229899 but I never get 4 casts in a row w/o crits. Even by rounding up to an approximate C of 0.23 the crit chance after 4 casts should be 0.92 and although rarely, we should see 4 casts in a row w/o crits.

I just want to understand what is the real C value for a 43% (WP8) and 49% (WP9) nominal chance on Phoenix.

MAYBE with WP8 the C value on Phoenix is 0.1 (base nominal chance) + C value of 33% nominal (0.141805) = 0.241805, perhaps rounded down to 0.24 ??

Plz shed some light when u have a sec, thanks!

0.229899 is about right. Having 4 non crits in a row is rather rare at 43%.

Here is an example run of 1 billion tries with a chance of 43%, like a crit or proc:

Success after 0 failures: 98_856_009
Success after 1 failures: 152_257_974
Success after 2 failures: 123_375_933
Success after 3 failures: 51_045_908
Success after 4 failures: 4_463_322

That's a total of 429_999_146 successes.

Just for comparison, here is a run with straight up rng without prd / streak prevention:

Success after 0 failures: 184_896_503
Success after 1 failures: 105_394_599
Success after 2 failures: 60_071_135
Success after 3 failures: 34_237_857
Success after 4 failures: 19_516_533
Success after 5 failures: 11_128_792
Success after 6 failures: 6_347_166
Success after 7 failures: 3_613_738
Success after 8 failures: 2_059_804
Success after 9 failures: 1_172_757
Success after 10 failures: 669_228
Success after 11 failures: 381_347
Success after 12 failures: 217_600
Success after 13 failures: 124_566
Success after 14 failures: 70_759
Success after 15 failures: 39_931
Success after 16 failures: 22_925
Success after 17 failures: 12_996
Success after 18 failures: 7_437
Success after 19 failures: 4_334
Success after 20 failures: 2_427
Success after 21 failures: 1_413
Success after 22 failures: 775
Success after 23 failures: 461
Success after 24 failures: 237
Success after 25 failures: 139
Success after 26 failures: 86
Success after 27 failures: 48
Success after 28 failures: 29
Success after 29 failures: 13
Success after 30 failures: 7
Success after 31 failures: 3
Success after 32 failures: 1
Success after 33 failures: 2
Success after 34 failures: 1
Success after 35 failures: 1

Sun 15 Mar 2020 11:44 AM by Killaloth

gruenesschaf wrote: ↑
Sat 14 Mar 2020 6:54 PM
Killaloth wrote: ↑
Fri 13 Mar 2020 11:44 AM
Hi Gruen,

I have been keeping an eye in rvr on my Enchanter crits streaks and done a lot of testing with some friends too.

I understand the concept of pseudo random distribution and something doesn't add up.

The C value for a 43% crit chance is 0.229899 but I never get 4 casts in a row w/o crits. Even by rounding up to an approximate C of 0.23 the crit chance after 4 casts should be 0.92 and although rarely, we should see 4 casts in a row w/o crits.

I just want to understand what is the real C value for a 43% (WP8) and 49% (WP9) nominal chance on Phoenix.

MAYBE with WP8 the C value on Phoenix is 0.1 (base nominal chance) + C value of 33% nominal (0.141805) = 0.241805, perhaps rounded down to 0.24 ??

Plz shed some light when u have a sec, thanks!

0.229899 is about right. Having 4 non crits in a row is rather rare at 43%.

Here is an example run of 1 billion tries with a chance of 43%, like a crit or proc:

Success after 0 failures: 98_856_009
Success after 1 failures: 152_257_974
Success after 2 failures: 123_375_933
Success after 3 failures: 51_045_908
Success after 4 failures: 4_463_322

That's a total of 429_999_146 successes.

Just for comparison, here is a run with straight up rng without prd / streak prevention:

Success after 0 failures: 184_896_503
Success after 1 failures: 105_394_599
Success after 2 failures: 60_071_135
Success after 3 failures: 34_237_857
Success after 4 failures: 19_516_533
Success after 5 failures: 11_128_792
Success after 6 failures: 6_347_166
Success after 7 failures: 3_613_738
Success after 8 failures: 2_059_804
Success after 9 failures: 1_172_757
Success after 10 failures: 669_228
Success after 11 failures: 381_347
Success after 12 failures: 217_600
Success after 13 failures: 124_566
Success after 14 failures: 70_759
Success after 15 failures: 39_931
Success after 16 failures: 22_925
Success after 17 failures: 12_996
Success after 18 failures: 7_437
Success after 19 failures: 4_334
Success after 20 failures: 2_427
Success after 21 failures: 1_413
Success after 22 failures: 775
Success after 23 failures: 461
Success after 24 failures: 237
Success after 25 failures: 139
Success after 26 failures: 86
Success after 27 failures: 48
Success after 28 failures: 29
Success after 29 failures: 13
Success after 30 failures: 7
Success after 31 failures: 3
Success after 32 failures: 1
Success after 33 failures: 2
Success after 34 failures: 1
Success after 35 failures: 1

Thanks Gruen, that helps.

I am killing some coronavirus self-isolation time by creating a caster spreadsheet to find out the best RAs combination to maximise magic damage.

Currently using Wild Power, Augmented Acuity and Mastery of Magery as variables.

Can you confirm if the below figures for augm acuity damage increase are correct?

Augm Acuity Damage %
0 100%
1 100,90%
2 101,80%
3 102,71%
4 103,83%
5 104,96%
6 106,31%
7 107,67%
8 109,25%
9 110,82%

I've found other tables online with slightly different figures. If not sure I will spend some plats and do some testing myself, thanks!

Mon 16 Mar 2020 2:48 AM by bigne88

Killaloth wrote: ↑
Sun 15 Mar 2020 11:44 AM
gruenesschaf wrote: ↑
Sat 14 Mar 2020 6:54 PM
Killaloth wrote: ↑
Fri 13 Mar 2020 11:44 AM
Hi Gruen,

I have been keeping an eye in rvr on my Enchanter crits streaks and done a lot of testing with some friends too.

I understand the concept of pseudo random distribution and something doesn't add up.

The C value for a 43% crit chance is 0.229899 but I never get 4 casts in a row w/o crits. Even by rounding up to an approximate C of 0.23 the crit chance after 4 casts should be 0.92 and although rarely, we should see 4 casts in a row w/o crits.

I just want to understand what is the real C value for a 43% (WP8) and 49% (WP9) nominal chance on Phoenix.

MAYBE with WP8 the C value on Phoenix is 0.1 (base nominal chance) + C value of 33% nominal (0.141805) = 0.241805, perhaps rounded down to 0.24 ??

Plz shed some light when u have a sec, thanks!

0.229899 is about right. Having 4 non crits in a row is rather rare at 43%.

Here is an example run of 1 billion tries with a chance of 43%, like a crit or proc:

Success after 0 failures: 98_856_009
Success after 1 failures: 152_257_974
Success after 2 failures: 123_375_933
Success after 3 failures: 51_045_908
Success after 4 failures: 4_463_322

That's a total of 429_999_146 successes.

Just for comparison, here is a run with straight up rng without prd / streak prevention:

Success after 0 failures: 184_896_503
Success after 1 failures: 105_394_599
Success after 2 failures: 60_071_135
Success after 3 failures: 34_237_857
Success after 4 failures: 19_516_533
Success after 5 failures: 11_128_792
Success after 6 failures: 6_347_166
Success after 7 failures: 3_613_738
Success after 8 failures: 2_059_804
Success after 9 failures: 1_172_757
Success after 10 failures: 669_228
Success after 11 failures: 381_347
Success after 12 failures: 217_600
Success after 13 failures: 124_566
Success after 14 failures: 70_759
Success after 15 failures: 39_931
Success after 16 failures: 22_925
Success after 17 failures: 12_996
Success after 18 failures: 7_437
Success after 19 failures: 4_334
Success after 20 failures: 2_427
Success after 21 failures: 1_413
Success after 22 failures: 775
Success after 23 failures: 461
Success after 24 failures: 237
Success after 25 failures: 139
Success after 26 failures: 86
Success after 27 failures: 48
Success after 28 failures: 29
Success after 29 failures: 13
Success after 30 failures: 7
Success after 31 failures: 3
Success after 32 failures: 1
Success after 33 failures: 2
Success after 34 failures: 1
Success after 35 failures: 1

Thanks Gruen, that helps.

I am killing some coronavirus self-isolation time by creating a caster spreadsheet to find out the best RAs combination to maximise magic damage.

Currently using Wild Power, Augmented Acuity and Mastery of Magery as variables.

Can you confirm if the below figures for augm acuity damage increase are correct?

Augm Acuity Damage %
0 100%
1 100,90%
2 101,80%
3 102,71%
4 103,83%
5 104,96%
6 106,31%
7 107,67%
8 109,25%
9 110,82%

I've found other tables online with slightly different figures. If not sure I will spend some plats and do some testing myself, thanks!

Norada made testa. Mom better than acu.