Guild Wars 2

http://gw2skills.net/editor/?vdAQFASncoClYh9ZBubB0ehlRk67UWlWBsWulAg+LzywE-ThRBABXt/o87hLAAlq/YmSQ80BAwDAAA-e

This is the build I use.
I have tested it on the DPS Golem multiple times.
On my initial burst I can get up to 30k dps, but after that I stay between 19k and 23k dps.

Sequence is:
Tool Belt skills 2,3 and 5>Hammer 2-> Hammer 4> Hammer 5> Grenades 2>Grenades 4> Grenades 5>
Bombs skill 1 about 7-10 times> Elixir Gun 4 ( let it blast back)>
Hammer 3 to get back in to fight> Start sequence over

You can use food/utility buffs to boost the dps a little bit also, but without them, I still do plenty of damage, and I love the build so much.

For Explosives Trait, I’m not sure if Grandmaster trait should be Shrapnel or if it should stay on Siege Rounds.

Please post any suggestions.

Lots’o skills there. I main an engineer myself, so I get where you’re coming from.

But as a raid leader, tell me, why should I bring a power engineer to my raid averaging on (at best) 26k dps, when I can bring a thief (32k for me), an elementalist (31k staff, 32k DW for me), or a guardian (27k dps for me)? All of which also brings other things to the raid, that the engineer does not.

I’m not saying you should, and I’m not saying the Engineer is any better then the other classes. I’m just making a point that the Engineer is still a very viable class to play and is not near as bad a most people think.

Allie.4925:

Lots’o skills there. I main an engineer myself, so I get where you’re coming from.

But as a raid leader, tell me, why should I bring a power engineer to my raid averaging on (at best) 26k dps, when I can bring a thief (32k for me), an elementalist (31k staff, 32k DW for me), or a guardian (27k dps for me)? All of which also brings other things to the raid, that the engineer does not.

Last I saw under similar sitautions Thief was ~29-30k, Power Engi ~25-26k, and D/W Ele 26-27k. The thing I don’t remember is the Alacrity on the Engi or not as the numbers on D/W rise with Alacrity but thief doesn’t. So not really all too bad. Why bring one over thief though? No reason, but it’s not bad at all.

I really think DPS is NOT the problem…

Use Siege Rounds.

Not sure what buffs you’re using in your tests, but if it’s all buffs, you’re not doing much better than a PS war.

https://www.reddit.com/r/Guildwars2/comments/4v76gq/qt_updated_guides_and_dps_benchmarks_for_all/

Heya!

Eragon.8234:

http://gw2skills.net/editor/?vdAQFASncoClYh9ZBubB0ehlRk67UWlWBsWulAg+LzywE-ThRBABXt/o87hLAAlq/YmSQ80BAwDAAA-e

This is the build I use.
I have tested it on the DPS Golem multiple times.
On my initial burst I can get up to 30k dps, but after that I stay between 19k and 23k dps.

I’m afraid that’s not good at all. :/

Eragon.8234:

Sequence is:
Tool Belt skills 2,3 and 5>Hammer 2-> Hammer 4> Hammer 5> Grenades 2>Grenades 4> Grenades 5>
Bombs skill 1 about 7-10 times> Elixir Gun 4 ( let it blast back)>
Hammer 3 to get back in to fight> Start sequence over

Hammer #4 is no longer a dps increase, yet bomb #2 is one. You should also interrupt your EG #4 and never use hammer #3, wich is also a dps loss, unless for combos.

You should check out this: https://www.reddit.com/r/Guildwars2/comments/4vzplt/engineerscrapper_updated_skill_priority_list_for/

Eragon.8234:

For Explosives Trait, I’m not sure if Grandmaster trait should be Shrapnel or if it should stay on Siege Rounds.

DGraves.3720:

Use Siege Rounds.

Shrapnel increases your dps by more than Siege Rounds ever will. You’ll have over 1k condi damage as a full zerker Engi in raids. Siege Rounds only allows you to use Orbital as a minor dps increase over auto attacks.

If you want the utility from Siege Rounds tough, just go for it.

Greez!
- Ziggy

Thank you Dahkeus for the link. And thank you Xyonon, you have helped me a lot.

Xyonon.3987:

Heya!

Shrapnel increases your dps by more than Siege Rounds ever will. You’ll have over 1k condi damage as a full zerker Engi in raids. Siege Rounds only allows you to use Orbital as a minor dps increase over auto attacks.

If you want the utility from Siege Rounds tough, just go for it.

---

Greez!
- Ziggy

=“Sequence”]Tool Belt skills 2,3 and 5>Hammer 2→ Hammer 4> Hammer 5> Grenades 2>Grenades 4> Grenades 5>
Bombs skill 1 about 7-10 times> Elixir Gun 4 ( let it blast back)>
Hammer 3 to get back in to fight> Start sequence over.

Even at 100% critical chance Shrapnel only gets 19 chances at most per cycle to activate in this sequence. Siege Rounds is predictable and jives better with the power layout plus it’s a toolbelt skill so it’s CD is only 34s I believe. It’s a strong AoE. Most standard sequences have greater odds iirc since it is the number of explosions that dictates the value of Shrapnel.

If we presumed even a true lucky streak of 25% that’s only 4 activations on average per cycle. At just 1,000 condition damage over the native 12s duration that’s 984 damage total per bleed cycle totaling a whole 3,936 after all was said and done per cycle. I think even one Orbital Strike critically hitting will top that. If we presumed two cycles per Orbital Strike CD that’s still only 7,872 damage overall.

I just hopped on and went to a PvP golem because i was in the PvP arena on my engi and set up just to use Orbital Strike, I averaged 3.55k in PvP Zerker w/ Scholar. In PvE obviously it’s going to be significantly greater and we’re not including things like might and vulnerability or additional damage through group effects so I’m not seeing how, even after giving the benefit of the doubt and a significant increase to the real occurrence rate, there’s going to be more damage.

But that’s just based on this rotation.

Heya!

That’s not really true – you could have auto attacked instead.

Siege Rounds doesn’t automaticly increase your dps by a two Orbital Strikes. You have to subtract the auto attacks you didn’t do aswell.

Let’s assume realistic raid buffs (qT standards):

Orbital Strike deals untraited 13,844 damage, traited 27,688 (2x). This with a cast time of 0.87s (quickness).
Your bomb aa deals 12,025 damage. With quickness, every 0.56s. In 0.87s you would deal 18,682 damage.

27,688 – 18,682 = 9,006. The cooldown is 34.87s (cooldowns start after the cast) or 26.44s with Alacrity.

Traiting Orbital strike truly increases your DpS by only 258-341.

What about Shrapnel? 1148 Condi Damage, 2 stacks GotL, 25 stacks Vuln, (wich all is included in the OS calculation aswell) makes each Shrapnel proc deal 1,923 damage, or increases each bomb attack by 289. 289 divided by 0.56 = 532 DpS.

Traiting Shrapnel increases your DpS by 532 if you’d only do bomb auto attacks.

What else do you use that procs Shrapnel multiple times?

• Shrapnel Grenade (3 times)
• Poison Grenade (3 times)
• Freeze Grenade (3 times)
• Fire Bomb (4 times)

If we are talking about DpS only, Shrapnel will always be better unless you bash some trash (badumtss).

However I’d drop it immediately if you can make use of the combo fields since the DpS differences aren’t really noticable.

Greez!
- Ziggy

That’s the main benefit to siege rounds to me, the ability to double blast a field and get two combos out of it. Not sure how often the chance to do this comes up, but Fire Bomb and two orbital strikes is 6 might stacks for everyone. Depends on your party composition on whether you can leave that sort of thing to others, or if it would be useful (while clearing away some trash).

Xyonon.3987:

Heya!

That’s not really true – you could have auto attacked instead.

Siege Rounds doesn’t automaticly increase your dps by a two Orbital Strikes. You have to subtract the auto attacks you didn’t do aswell.

---

Let’s assume realistic raid buffs (qT standards):

Orbital Strike deals untraited 13,844 damage, traited 27,688 (2x). This with a cast time of 0.87s (quickness).
Your bomb aa deals 12,025 damage. With quickness, every 0.56s. In 0.87s you would deal 18,682 damage.

27,688 – 18,682 = 9,006. The cooldown is 34.87s (cooldowns start after the cast) or 26.44s with Alacrity.

Traiting Orbital strike truly increases your DpS by only 258-341.

What about Shrapnel? 1148 Condi Damage, 2 stacks GotL, 25 stacks Vuln, (wich all is included in the OS calculation aswell) makes each Shrapnel proc deal 1,923 damage, or increases each bomb attack by 289. 289 divided by 0.56 = 532 DpS.

Traiting Shrapnel increases your DpS by 532 if you’d only do bomb auto attacks.

What else do you use that procs Shrapnel multiple times?

• Shrapnel Grenade (3 times)
• Poison Grenade (3 times)
• Freeze Grenade (3 times)
• Fire Bomb (4 times)

---

If we are talking about DpS only, Shrapnel will always be better unless you bash some trash (badumtss).

However I’d drop it immediately if you can make use of the combo fields since the DpS differences aren’t really noticable.

---

Greez!
- Ziggy

+ SD on the orbital, and Grenade Barrage as well for Shrapnel

The main strengths of the rifle meta build.

That is to say : Explosions(3,2,2),Firearms(3,1,2),Tools(1,3,2), Rifle -Healing Turret, Nades/Bomb, Battering Ram, Rifleturret, Mortar Kit.

The main strengths of this build are
1) Decent dps, right under Ele and Thief.
2) Super high breakbar power. (Rifle 4, Big Bomb, Battering Ram, Destroy Healing Turret, Destroy Rifle Turret, Launch Battering Ram).
3) Mobility, Rifle and enhanced endurance regen allow you to travel really quickly.
4) Group Healing from HT, Mortar 5, + 2-3 blast finishers.

The problem is though, PS Warrior and Rev bring something unique to the table and together can adequately do BB Damage. (Reason #2 becomes obsolete)
(Reason 3 is only good on VG, at which point you might as well fill the spot with a condi engi)
(Reason 4, a druid healer usually suffices making this obsolete)
(Reason 1, it’s better to take a thief/ele seeing as they are higher DPS)

For every reason there is to take an Engi, there is a reason to take something else.

Our biggest hope, at all, to get a perma raid spot, is that Anet makes every raid encounter require condi to some degree, and that Anet ends up nerfing epidemic on Necros. If that happens, then we out condi necros by a good 6k damage and can take over the condi spots.

Jerus.4350:

+ SD on the orbital, and Grenade Barrage as well for Shrapnel

The OP build’s playing Scrapper.

@Ging:

The one big selling point the Engineer has is blind and cripple/immob/chill/control. Unfortunately this isn’t enough. :/

I have looked at the build I was using, and also looked at the SD build, and I have to say, the SD build is quite fun so far.

Xyonon.3987:

Heya!

That’s not really true – you could have auto attacked instead.

Siege Rounds doesn’t automaticly increase your dps by a two Orbital Strikes. You have to subtract the auto attacks you didn’t do aswell.

---

Let’s assume realistic raid buffs (qT standards):

Orbital Strike deals untraited 13,844 damage, traited 27,688 (2x). This with a cast time of 0.87s (quickness).
Your bomb aa deals 12,025 damage. With quickness, every 0.56s. In 0.87s you would deal 18,682 damage.

27,688 – 18,682 = 9,006. The cooldown is 34.87s (cooldowns start after the cast) or 26.44s with Alacrity.

Traiting Orbital strike truly increases your DpS by only 258-341.

What about Shrapnel? 1148 Condi Damage, 2 stacks GotL, 25 stacks Vuln, (wich all is included in the OS calculation aswell) makes each Shrapnel proc deal 1,923 damage, or increases each bomb attack by 289. 289 divided by 0.56 = 532 DpS.

Traiting Shrapnel increases your DpS by 532 if you’d only do bomb auto attacks.

What else do you use that procs Shrapnel multiple times?

• Shrapnel Grenade (3 times)
• Poison Grenade (3 times)
• Freeze Grenade (3 times)
• Fire Bomb (4 times)

---

If we are talking about DpS only, Shrapnel will always be better unless you bash some trash (badumtss).

However I’d drop it immediately if you can make use of the combo fields since the DpS differences aren’t really noticable.

---

Greez!
- Ziggy

But his rotation only has 19 total explosions in it (opportunities). That’s all. I’m looking at his rotation in totality and of those explosions “7-10” are really just bomb aa’s. 3 per grenade salvo and 4 for firebomb would range 19-23 if I counted correctly.

As for damage lost during the strike I need to check on that. I think you actually can attack during that two second charge meaning that the “loss” is really the difference between the AA (varies) and the First Orbital Strike. The second is not effected by whatever you’re doing so it is actually free damage from my time just experimenting with it.

So met gain should be equal to ( Orbital 1 – AA Potential ) + Orbital 2. Parenthesis for emphasis.

Also you can’t (shouldn’t) use partial attacks. In .87s you’d only get one bomb blast. This matters because it greatly effects how rotations are set up based on concurrent damage ( for instance starting orbital then dropping a bomb will yield much higher damage than dropping two bombs and skipping orbital because of “mathematical averaging” instead of real-time behavior ) and the like.

I’m so boring!?

http://gw2skills.net/editor/?vdAQFAUlUUh+sYNVwELQ7FLvFFYH+hp14acjZx3DOhBAA-ThRBABUq+jZKBva/R53DXAAAeAA4pDgUADMsC-e

This is the build I’m using now.
Ging showed using battering ram also. If I use battering ram, should I get rid of bombs, or nades. I’ve been using the nades only for skills 2/4/5 and the toolbelt skill, and just spamming bomb skill 1 when not using the toolbelt skills to proc SD. If I use battering ram, does bombs or nades have better damage? Or should I just stay with the build I have now?(keep in mind the shrapnel trait plays a good part in this)

Bombs will produce more damage using that build.

DGraves.3720:

So met gain should be equal to ( Orbital 1 – AA Potential ) + Orbital 2. Parenthesis for emphasis.

That’s what I did hon’.

DGraves.3720:

Also you can’t (shouldn’t) use partial attacks. In .87s you’d only get one bomb blast. This matters because it greatly effects how rotations are set up based on concurrent damage ( for instance starting orbital then dropping a bomb will yield much higher damage than dropping two bombs and skipping orbital because of “mathematical averaging” instead of real-time behavior ) and the like.

That’s not true, no. I believe you are aware that there is no real “rotation” for the engineer, rather a skill priority list in your mind. So you won’t “mess up” anything. Your main damage skills should always be on cd when they are up, regardless of using OS or bombs. The only thing that happens is that you may or may not shift this one bomb aa to a later moment. It matters after the x-th time you’ve used bomb aa instead of OS.

Eragon.8234:

does bombs or nades have better damage? Or should I just stay with the build I have now?

Take a look at the image attached.

Skill priority ( opportunity cost ) just uses highest damage per time slot. In .87 seconds you can drop one bomb for let’s say 14k damage. Or you can cast traited Orbital Strike for 27k.

The DPS has no value in opportunity cost prioritization.

Speaking of which looking at the 19 explosions let’s say one full rotation (he posted a rotation) takes 15s. So 4 in a minute. 19 × 4 = 76 opportunities. 15% of these will produce a proc, 11.4, so we round down to a real 11.

Taking it raw 11 times, 15.96s (he’s in seeker with no additional expertise but has Firearms) a bleed, and for giggles I’ll give 1800 condition damage (much more than what he’d have even effectively using buffs from any sources:

A second of bleed is worth 22 + .06 × 1800 = 130.

A single Shrapnel is worth 130 × 15.96s = 2074.8

11 of these is worth 2074.8 × 11= 22,822.8

And divide that by 60 for 380.38 dps.

-shrug-

If we just take your 13.5k per orbital strike and multiply that by 2 we get 27k. But under alacrity the 40s ’s becomes 26.4 seconds meaning it goes off twice a minute.

That 27k becomes 54k. 54,000 / 60 = 900 dps.

So in this custom made situation he doesn’t have enough explosions or enough effective condition damage to hit 532 with shrapnel.

Thank you Xyonon, from the information I gathered from the image you attached, and please correct me if i’m wrong, I should stick with build I have now, with both bombs and grenades.

Very well done with the spreadsheet by the way.

Yes, Nades, Bombs and the Rifle Turret for it’s toolbelt are currently max dps.

If we talk about a raid situation (realistic) I suggest you act like this:

Start with Shrapnel Nades and Blunderbuss, then use all your strong skills like Barrage and Big Ol’ Bomb (and Jump Shot if you don’t have Quickness). Now you swap to the bomb kit, use Fire Bomb and then Bomb aa.
Careful with the Fire Bomb, since it will cancel the cast of auto attacks. So best to use it at the very beginning of swapping to the Bomb Kit.

Shrapnel Nades are a massive DpS increase, yet you can’t swap to the Nade Kit and then back to the Bomb Kit immediately again (cd). Assuming no Alacrity, count to 6 in your head, then swap to the Nade kit and use Shrapnel, then swap to the rifle and use Blunderbuss, then back to Bomb Kit. Next time use Shrapnel → Poison Nades. Then Shrapnel → Blunderbuss. Then Shrapnel → Freeze Nades. Then Shrapnel Blunderbuss and so on. I think you get the idea. It’s about always being able to use something that is better than the Bomb aa together with Shrapnel Nades, to not lose DpS.

Oh yea, and always use the Rifle Turrets toolbelt (Surprise Shot) on cd!

Greez!

Awesome! Thank you VERY much for your help.

So basically OP is scraping their posted rotation right? Because this has nothing to do with that rotation. As a master of fact OP now uses rifle instead of hammer.

Bypass answering questions: Tell others how to play.

@DGraves
When I first posted this, I was using Scrapper, now I’m not. My original post was to prove the Engineer was still great, like the headline says, and I linked the build I used with information to prove it, and the others that have posted were making comments on how to improve the build I was using, and the build I’m using now. This forum was not originally meant to find the best build, it was meant to prove the Engineer is still great.

Very well. What a bummer. I miss the days when we actually explored builds instead of just talking about DPS but times have changed. So should I.

Maybe I should have explained this better, I am sorry for not doing so. Most people say that the Engineer sucks when I comes to dps, this forum was to prove that they were wrong.
There is nothing wrong with playing your own build, and exploring other builds, and plenty of people still do so. I have tried creating multiple kinds of builds for my Engineer, some have worked out, some haven’t. You shouldn’t change just because others have or are changing, you should change because you want too.
You should play what u find most fun, not what others do, and the dps builds people want nowadays … well let me quote Nemesis> “There is no best, there is only best at…”, so the dps builds people want nowadays are not the best builds out there.

My concern is with people becoming so incredibly theoretical that they lose touch with the game itself. The DPS arms race is the worst thing to happen to gaming overall simply because it creates a false standard of superiority through numerical mastery.

The only problem is that numbers are so easy to manipulate and all it takes is a “solid reputation” and you can tell people anything and create an equation to equal whatever you want. I just sit quietly and watch most of the numbers I see are either flat-out wrong or completely theoretical and ignoring many realities that you’ll face in encounters rather regularly.

Not that I would be against GW2 being played on a spreadsheet. I am good with spreadsheets.

Hmm, I have to agree with you on the first part.
The second part… can you explain that a little more, it has got me a little curious, and I would like to talk a little more about this.

Elementalist has the highest DPS in the game right? Or potential highest etc. People used to believe they were utter trash. Then you had people with a little fame coming out with numbers that were impossibly high DPS before ascended ever hit the scene and others believing they did insane DPS. Then you had people believing that they could optimize a super rotation when, in fact, even within the same quarter ( meaning before any quarterly updates were applied ) they were shown to be wrong.

Things noticed in year one like how you don’t have 100% damage uptime and the like were blatantly ignored, mechanics of the game were declared unsound, random spreadsheets were, are, and probably will forever be “the answer” to questions that really require more than a spreadsheet to answer in any given situation and you often have bad metrics to boot.

For instance, Shrapnel, if you throw one volley of grenades what is the probably that two of those grenades will proc Shrapnel? Simple probability:

.15^2 = .0225

All three? .15^3 = .003375.

Basically it’s so rare for it to happen it’s not even worth discussing as a possibility and it is incredibly unpredictable. But people do it. People somehow “roll” this into their calculations. I’ve seen too many times where people take and mix principles that don’t belong or account for things wrong or ignore basic pragmatism for theoretical propositions. They overcomplicate it and make claims that don’t even make sense about their skills and abilities for things they actually can’t calculate or, if they do, don’t come out anywhere near what they say it should using commonality.

But what makes this possible is simply reputation. We will believe anyone not based on the merit of their claim but based on the warmth of their ideas. An easy to read spreadsheet, even if wrong, is the best way to win fans. Show a video of you doing something people consider difficult or some crazy large numbers and you’ve got people hooked. After that you can go the way of L. Pauling (the source Vitamin C myth) and just say whatever the heck you want and people will hang on your every word.

DGraves.3720:

Elementalist has the highest DPS in the game right? Or potential highest etc. People used to believe they were utter trash. Then you had people with a little fame coming out with numbers that were impossibly high DPS before ascended ever hit the scene and others believing they did insane DPS. Then you had people believing that they could optimize a super rotation when, in fact, even within the same quarter ( meaning before any quarterly updates were applied ) they were shown to be wrong.

Things noticed in year one like how you don’t have 100% damage uptime and the like were blatantly ignored, mechanics of the game were declared unsound, random spreadsheets were, are, and probably will forever be “the answer” to questions that really require more than a spreadsheet to answer in any given situation and you often have bad metrics to boot.

For instance, Shrapnel, if you throw one volley of grenades what is the probably that two of those grenades will proc Shrapnel? Simple probability:

.15^2 = .0225

All three? .15^3 = .003375.

Basically it’s so rare for it to happen it’s not even worth discussing as a possibility and it is incredibly unpredictable. But people do it. People somehow “roll” this into their calculations. I’ve seen too many times where people take and mix principles that don’t belong or account for things wrong or ignore basic pragmatism for theoretical propositions. They overcomplicate it and make claims that don’t even make sense about their skills and abilities for things they actually can’t calculate or, if they do, don’t come out anywhere near what they say it should using commonality.

But what makes this possible is simply reputation. We will believe anyone not based on the merit of their claim but based on the warmth of their ideas. An easy to read spreadsheet, even if wrong, is the best way to win fans. Show a video of you doing something people consider difficult or some crazy large numbers and you’ve got people hooked. After that you can go the way of L. Pauling (the source Vitamin C myth) and just say whatever the heck you want and people will hang on your every word.

The chance that at least one grenade in a group of 3 will proc shrapnel is 38.5%. The chance that exactly 2 will is 5.74%.

It’s not 0.15^2, it’s 3c2*0.15^2*0.85 = 3*0.15^2*0.85.

Ging.6485:

The chance that at least one grenade in a group of 3 will proc shrapnel is 38.5%. The chance that exactly 2 will is 5.74%.

It’s not 0.15^2, it’s 3c2*0.15^2*0.85 = 3*0.15^2*0.85.

It should be just 45% since they are independent events for “at least one”.

Basically it runs like any other form of basic dice game; if you have two twenty sided dice the odds are additive for rolling a twenty on either and multiplicative for rolling a twenty on both, I.E.

5% + 5% = 10%

5% x 5% = .25%

In fraction form:

1/20 + 1/20 = 2/20 = 1/10 = 10/100 = 10%

1/20 × 1/20 = 1/400 = .25%

I’d be lying to say I knew where you gathered your numbers from. Combinatorics doesn’t really support the idea as far as I know and we’ve no evidence that isn’t a fair throw though it is possible that you are correct via the algorithmic nature itself and maybe you know something I don’t about the algorithm but that wouldn’t explain multiplying in the failure rate.

DGraves.3720:

Ging.6485:

The chance that at least one grenade in a group of 3 will proc shrapnel is 38.5%. The chance that exactly 2 will is 5.74%.

It’s not 0.15^2, it’s 3c2*0.15^2*0.85 = 3*0.15^2*0.85.

It should be just 45% since they are independent events for “at least one”.

Basically it runs like any other form of basic dice game; if you have two twenty sided dice the odds are additive for rolling a twenty on either and multiplicative for rolling a twenty on both, I.E.

5% + 5% = 10%

5% x 5% = .25%

In fraction form:

1/20 + 1/20 = 2/20 = 1/10 = 10/100 = 10%

1/20 × 1/20 = 1/400 = .25%

I’d be lying to say I knew where you gathered your numbers from. Combinatorics doesn’t really support the idea as far as I know and we’ve no evidence that isn’t a fair throw though it is possible that you are correct via the algorithmic nature itself and maybe you know something I don’t about the algorithm but that wouldn’t explain multiplying in the failure rate.

Well the chance of all three grenades failing to proc would be 0.85^3. 1 minus that number is the chance that no failure occurs, meaning that at least one succeeds. I did it in excel and am writing from my phone right now, but the chance for a complete failure was 61% and change.

The chance for exactly two to proc shrapnel is 0.15*0.15*0.85, the chance of 2 successes and one failure. There are three ways that could happen. 12 23 13. that number comes out to 0.05735 or something close to that.

The point is though, its about a 40% chance that you will get at least one 32 second bleed worth 4100 damage every time you throw a grenade. Each grenade throw is .5 seconds, meaning that in the time it takes you to cast Orbital Strike once, you have about 240 chances to proc shrapnel.

Ging.6485:

Well the chance of all three grenades failing to proc would be 0.85^3. 1 minus that number is the chance that no failure occurs, meaning that at least one succeeds. I did it in excel and am writing from my phone right now, but the chance for a complete failure was 61% and change.

That isn’t true.

Let’s take two six sided dice. We know that the probability of getting two sixes (or two of any number really) is 1/36. We know it’s 1/6 × 1/6 right?

So if that’s the case using your logic 5/6 × 5/6 or 5/6^2 would be the direct opposite, right? Well, let’s do it: 5/6^2 = .69444 or 69.44%. So using your logic if we take 100% and subtract 69.4% there’s a 30.56% of getting two sixes?

We know that is wrong. So why is it wrong?

Actually it is because you can’t replace the odds of something not happening with the odds of something happening when setting up probability matrices. You should always take the odds of the event occurring and find the values you’re looking for, for instance:

1 – .15 = .85. Easy.
1 – (.15 × 2 ) = 1 – .30 = .70. Easy.
1 – (.15 × 3) = 1 – .45 = .55. Easy.

So there is kitten percent chance that none of the grenades in a salvo will produce Shrapnel. It’s that simple. What I want you to note is that the final results of those problems changes but the actual probability of the event does not. It’s still 15% in all three problems but 85% is not the unified odds of failure, and in fact, changes the more iterations you have.

This makes sense if you think about it. Again with six sided dice:

If I roll 1 the odds are 1/6 that I’ll get (number)
If I roll 2 the odds are 1/6 × 1/6 that I’ll get (number combination)
If I roll 3 the odds are 1/6 × 1/6 × 1/6 that I’ll get (number combination)

But note that 1/6th doesn’t change. I guarantee you the result does:

1/6
1/36
1/216

Now if you took 5/6 and tried to apply it to this you would get:

5/6
25/36
125/216

That gap is worsening by the way.

I feel so boring!

The chance for exactly two to proc shrapnel is 0.15*0.15*0.85, the chance of 2 successes and one failure. There are three ways that could happen. 12 23 13. that number comes out to 0.05735 or something close to that.

Alongside the above this is wrong for two distinct reasons; first you never ever double-attribute a behavior. This means that if you need only to find the probability of an event the size of the pool of events doesn’t matter. Let’s say that instead of throwing 3 grenades we throw 4 at the time. Does that mean it becomes .15 × .15 × .85 x .85? If so, how does that look?

Well combining those four numbers produces the value 1.6% but if it were 5 grenades looking for the same thing …

That value becomes 1.4% ( .15 × .15 × .85 x .85 x.85 ).

It’s going the wrong way. Your odds should increase not decrease so what is happening? You’re not checking to see if two will succeed you are finding the odds of two succeeding and specifically one failing, or defining the odds by specifically searching out the distinct behavior of two unified events.

The point is though, its about a 40% chance that you will get at least one 32 second bleed worth 4100 damage every time you throw a grenade. Each grenade throw is .5 seconds, meaning that in the time it takes you to cast Orbital Strike once, you have about 240 chances to proc shrapnel.

How are you getting 32s? First, the base for Shrapnel is 12s, when you take Firearms trait line you are forced to take a trait that makes your bleeds last 33% longer, so it’s 16s after that trait is forcefully applied not before. So you’ll cap out at 24s.

That said if every grenade throw is .5s and you get three at a time it looks like this:

60 / .5 = 120

120 × 3 = 360

So it’s 360 not 240 presuming continuous attack.

So 15% of 360 would be 54 activations? That’s amazingly high, no doubt, but in the rotation we were using bombs, not grenades, because the answer is completely different when you use grenades over bombs due to more opportunities. Let me think, uh, 360 – 107 (number of bombs in one minute under 100% quickness) = 253 additional chances with grenades?

I feel so old!

Oh, and let me correct one more thing, the cast time of orbital strike would be .87s so you wouldn’t gain 240 opportunities in that time, and in fact, lose a grand total of six, or two salvos altogether, because of the CD itself. That was spun way out of control there.

DGraves.3720:

Ging.6485:

Well the chance of all three grenades failing to proc would be 0.85^3. 1 minus that number is the chance that no failure occurs, meaning that at least one succeeds. I did it in excel and am writing from my phone right now, but the chance for a complete failure was 61% and change.

That isn’t true.

Let’s take two six sided dice. We know that the probability of getting two sixes (or two of any number really) is 1/36. We know it’s 1/6 × 1/6 right?

So if that’s the case using your logic 5/6 × 5/6 or 5/6^2 would be the direct opposite, right? Well, let’s do it: 5/6^2 = .69444 or 69.44%. So using your logic if we take 100% and subtract 69.4% there’s a 30.56% of getting two sixes?

We know that is wrong. So why is it wrong?

Actually it is because you can’t replace the odds of something not happening with the odds of something happening when setting up probability matrices. You should always take the odds of the event occurring and find the values you’re looking for, for instance:

1 – .15 = .85. Easy.
1 – (.15 × 2 ) = 1 – .30 = .70. Easy.
1 – (.15 × 3) = 1 – .45 = .55. Easy.

So there is kitten percent chance that none of the grenades in a salvo will produce Shrapnel. It’s that simple. What I want you to note is that the final results of those problems changes but the actual probability of the event does not. It’s still 15% in all three problems but 85% is not the unified odds of failure, and in fact, changes the more iterations you have.

This makes sense if you think about it. Again with six sided dice:

If I roll 1 the odds are 1/6 that I’ll get (number)
If I roll 2 the odds are 1/6 × 1/6 that I’ll get (number combination)
If I roll 3 the odds are 1/6 × 1/6 × 1/6 that I’ll get (number combination)

But note that 1/6th doesn’t change. I guarantee you the result does:

1/6
1/36
1/216

Now if you took 5/6 and tried to apply it to this you would get:

5/6
25/36
125/216

That gap is worsening by the way.

I feel so boring!

The chance for exactly two to proc shrapnel is 0.15*0.15*0.85, the chance of 2 successes and one failure. There are three ways that could happen. 12 23 13. that number comes out to 0.05735 or something close to that.

Alongside the above this is wrong for two distinct reasons; first you never ever double-attribute a behavior. This means that if you need only to find the probability of an event the size of the pool of events doesn’t matter. Let’s say that instead of throwing 3 grenades we throw 4 at the time. Does that mean it becomes .15 × .15 × .85 x .85? If so, how does that look?

Well combining those four numbers produces the value 1.6% but if it were 5 grenades looking for the same thing …

That value becomes 1.4% ( .15 × .15 × .85 x .85 x.85 ).

It’s going the wrong way. Your odds should increase not decrease so what is happening? You’re not checking to see if two will succeed you are finding the odds of two succeeding and specifically one failing, or defining the odds by specifically searching out the distinct behavior of two unified events.

The point is though, its about a 40% chance that you will get at least one 32 second bleed worth 4100 damage every time you throw a grenade. Each grenade throw is .5 seconds, meaning that in the time it takes you to cast Orbital Strike once, you have about 240 chances to proc shrapnel.

How are you getting 32s? First, the base for Shrapnel is 12s, when you take Firearms trait line you are forced to take a trait that makes your bleeds last 33% longer, so it’s 16s after that trait is forcefully applied not before. So you’ll cap out at 24s.

That said if every grenade throw is .5s and you get three at a time it looks like this:

60 / .5 = 120

120 × 3 = 360

So it’s 360 not 240 presuming continuous attack.

So 15% of 360 would be 54 activations? That’s amazingly high, no doubt, but in the rotation we were using bombs, not grenades, because the answer is completely different when you use grenades over bombs due to more opportunities. Let me think, uh, 360 – 107 (number of bombs in one minute under 100% quickness) = 253 additional chances with grenades?

I feel so old!

Oh, and let me correct one more thing, the cast time of orbital strike would be .87s so you wouldn’t gain 240 opportunities in that time, and in fact, lose a grand total of six, or two salvos altogether, because of the CD itself. That was spun way out of control there.

First of all by my logic the chance for getting two 6’s would be 1/36, but the chance of not getting 2 6’s , meaning the chance to get anything aside from 2 6’s would be, 1 – 1/36, or 35/36.

The chance of getting 2 procs of shrapnel in 4 trials is (4c2) x 0.15^2 × 0.85^2 = 9.75%, the chance of getting 2 procs of shrapnel in 5 trials is (5c2) x 0.15^2 × 0.85^3 = 13.8%. So it does increase.

I got confused though what you were talking about. When you do a grenade throw you throw 3 grenades. Each of them have a 0.15% chance to proc shrapnel, and each have a 0.85% chance to do nothing. The chance that ALL 3, do nothing is 0.85^3 . That is the chance that in a throw you proc none. Meaning that 100% minus that chance as a percent, is the chance that you do NOT get nothing, that you get something, anything.

Now if we are talking about a pure Berserker bomb build. I have had Orbital Strike hit up to 20k on the golem during a perfect non realistic trial. With that build, it would have a 34 second cooldown and a 0.75 cast time. In which case, 20,000/34.75 = 575.54 dps.

In the same build , according to GW2skills.net numbers, Shrapnel would have a bleed with 21.25 seconds duration and 1654 damage or 77.84 dps. 575.54/77.84 = 7.4. That means it would take more than 7.4 procs of shrapnel to outdo the damage done by the extra Orbital strike.

In 34 seconds you can place down 68 bombs. You need at least 8 procs of shrapnel for there to be a DPS increase of shrapnel over Orbital strike. The probability that at least 8 of the 68 bombs you place will proc shrapnel is 82%. I placed an excel snippet to show how I got that number.

For clarity, the reason it only goes up to 24, is because the chance of getting 24 or more procs of Shrapnel out of 68 trials is extremely close to 0. As you can see, even 24 shows up as 0 to 4 decimal places. Obviously it’s not 0, but it’s not really a number of concern either. Pretty much there is a 99.99999% chance that you will get less than 24 procs of shrapnel during 68 trials.

DGraves.3720:

My concern is with people becoming so incredibly theoretical that they lose touch with the game itself. The DPS arms race is the worst thing to happen to gaming overall simply because it creates a false standard of superiority through numerical mastery.

This is the first time I am reading this. Ever thought I am the only one. I’ve been seeing this in many other games, mostly with a big streaming population on Twitch. They build their characters exactly the same as the streamers do, but their actual gameplay is just clumsy and overall far, far worse compared to their gameplay they would develop during gameplay without following others or guides. They calculate everything and averages – ingame however, they land only 12.5% of the skillshots, or panic and just mash buttons, or even their PC start to have framerate issues and can’t aim properly etc….

Granted, I have asked for help (waves at Ziggy), but in the end I picked up only things that did not interfere with my gameplay I developed myself during my time here. When I see all these rotations I would lose a big junk of fun – in exchange of damage of course, but that ain’t worth it to me.

(Also, the fun should be #1, even though sacrificing damage for it is fine, but that is not the point of the thread)

Ging.6485:

DGraves.3720:

Ging.6485:

Well the chance of all three grenades failing to proc would be 0.85^3. 1 minus that number is the chance that no failure occurs, meaning that at least one succeeds. I did it in excel and am writing from my phone right now, but the chance for a complete failure was 61% and change.

That isn’t true.

Let’s take two six sided dice. We know that the probability of getting two sixes (or two of any number really) is 1/36. We know it’s 1/6 × 1/6 right?

So if that’s the case using your logic 5/6 × 5/6 or 5/6^2 would be the direct opposite, right? Well, let’s do it: 5/6^2 = .69444 or 69.44%. So using your logic if we take 100% and subtract 69.4% there’s a 30.56% of getting two sixes?

We know that is wrong. So why is it wrong?

Actually it is because you can’t replace the odds of something not happening with the odds of something happening when setting up probability matrices. You should always take the odds of the event occurring and find the values you’re looking for, for instance:

1 – .15 = .85. Easy.
1 – (.15 × 2 ) = 1 – .30 = .70. Easy.
1 – (.15 × 3) = 1 – .45 = .55. Easy.

So there is kitten percent chance that none of the grenades in a salvo will produce Shrapnel. It’s that simple. What I want you to note is that the final results of those problems changes but the actual probability of the event does not. It’s still 15% in all three problems but 85% is not the unified odds of failure, and in fact, changes the more iterations you have.

This makes sense if you think about it. Again with six sided dice:

If I roll 1 the odds are 1/6 that I’ll get (number)
If I roll 2 the odds are 1/6 × 1/6 that I’ll get (number combination)
If I roll 3 the odds are 1/6 × 1/6 × 1/6 that I’ll get (number combination)

But note that 1/6th doesn’t change. I guarantee you the result does:

1/6
1/36
1/216

Now if you took 5/6 and tried to apply it to this you would get:

5/6
25/36
125/216

That gap is worsening by the way.

I feel so boring!

The chance for exactly two to proc shrapnel is 0.15*0.15*0.85, the chance of 2 successes and one failure. There are three ways that could happen. 12 23 13. that number comes out to 0.05735 or something close to that.

Alongside the above this is wrong for two distinct reasons; first you never ever double-attribute a behavior. This means that if you need only to find the probability of an event the size of the pool of events doesn’t matter. Let’s say that instead of throwing 3 grenades we throw 4 at the time. Does that mean it becomes .15 × .15 × .85 x .85? If so, how does that look?

Well combining those four numbers produces the value 1.6% but if it were 5 grenades looking for the same thing …

That value becomes 1.4% ( .15 × .15 × .85 x .85 x.85 ).

It’s going the wrong way. Your odds should increase not decrease so what is happening? You’re not checking to see if two will succeed you are finding the odds of two succeeding and specifically one failing, or defining the odds by specifically searching out the distinct behavior of two unified events.

The point is though, its about a 40% chance that you will get at least one 32 second bleed worth 4100 damage every time you throw a grenade. Each grenade throw is .5 seconds, meaning that in the time it takes you to cast Orbital Strike once, you have about 240 chances to proc shrapnel.

How are you getting 32s? First, the base for Shrapnel is 12s, when you take Firearms trait line you are forced to take a trait that makes your bleeds last 33% longer, so it’s 16s after that trait is forcefully applied not before. So you’ll cap out at 24s.

That said if every grenade throw is .5s and you get three at a time it looks like this:

60 / .5 = 120

120 × 3 = 360

So it’s 360 not 240 presuming continuous attack.

So 15% of 360 would be 54 activations? That’s amazingly high, no doubt, but in the rotation we were using bombs, not grenades, because the answer is completely different when you use grenades over bombs due to more opportunities. Let me think, uh, 360 – 107 (number of bombs in one minute under 100% quickness) = 253 additional chances with grenades?

I feel so old!

Oh, and let me correct one more thing, the cast time of orbital strike would be .87s so you wouldn’t gain 240 opportunities in that time, and in fact, lose a grand total of six, or two salvos altogether, because of the CD itself. That was spun way out of control there.

First of all by my logic the chance for getting two 6’s would be 1/36, but the chance of not getting 2 6’s , meaning the chance to get anything aside from 2 6’s would be, 1 – 1/36, or 35/36.

The chance of getting 2 procs of shrapnel in 4 trials is (4c2) x 0.15^2 × 0.85^2 = 9.75%, the chance of getting 2 procs of shrapnel in 5 trials is (5c2) x 0.15^2 × 0.85^3 = 13.8%. So it does increase.

I got confused though what you were talking about. When you do a grenade throw you throw 3 grenades. Each of them have a 0.15% chance to proc shrapnel, and each have a 0.85% chance to do nothing. The chance that ALL 3, do nothing is 0.85^3 . That is the chance that in a throw you proc none. Meaning that 100% minus that chance as a percent, is the chance that you do NOT get nothing, that you get something, anything.

Now if we are talking about a pure Berserker bomb build. I have had Orbital Strike hit up to 20k on the golem during a perfect non realistic trial. With that build, it would have a 34 second cooldown and a 0.75 cast time. In which case, 20,000/34.75 = 575.54 dps.

In the same build , according to GW2skills.net numbers, Shrapnel would have a bleed with 21.25 seconds duration and 1654 damage or 77.84 dps. 575.54/77.84 = 7.4. That means it would take more than 7.4 procs of shrapnel to outdo the damage done by the extra Orbital strike.

In 34 seconds you can place down 68 bombs. You need at least 8 procs of shrapnel for there to be a DPS increase of shrapnel over Orbital strike. The probability that at least 8 of the 68 bombs you place will proc shrapnel is 82%. I placed an excel snippet to show how I got that number.

For clarity, the reason it only goes up to 24, is because the chance of getting 24 or more procs of Shrapnel out of 68 trials is extremely close to 0. As you can see, even 24 shows up as 0 to 4 decimal places. Obviously it’s not 0, but it’s not really a number of concern either. Pretty much there is a 99.99999% chance that you will get less than 24 procs of shrapnel during 68 trials.

Have you ever actually tested your numbers?

I’ll go on regardless but before I do I need to know. It’ll help me know how to explain.

DGraves.3720:

Ging.6485:

Well the chance of all three grenades failing to proc would be 0.85^3. 1 minus that number is the chance that no failure occurs, meaning that at least one succeeds. I did it in excel and am writing from my phone right now, but the chance for a complete failure was 61% and change.

That isn’t true.

Let’s take two six sided dice. We know that the probability of getting two sixes (or two of any number really) is 1/36. We know it’s 1/6 × 1/6 right?

So if that’s the case using your logic 5/6 × 5/6 or 5/6^2 would be the direct opposite, right? Well, let’s do it: 5/6^2 = .69444 or 69.44%. So using your logic if we take 100% and subtract 69.4% there’s a 30.56% of getting two sixes?

We know that is wrong. So why is it wrong?

…

Your math here is missing something: the event where one six is rolled.
The probabilities of rolling no sixes and rolling two sixes therefore do not add up to 1.

Two sixes: 1/6 × 1/6 = 1/36 or 2.8%
No sixes: 5/6 × 5/6 = 25/36 or 69.4%
One six: 5/6 × 1/6 + 1/6 × 5/6 = 10/36 or 27.8%

So the probability for one or more sixes is the total (1) minus the probability for no sixes = 30,6%

The same applies to Shrapnel. The probability to get 0 procs in one grenade throw is 0.85^3 = 61,4%.
Therefore the probability to get at least one Shrapnel bleed is 38,6%.

The chances for a single grenade throw are:

61.41% no bleed
32.51% one bleed
5.74% two bleeds
0.34% three bleeds

@DGraves
You went into far more detail then I thought you would, and I appreciate you explaining this in great detail, most people wouldn’t do that.
After reading your posts, I have to say, it makes sense. A lot of people blindly accept what some people with a little fame say. I was one of them, sad to say, until I tried to experiment with my own builds. Yes, I use the SD build right now, but I’m not sure how long i’ll be using it.
Do you have any builds that might worth for me to try?
I would really like to see if you have come up with any unique builds.

Mork vom Ork.2598:

DGraves.3720:

Ging.6485:

Well the chance of all three grenades failing to proc would be 0.85^3. 1 minus that number is the chance that no failure occurs, meaning that at least one succeeds. I did it in excel and am writing from my phone right now, but the chance for a complete failure was 61% and change.

That isn’t true.

Let’s take two six sided dice. We know that the probability of getting two sixes (or two of any number really) is 1/36. We know it’s 1/6 × 1/6 right?

So if that’s the case using your logic 5/6 × 5/6 or 5/6^2 would be the direct opposite, right? Well, let’s do it: 5/6^2 = .69444 or 69.44%. So using your logic if we take 100% and subtract 69.4% there’s a 30.56% of getting two sixes?

We know that is wrong. So why is it wrong?

…

Your math here is missing something: the event where one six is rolled.
The probabilities of rolling no sixes and rolling two sixes therefore do not add up to 1.

Two sixes: 1/6 × 1/6 = 1/36 or 2.8%
No sixes: 5/6 × 5/6 = 25/36 or 69.4%
One six: 5/6 × 1/6 + 1/6 × 5/6 = 10/36 or 27.8%

So the probability for one or more sixes is the total (1) minus the probability for no sixes = 30,6%

The same applies to Shrapnel. The probability to get 0 procs in one grenade throw is 0.85^3 = 61,4%.
Therefore the probability to get at least one Shrapnel bleed is 38,6%.

The chances for a single grenade throw are:

61.41% no bleed
32.51% one bleed
5.74% two bleeds
0.34% three bleeds

Let’s say you had 100 slips that told you the outcome of your attack. You’d pick a slip and afterwards you’d take it out and continue picking until you had none left. You have a 15% chance of success at whatever.

61.41% of those slips would say “total failure”.
32.51% of those slips would say “meager success”.
5.74% of those slips would say “astounding success”.
.34% of those slips would say “miraculous success”.

This is a dependent variable matrix. You will eventually get all of the outcomes because every time you choose a slip you get to remove an outcome.

However in independent variables the success of your previous action does not dictate the success of your next. That means that when you pick a slip you simply have a 15% chance of getting “You win” and an 85% chance of getting “You lose”. The event either happens or it does not regardless of the previous pull.

I stopped using matrices like that a long time ago but used to in order to dissect critical strikes. Turns out it was total crap and never panned out correctly because the matrix size wasn’t big enough. It was great when I wanted to know the odds of continuing to get critical strikes in a specific number in a row but otherwise it doesn’t tell you anything about the actual event’s expression itself.

Some things you just take at face value. Don’t overcomplicate it.

Sidenote: the purpose of showing that 1/36 and 25/36 =/= 36/36 was specifically to acknowledge the fact that it was a dependent matrix table. I just didn’t know how to explain that at the time because I’m not a teacher; I do math all day and can see errors but don’t expect me to be able to relay my thoughts to others on the matter? Ha! Pathetique!

https://www.wyzant.com/resources/lessons/math/statistics_and_probability/probability/further_concepts_in_probability

Eragon.8234:

@DGraves
You went into far more detail then I thought you would, and I appreciate you explaining this in great detail, most people wouldn’t do that.
After reading your posts, I have to say, it makes sense. A lot of people blindly accept what some people with a little fame say. I was one of them, sad to say, until I tried to experiment with my own builds. Yes, I use the SD build right now, but I’m not sure how long i’ll be using it.
Do you have any builds that might worth for me to try?
I would really like to see if you have come up with any unique builds.

Everyone has thoughts just as deep. It’s all a matter of making the choice to try your best to express them. That said I don’t think it’s bad that builds exist and I don’t want to sound like I am “against” people making general observations or progressing theorycrafting and gameplay by being honest I just don’t see the point in not verifying people’s claims. It becomes dangerous to take everything people commonly believe as just outright truth because it hurts the community since it takes time to pluck the idea and then reinsert the newer better idea no matter who it is endorsed by or how true it is.

But as for builds, ha! I’ve none. I change mine constantly looking at all sorts of things and testing ideas all the time. I help build some builds but I don’t really have any to give out simply because I’m always doing something different all the time. I really should take notes though; the only thing that I’m proud of lately is the Valk/Knights hybrid instead of using Soldier’s in WvWvW which uses an idea I came up with.

It’s not bad. But it’s definitely not MetaBattle worthy. :p

Well, thank you for taking the time to explain this.
And there’s nothing wrong if your build isn’t MetaBattle worthy,lol :P

DGraves.3720:

Mork vom Ork.2598:

DGraves.3720:

Ging.6485:

Well the chance of all three grenades failing to proc would be 0.85^3. 1 minus that number is the chance that no failure occurs, meaning that at least one succeeds. I did it in excel and am writing from my phone right now, but the chance for a complete failure was 61% and change.

That isn’t true.

Let’s take two six sided dice. We know that the probability of getting two sixes (or two of any number really) is 1/36. We know it’s 1/6 × 1/6 right?

So if that’s the case using your logic 5/6 × 5/6 or 5/6^2 would be the direct opposite, right? Well, let’s do it: 5/6^2 = .69444 or 69.44%. So using your logic if we take 100% and subtract 69.4% there’s a 30.56% of getting two sixes?

We know that is wrong. So why is it wrong?

…

Your math here is missing something: the event where one six is rolled.
The probabilities of rolling no sixes and rolling two sixes therefore do not add up to 1.

Two sixes: 1/6 × 1/6 = 1/36 or 2.8%
No sixes: 5/6 × 5/6 = 25/36 or 69.4%
One six: 5/6 × 1/6 + 1/6 × 5/6 = 10/36 or 27.8%

So the probability for one or more sixes is the total (1) minus the probability for no sixes = 30,6%

The same applies to Shrapnel. The probability to get 0 procs in one grenade throw is 0.85^3 = 61,4%.
Therefore the probability to get at least one Shrapnel bleed is 38,6%.

The chances for a single grenade throw are:

61.41% no bleed
32.51% one bleed
5.74% two bleeds
0.34% three bleeds

Let’s say you had 100 slips that told you the outcome of your attack. You’d pick a slip and afterwards you’d take it out and continue picking until you had none left. You have a 15% chance of success at whatever.

61.41% of those slips would say “total failure”.
32.51% of those slips would say “meager success”.
5.74% of those slips would say “astounding success”.
.34% of those slips would say “miraculous success”.

This is a dependent variable matrix. You will eventually get all of the outcomes because every time you choose a slip you get to remove an outcome.

However in independent variables the success of your previous action does not dictate the success of your next. That means that when you pick a slip you simply have a 15% chance of getting “You win” and an 85% chance of getting “You lose”. The event either happens or it does not regardless of the previous pull.

I stopped using matrices like that a long time ago but used to in order to dissect critical strikes. Turns out it was total crap and never panned out correctly because the matrix size wasn’t big enough. It was great when I wanted to know the odds of continuing to get critical strikes in a specific number in a row but otherwise it doesn’t tell you anything about the actual event’s expression itself.

Some things you just take at face value. Don’t overcomplicate it.

Sidenote: the purpose of showing that 1/36 and 25/36 =/= 36/36 was specifically to acknowledge the fact that it was a dependent matrix table. I just didn’t know how to explain that at the time because I’m not a teacher; I do math all day and can see errors but don’t expect me to be able to relay my thoughts to others on the matter? Ha! Pathetique!

https://www.wyzant.com/resources/lessons/math/statistics_and_probability/probability/further_concepts_in_probability

Well, each explosion is an independent event that’s right.

But calculating the chance for x outcomes within y events is what you call a dependent matrix. And the probabilities of all possible outcomes always add up to 1.

By your calculation the chance for at least one bleed within 3 explosions is 15% + 15% +15% =45%.
Let’s try to simplify this and say we toss a coin three times. each time it’s 50% heads or tails.
By your calculation the probability for at least one heads toss would be 150%. And so would be for at least one tails.
Both by themselves already are above the total possible sum of probabilities however.

Well, each explosion is an independent event that’s right.

But calculating the chance for x outcomes within y events is what you call a dependent matrix. And the probabilities of all possible outcomes always add up to 1.

By your calculation the chance for at least one bleed within 3 explosions is 15% + 15% +15% =45%.
Let’s try to simplify this and say we toss a coin three times. each time it’s 50% heads or tails.
By your calculation the probability for at least one heads toss would be 150%. And so would be for at least one tails.
Both by themselves already are above the total possible sum of probabilities however.

No.

You’re asking a set question and using independent event math. Can’t mix them (which is my point all along).

First it’s not 150% because what you do to one side of An equation you must do to another.

(X + Y) = 100 is the formula.

In coins that’s 50 + 50.

3(X + Y) = 3(100) is what you’d use to maintain that equilibrium and if you removed the iterations you’ll get the same base probability.

(150+150)/3=(300)/3 = (50 + 50) = 100.

Same is true for Shrapnel.

(15 + 85) = 100.

3(15 + 85) = 300

(45 + 255) = 300

(45 + 255)/3 =300/3

(15 + 85) = 100.

You can think of the above this way: if you had one grenade per attack and made 100 attacks that is 100 chances set at 15%.

If you have three grenades per attack over the same 100 attacks that’s 300 chances at 15%.

15% of 300 is 45 positive results.

85% of 300 is 255 negative results.

In a set matrix this changes drastically because you’re looking for a specific outcome while linking all the events dependently.

In coins the odds of getting at least one heads on two flips isn’t 50% in a matrix, it’s 75%:

H + T
H + H
T + H
T + T

Think punnet square or Nash equilibrium.

Another way it differs is that the number of trials doesn’t matter in independent events. I can throw 15b grenades and get the same results. In sets that’s not the case. But sets are not dynamic either, it produces a table of results that only works for that specified set of trials.

Your odds will be proportional in independent events but not dependent ones.

I agree with DGraves. 15% chance on Shrapnel is an individual event occuring 3 times. If you throw 100 times a grenade auto attack (3 nades each), it would be the same as if you’d throw 100 times a solo grenade with 45% chance to proc Shrapnel.

Grenades have a 1s cast time, not 0.5 as mentioned ingame. Someone mentioned it somewhere :P So don’t get confused by ingame cast times – they are almost ALL wrong.

Xyonon.3987:

I agree with DGraves. 15% chance on Shrapnel is an individual event occuring 3 times. If you throw 100 times a grenade auto attack (3 nades each), it would be the same as if you’d throw 100 times a solo grenade with 45% chance to proc Shrapnel.

---

Grenades have a 1s cast time, not 0.5 as mentioned ingame. Someone mentioned it somewhere :P So don’t get confused by ingame cast times – they are almost ALL wrong.

I ran a test on Excel with a random number generator. 3 random numbers between 1 and 100, result on average with at least 1 less than 16, about 39 times out of 100.

With a sample space of 100, it fluctuate between 26 and 47.

With a sample space of 1000, it’s 370 and 400.

Not 3 random numbers at once. 1 random number times 3. As DGraves mentioned mutltiple times now → “each explosion is an independent event”

DGraves.3720:

Ging.6485:

Well the chance of all three grenades failing to proc would be 0.85^3. 1 minus that number is the chance that no failure occurs, meaning that at least one succeeds. I did it in excel and am writing from my phone right now, but the chance for a complete failure was 61% and change.

That isn’t true.

Let’s take two six sided dice. We know that the probability of getting two sixes (or two of any number really) is 1/36. We know it’s 1/6 × 1/6 right?

So if that’s the case using your logic 5/6 × 5/6 or 5/6^2 would be the direct opposite, right? Well, let’s do it: 5/6^2 = .69444 or 69.44%. So using your logic if we take 100% and subtract 69.4% there’s a 30.56% of getting two sixes?

No. The direct opposite of “having two 6” is “not having two 6”.
5*6^2 describe the probability of “not having any 6 at all”, which doesn’t encompass “having only one 6”.
Take your head/tail matrix :
H | H
H | T
T | T
T | H
The odds of getting exactly two heads equate to the first case among the four, so 1/4. The odds of not having exactly two heads are all of the other repartitions, which are 3/4.
1-(1/4) does indeed result in 3/4.

DGraves.3720:

This makes sense if you think about it. Again with six sided dice:

If I roll 1 the odds are 1/6 that I’ll get (number)
If I roll 2 the odds are 1/6 × 1/6 that I’ll get (number combination)
If I roll 3 the odds are 1/6 × 1/6 × 1/6 that I’ll get (number combination)

But note that 1/6th doesn’t change. I guarantee you the result does:

1/6
1/36
1/216

Now if you took 5/6 and tried to apply it to this you would get:

5/6
25/36
125/216

That gap is worsening by the way.

Of course, it doesn’t works, since you aren’t applying the discussed formula. As you rightfully explain, the odds of getting a specific combination on three dice is 1/216. The event considered is “having the combination”, not “having a certain result on a certain dice”. So the odd of “not having the combination” is 1-(the odds of the combination) = 1-(1/216) = 215/216, which is totally reasonnable.
In the case of the grenades, not having any grenade procing is your case of throwing three dice and expecting the combination (1,1,1), with 1 having a probability of 85% instead of 1/6. Which gives a probability for the combination of: 0.85^3 = 61% (according to your own math).
The odds of having at least one proc is then everything not in those 61%, so: 1-0.61 = 39% (as Ging stated, correctly).
The odds of having exactly one proc for each position is: 0.15*0.85^2 = 11%, hence the odds of two or three procs are 39% – (11% * 3 position for the proc) = 6%.
In the same way, the odds of exactly two procs is: 0.15*0.15*0.85 = 1.9%. Each means a certain grenade didn’t proc, so for three grenade you have a total chance of: 1.9*3 = 5.7% for exactly two procs in three grenades.
That leaves 0.3% for the odds of having three procs, and indeed: 0.15^3 = .3% (which you did calculate yourself earlier).
The math adds up. There is no repartition that would gives you the 2% of getting the two proc you determined.

DGraves.3720:

The chance for exactly two to proc shrapnel is 0.15*0.15*0.85, the chance of 2 successes and one failure. There are three ways that could happen. 12 23 13. that number comes out to 0.05735 or something close to that.

Alongside the above this is wrong for two distinct reasons; first you never ever double-attribute a behavior. This means that if you need only to find the probability of an event the size of the pool of events doesn’t matter.

First, you totally do double attribute a behaviour (if I understand that the same way as you do, as in taking 12/13/23 instead of just 12). That’s why in your head/tail table you have an entry for H|T and one for T|H. It does matter, otherwise the repartition of two coin throws would be 1/3, which is provably wrong.
And for the second point, that would be true only if you forget he said “exactly” two procs. When you search for a set number of occurence, the size of the pool does matter. More on that later.

DGraves.3720:

Let’s say that instead of throwing 3 grenades we throw 4 at the time. Does that mean it becomes .15 × .15 × .85 x .85

Indeed.

DGraves.3720:

If so, how does that look?

Well combining those four numbers produces the value 1.6% but if it were 5 grenades looking for the same thing …

That value becomes 1.4% ( .15 × .15 × .85 x .85 x.85 ).

It’s going the wrong way. Your odds should increase not decrease so what is happening?

It is happening that math works. The more you throw a coin, the less the chances of having only two heads in the result. With two throws, the odds are 1/4. With ten thousands of throw, the odds of having two heads are astonomically low.
You are computing the odds of exactly two occurences in a variable set, and trying to conclude of the odds of having at least two occurences on the same set. Unsurprisingly, the result is wrong.

Xyonon.3987:

15% chance on Shrapnel is an individual event occuring 3 times. If you throw 100 times a grenade auto attack (3 nades each), it would be the same as if you’d throw 100 times a solo grenade with 45% chance to proc Shrapnel.

Easy way to see it is wrong, is that while improbable, we can get up to 300 procs with nades, and your case would prevent getting more than 100. The bell curves would be wildly different in both cases.

I won’t say anything about DPS and rotation, but as probabilities are concerned, Ging is right. You have a 61/33/5.7/0.3 repartition.

Alongside the above this is wrong for two distinct reasons; first you never ever double-attribute a behavior. This means that if you need only to find the probability of an event the size of the pool of events doesn’t matter.

First, you totally do double attribute a behaviour (if I understand that the same way as you do, as in taking 12/13/23 instead of just 12). That’s why in your head/tail table you have an entry for H|T and one for T|H. It does matter, otherwise the repartition of two coin throws would be 1/3, which is provably wrong.
And for the second point, that would be true only if you forget he said “exactly” two procs. When you search for a set number of occurence, the size of the pool does matter. More on that later.

You do not double attribute behavior.

A coins chance is 50/50 no matter how many coins you flip. Double attribution specifically ruins outcomes by adding unnecessary data.

It’s as simple as asking “what are the odds of getting heads on a fair coin if I flip it 900 times on any given throw?”

A: 50/50

This is NOT “what are the odds of getting at least one heads if I flip 900 fair coins?”

A: (.5^900)-1= A very, very small number.

The rest has been previously addressed.