[ QUOTE ]
Final to-hit : misses allowed
>.9 : 1
.8-.9 : 2
.6-.8 : 3
.4-.6 : 4
.3-.4 : 6
.2-.3 : 8
0 -.2 : 100
[/ QUOTE ]
So, accounting for the streak breaker, the chance to hit an even-level opponent opponent is going to be 75.29% rather than 75%, which does explain _Castle_'s slightly high chance to hit.
Notably, the description is quite ambiguous about what happens with multiple attack powers that have varying chances to hit. Here's a simple example:
Let's say we have an attack with a massive to-hit penalty that has a .75 chance to hit, followed by two attacks that are at the to-hit cap. Can all three attacks miss?
[ QUOTE ]
Notably, the description is quite ambiguous about what happens with multiple attack powers that have varying chances to hit.
[/ QUOTE ]
I started a thread in the AT/Powers forum that has some examples, hopefully it is more clear. Plus it's in a forum that makes more sense than this one for a topic that affects all ATs.
[ QUOTE ]
Let's say we have an attack with a massive to-hit penalty that has a .75 chance to hit, followed by two attacks that are at the to-hit cap. Can all three attacks miss?
[/ QUOTE ]
Yes. That .75 attack falls into the "3 misses allowed" category. The two following attacks are allowed to miss, no matter what their to-hit is, because there was a .75 attack in the chain. That assumes that all three fail to-hit rolls, of course - things reset as soon as one attack hits.
Skip
My Char. List and Market Transactions
HeroStats Developer
Legion of Valor
Iron Eagles
[ QUOTE ]
For instance, with a 33% chance to hit, you'd have a 64/729 chance to miss 5 times in a row, or roughly a 1/12 chance to miss 6 times in a row and automatically hit the 7th time. 1/12 is pretty significant.
[/ QUOTE ]
I think you're overstating the effect:
Let's calculate the average number of attacks to get a hit with a base to hit of 1/3 and at most 4 consecutive misses:
1/3 of the time it takes 1 attack to get 1 hit
2/9 of the time it takes 2 attacks to get 1 hit
4/27 of the time it takes 3 attacks to get 1 hit
8/81 of the time it takes 4 attacks to get 1 hit
and
16/81 of the time it takes 5 attacks to get 1 hit.
So the average number of attacks necessary is
27/81+36/81+36/81+32/81+80/81=211/81
So the modified chance to hit is 81/211 - roughly .38.
That's a difference of roughly .05 in the chance to hit, rather than .08 you had.
[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
In another redname post several months back, it was clearly explained that aim+buildup does NOT stack. IE. if aim gives 100% to accuracy and 25% to damage, and buildup gives 100% to damage and 25% to accuracy, the total would be the combined best of both abilities, IE. 100% accuracy and damage.
[/ QUOTE ]
Wrong.
[/ QUOTE ]
I'm pretty sure +100% to both was right. But instead of just flatly stating WRONG and leaving it at that, why don't you state what the correct values would be? Are you saying both Aim+BU would give +125% to dmg and acc?
[/ QUOTE ]
I think Stargazer is saying, wrong, no dev ever said that.
Build Up is +100% damage and (the generally accepted number is) +60% tohit buff. Aim is the reverse.
For blasters, scrappers, and defenders, at least. Tanker build up (and apparently stalker build up) is +80% damage and (the generally accepted number seems to be) +50% tohit buff.
And they've stacked since the beginning of time. The damage has always stacked. The accuracy has stacked in every instance I've had an opportunity to test: in particular, testing Aim + BU (verses each separately) against PFF in the arena (circa I3-I4), I can tell you that either they stack, or I had a remarkable coincidence of Aim+BU attacks hitting while Aim or BU alone were missing.
[Guide to Defense] [Scrapper Secondaries Comparison] [Archetype Popularity Analysis]
In one little corner of the universe, there's nothing more irritating than a misfile...
(Please support the best webcomic about a cosmic universal realignment by impaired angelic interference resulting in identity crisis angst. Or I release the pigmy water thieves.)
[ QUOTE ]
I have to ask. Was this done on a test server or a live one?
[/ QUOTE ]
Doesn't matter. The stuff on Test winds up on Live anyway.
So, given that the strikebreaker is in effect, an accuracy of 75% will actually hit 76.9%... which means that _Castle_'s simulated all night run was a little off for some reason (he only got 75.29%). We should be expecting upward percentage changes of the following order:
[*] 5% -> 5.9% [*]15% -> 15.7% [*]25% -> 31.4% [*]35% -> 41.4% [*]45% -> 51.6% [*]55% -> 59.5% [*]65% -> 68.6% [*]75% -> 76.9% [*]85% -> 86.0% [*]95% -> 95.2%
This is based on computational runs of 10,000,000 runs with the given odds and the strikebreaker thrown in. The simulation code basicly looked like this:
<font class="small">Code:[/color]<hr /><pre>
hit = 0
miss = 0
misscount = 0
for (i = 0; i<10000000; i++)
{
if (misscount == STREAKBREAKER)
{
misscount = 0
hit++
}
else
{
if (rand() >= HITCHANCE)
{
miss++
misscount++
}
else
hit++
}
}
print (hit / (miss + hit))
</pre><hr />
[ QUOTE ]
Second, there's a more significant implication here that may not have been considered. To illustrate, here's a phenomenon that I've experienced many times with my scrapper and stalker (especially in the low to mid levels): I spot a mob of say 10 white or yellow minions with maybe a LT or two. I jump in the middle and take few or no hits for 20 seconds or more, then all of a sudden - WAM - I suffer several hits all at the same time. This concern has been voiced by myself and others in the past, but it has generally been shrugged off as random bad luck, or biased observation. However, it now appears that what's happening is that during those first 20 seconds or so the misses are all legitimate, then suddenly the streak-breaker kicks in for all of them at the same time! This concentrates the damage into a very short window and makes it very difficult to deal with. I think this is completely unfair to defensive sets. I would think this problem would be even more apparent with Ice Tankers, who are supposed to have many minions attacking them at the same time.
[/ QUOTE ]
The streakbreaker can definitely cause this type of behavior: its been fingered as the culprit for the sudden "everyone hits at once" phenomenon in the past. Now that we have confirmed numbers for the streakbreaker and absolute confirmation as to its precise behavior, especially as it pertains to NPCs (which has always been a point of dispute on the forums despite WB's previous declaration that the streakbreaker does affect villains), this point is worth reiterating, since it can now be quantified.
Given the numbers above, does the streakbreaker "cluster" hits from multiple sources? It almost certainly does, but how much is the question: its effect on multiple attackers is very different than its (relatively) slight behavior on a single attacker. How much clustering? Well, thats a bit trickier to analyze, and I won't be in a position to simulate its effects to Monte Carlo the thing for at least a week.
Its worth noting that it isn't quite this simple: for the streakbreaker to kick in on all of the villains simultaneously, the implication is that all of them were unlucky enough to miss often enough to *force* the streakbreaker to kick in simultaneously. In other words, a really bad bad-luck streak on the villains suddenly turns into a bad luck event for us.
If this happens too many times, thats tantamount to saying something is wrong with the tohit calculator, since the NPC villains would, in essence, have something to complain about if they always needed the streakbreaker to hand them a hit.
[Guide to Defense] [Scrapper Secondaries Comparison] [Archetype Popularity Analysis]
In one little corner of the universe, there's nothing more irritating than a misfile...
(Please support the best webcomic about a cosmic universal realignment by impaired angelic interference resulting in identity crisis angst. Or I release the pigmy water thieves.)
Couple clarifications.
By critter, I mean anything run by the AI. Your pets, Skuls, exploding crates and everything in between. If it can activate a power, the streak breaker will help make it a tad more accurate.
In the note about AE attacks, I forgot to specify that the hit guarantee is only in the case that you don't have an existing miss steak with a lower to-hit. That should have been clear from the rest of the post, but I wanted to make sure no one was confused by that omission.
I did locate a small bug, which is now fixed internally. Essentially the streak breaker would not become active until you landed your first hit (so you have to do it yourself, no freebies). Pretty minor in the grand scheme of all things accuracy-related, but it's fixed now regardless. Note that this bug affected players and non-players alike.
WeirdBeard, I see one glaring flaw with the way you say this system works:
[ QUOTE ]
To determine the to-hit used in the table above, you take either the current to-hit, or the worst to-hit in your current miss series, whichever is lower.
[/ QUOTE ]
That means that if I have started a miss streak, using a power such as Aim or Build Up, or using an Accuracy inspiration will actually have no appreciable effect- (as they increase my current accuracy, but do nothing for the accuracy of previous attacks in my streak) Is that intended? Perhaps activating one of the above should force a streakbreaker to take effect. Err, in PVE only, I mean...
(yeah, that would make your next attack 'auto-hit', but heck, it's only 1 attack)
Also, whie the streakbreaker code supposedly breaks any sufficiently long miss streak, there's nothing to stop me from having a series of miss streaks punctuated by the occasional streakbreaker auto-hit.
Missing 3 times, hitting once, and then missing another 3 times is only marginally less annoying than missing all 7 times in a row. Even more annoying sometimes, considering that 1 hit is usually brawl. It's been my experience that, even unslotted, Brawl seems to have a higer overall accuracy than any other attack.
And another thing: if your streakbreaker code is based on "average" accuracy over a series of attacks, is it possible that this same sort of calculation is being used for overall accuracy? That might go a long way to explaining some of the issues we've seen brought up time and again.
I spent a few hours last night with a new character, brawling my way to lvl 3. The only power I used was brawl, and I only attacked even-con minions (with the exception of the guys I needed to get my Isolator Badge after finishing my door mission in Outbreak). Aside from a few trips to the hospital, as far as I can tell, the streakbreaker on is working, at this level. I never saw a streak of more than 3 misses, though I did have a number of the aforementioned series of 3 miss, 1 hit, 3 miss- and at least 1 of 3 miss, 1 hit, 3 miss, 1 hit, 3 miss, hospital. One thing of note I did observe: an inordinate amount of misses seemed to be my 1st attack against a given critter. (I like that word so much better than 'mob'- to me a mob has always been a group of people, not one)
Tonight I plan on switching to Jab for my trip to lvl 5, and seeing if I notice a difference.
If anyone's interested, I believe I recorded some demos, though im not 100% familiar with that process yet. (I even got the part where some guy started KSing me, so I put him on Follow and went to make a sandwich- good fun!)
I'm sorry if this has already been mentioned, but how does this apply to Defense-based characters? If a character normally has a .9 percent to hit chance, and is attacking someone with a .6 defense percentage, does the streak breaker still only allow them one miss in a row? Or does the defense directly affect the atacker's to hit number, in turn increasing the amount of misses allowed before a streakbreaker trigger? If it's the first option, then that's definitely a problem for all defense-based builds.
[ QUOTE ]
That means that if I have started a miss streak, using a power such as Aim or Build Up, or using an Accuracy inspiration will actually have no appreciable effect- (as they increase my current accuracy, but do nothing for the accuracy of previous attacks in my streak)
[/ QUOTE ]
Well, it's true that those sorts of accuracy/tohit buffs do nothing to effect the potential length of a miss streak. However, they do a great deal to increase the likelihood that the miss streak will end with a randomly (not streak breaker) generated hit. Once your to hit is improved, the likelihood of you hitting without aid of the streak breaker increases, thus decreasing the likelihood of any given miss streak continuing. Assuming the streak breaker is working bug free as described, I think that's sufficient, though of course YMMV.
Oh WeirdBeard. Why is your beard so weird?
Thanks for the explanations. Any chance of getting a look at the base to-hit progressions vs. minions? My feeling is even-con minions are working fine at the 75% chance for us to hit them, but I get the feeling that the increase/decrease vs. +/- enemies of varying levels has changed from what it was post Issue 1 "Purple Patch".
Thanks!
[ QUOTE ]
I did locate a small bug, which is now fixed internally. Essentially the streak breaker would not become active until you landed your first hit (so you have to do it yourself, no freebies). Pretty minor in the grand scheme of all things accuracy-related, but it's fixed now regardless. Note that this bug affected players and non-players alike.
[/ QUOTE ]
Ah. So that explains why I could miss my first target in Breakout six times in a row. I was thinking of complaining about that. Good to know.
[ QUOTE ]
I'm sorry if this has already been mentioned, but how does this apply to Defense-based characters? If a character normally has a .9 percent to hit chance, and is attacking someone with a .6 defense percentage, does the streak breaker still only allow them one miss in a row? Or does the defense directly affect the atacker's to hit number, in turn increasing the amount of misses allowed before a streakbreaker trigger? If it's the first option, then that's definitely a problem for all defense-based builds.
[/ QUOTE ]
It's based on the final to-hit, not the attacker's accuracy - your second option. This also means that simply slotting lots of accuracy in your attacks won't help you as far as streak-breaking is concerned - you also have to take into account the defense of your target.
Skip
My Char. List and Market Transactions
HeroStats Developer
Legion of Valor
Iron Eagles
[ QUOTE ]
I'm sorry if this has already been mentioned, but how does this apply to Defense-based characters? If a character normally has a .9 percent to hit chance, and is attacking someone with a .6 defense percentage, does the streak breaker still only allow them one miss in a row? Or does the defense directly affect the atacker's to hit number, in turn increasing the amount of misses allowed before a streakbreaker trigger? If it's the first option, then that's definitely a problem for all defense-based builds.
[/ QUOTE ]
His explanation already says it's the second option AFAICT.
A 90% to hit vs a 60% def puts us into the 30% area of the streakbreaker.
@JohnP - Victory
[ QUOTE ]
I'm sorry if this has already been mentioned, but how does this apply to Defense-based characters? If a character normally has a .9 percent to hit chance, and is attacking someone with a .6 defense percentage, does the streak breaker still only allow them one miss in a row?
[/ QUOTE ]
It's based on final ToHit
(.9 - .6) * (0 + 1) = 20-30 Streak Bracket
Phantom Rose: Ill / Kin / Psi
Soleau: Ice / Icy / Ice / Core: Ice / Fire / Pyre / Wind / Eclipse / Flare / Corona
---------------
Solo Space
Regarding the effect of the streakbreaker on accuracy:
I worked it out analytically a little while ago, and checked it against my own code. My analytical result follows:
[ QUOTE ]
x = chance to hit
y = 1- x = chance to miss
s = # of misses before streakbreaker kicks in
New accuracy, accounting for effect of streakbreaker:
x = x / (1-y^(s+1))
[/ QUOTE ]
So, for example, at 55%, you can miss up to 4 times. Then
x = 0.55, y = 0.45, s = 4, and the new accuracy is
x' = 55%/(1-0.45^5) =56.03%
At 75%, with s = 3, one finds an accuracy of
75%/(1-0.25^3) = 75.29%, which corresponds nicely with Castle's result.
These exactly match the results of my own simulations of 10,000,000 random numbers with the streakbreaker. I can't account for the discrepancy between my result and yours, except to say that I'm using the random number generator from Matlab rather than the ANSI C generator. I'm not sure why it would make sure a large difference, though.
Cheers,
Jonathan
<font class="small">Code:[/color]<hr /><pre>
hit = 0
miss = 0
misscount = 0
for (i = 0; i<10000000; i++)
{
if (misscount == STREAKBREAKER)
{
misscount = 0
hit++
}
else
{
if (rand() >= HITCHANCE)
{
miss++
misscount++
}
else
{
hit++
misscount = 0
}
}
}
print (hit / (miss + hit))
</pre><hr />
d'oh! bug in my simulator. Marked above in red. Simulator now agrees with your findings (75.27%). *shame*
[ QUOTE ]
Okay now based off that chart
How can I make a NEW character, run up to PRISONERS in breakout and miss 10 times in a row?
Explain that one please. If you need a fraps recording I will be HAPPY to provide this proof.
[/ QUOTE ]
A video would be great.
I'm willing to dismiss this as perceptions not matching reality untill someone gathers something more concrete, such as a video.
what are HITCHANCE and STREAKBREAKER in your code?
Ok, given the explanation by Weird, let me ask the following question:
The table provided says that the streakbreaker count for accuracy 5-10% is 100, and Wierdbeard (I think) said that the count used the worst accuracy for the streak. Does this, then, mean that if *at any point*, I make an attack attempt against such a hard to hit target, then my streakbreaker value may stay at 100 for up to the next 100 attacks, assuming I miss?
So let's use a completely hypothical example. Let's say I'm a level 28 hero working on Croatoa missions set to heroic. Lets say (again, purely hypothetical) that the mission is populated by a mix of level 28 foes with level 43 pets, say. If I swing at that pet, my streakbreaker will become 100 (a) until I hit, or (b) until 100 attacks occur.
Another example. Let's say I'm fighting Paragon Protectors. If one of them MOGs, does the same thing occur?
[ QUOTE ]
Its worth noting that it isn't quite this simple: for the streakbreaker to kick in on all of the villains simultaneously, the implication is that all of them were unlucky enough to miss often enough to *force* the streakbreaker to kick in simultaneously. In other words, a really bad bad-luck streak on the villains suddenly turns into a bad luck event for us.
[/ QUOTE ]
Well, I don't think it would have to be all of them for a noticeable effect. For example, let's say my SR scrapper is surrounded by 10 identical even-con minions. Let's also say that my scrapper's melee defense is enough to lower their melee attacks to 20% to-hit, and that they are using only melee attacks. I'm not sure what the odds are on missing 8 times in a row with a 20% chance to-hit, but it can't be that low. So I don't think it's that unlikely that after 8 rounds (assuming one attack per minon per round) of attacks, that maybe only 5 of the minions hit me during that time. Now comes the 9th round and suddenly 5 of the minions (the 5 that haven't hit me yet) are all going to hit me at the same time. Statistically, one of the 5 that already hit me, will hit me again this round, too. So that's at least 6 roughly simultaneous hits coming at me. That's a lot. Without the streak-breaker I should only expect to be hit twice per round.
If someone can quickly compute the odds on 5 out of 10 minions, with a 20% chance to hit, missing 8 times in a row, and also the odds of 6 of those minions hitting at the same time normally. It might be interesting to compare the numbers.
However, you are correct in that some simulations need to be run to determine the real effect. I'll see if I can work on that.
"I like that word so much better than 'mob'- to me a mob has always been a group of people, not one."
Just for general clarification, the term "MOB" (or "mob") in game design is an acronym for "Mobile OBject" rather than a reference to the english term for a throng or disorderly crowd. Early games had more of a programming split between static objects (like walls and things you don't want to run into) and mobile objects (like space invaders, berzerk robots, etc). Coming from a pen and paper gaming background, I've always preferred the term "NPC" for "critters".
"I'm not sure what the odds are on missing 8 times in a row with a 20% chance to-hit, but it can't be that low."
I believe the odds are 0.1677, or just around one in six, so no, it's not that low.
The odds of six 20% minions all hitting you with their next attack is one in 15,625. (0.000064)
I'm not sure what this all means for you, aside from that you need to kill minions faster. 
The odds of missing twice in a row with a 90% chance to-hit, by contrast, is .01, or one in one hundred.
[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
In another redname post several months back, it was clearly explained that aim+buildup does NOT stack. IE. if aim gives 100% to accuracy and 25% to damage, and buildup gives 100% to damage and 25% to accuracy, the total would be the combined best of both abilities, IE. 100% accuracy and damage.
[/ QUOTE ]
Wrong.
[/ QUOTE ]
I'm pretty sure +100% to both was right. But instead of just flatly stating WRONG and leaving it at that, why don't you state what the correct values would be? Are you saying both Aim+BU would give +125% to dmg and acc?
[/ QUOTE ]
The non-stacking part is wrong. Aim and BU stack just fine.
BU has a 100% Dmg buff, and Aim 60-ish %. It's often said that the tohit buffs are 60/100%, but I can't verify that to a large degree of certainty. It could be 60/100, or it could be something else.
With Aim+BU you have a 160-ish % dmg buff.