Remove Advertisements

Testing the combat table

Warning: Theorycraft inside.

Moderators: Fridmarr, Worldie, Aergis, theckhd

Re: Testing the combat table

Postby Salamandra » Sun May 03, 2009 10:17 am

It would start from 0 somewhat because it's not a 100-roll system, it would be more accurately desctribed as a 10,000 roll system (given that tooltips go to 2 decimal places, and actual avoidance may go into further decimal places), and so starting at 1 would be like starting at 100, thereby removing legitimate possibilities from your theorycrafting.

Also, your data doesn't suggest any changes, still. It just shows a shuffling around of the combat table that has no actual change on the probability of anything. You say things like something that would have otherwise been a block being a parry, where if it's the other way round it would have been a dodge, but the chance of getting a parry, the chance of getting a dodge, the chance of getting a miss, the chance of getting a block, they're all exactly the same regardless of the order of miss dodge and parry in both no-bladewarding and the bladewarding examples.

You're sort of arguing it would matter retroactively given a specific outcome of a specific hit. Which, yes, I suppose it would, but it's totally irrelevant because it already happened and you couldn't have known if you would have gotten a dodge or a parry until the hit happened.
Image
User avatar
Salamandra
 
Posts: 373
Joined: Mon May 12, 2008 5:45 pm

Re: Testing the combat table

Postby toothdecaykills » Sun May 03, 2009 10:46 am

Salamandra wrote:It would start from 0 somewhat because it's not a 100-roll system, it would be more accurately desctribed as a 10,000 roll system (given that tooltips go to 2 decimal places, and actual avoidance may go into further decimal places), and so starting at 1 would be like starting at 100, thereby removing legitimate possibilities from your theorycrafting.


Except you can just assume I truncated those decimal places because the theoretical values I threw in there were all whole numbers. Although starting at 1 might be like starting at 100, I've extended the opposite direction, ending at 100 when you would have ended at 99.

Salamandra wrote:Also, your data doesn't suggest any changes, still. It just shows a shuffling around of the combat table that has no actual change on the probability of anything. You say things like something that would have otherwise been a block being a parry, where if it's the other way round it would have been a dodge, but the chance of getting a parry, the chance of getting a dodge, the chance of getting a miss, the chance of getting a block, they're all exactly the same regardless of the order of miss dodge and parry in both no-bladewarding and the bladewarding examples.


The tables all represent what happens to a very simplified combat table when Blade Ward is applied, so yes, they do change by increasing your total avoidance. I'm pretty certain that is what the enchant does. So, if it increases your avoidance, and we know the table has a limit, then obviously it changes the probability of things occurring. In fact, Blade Ward specifically changes the probability of Parries occurring. However, depending on the placement of the different outcomes on the table, it has an effect on the entire table, because its a closed system.

Salamandra wrote:You're sort of arguing it would matter retroactively given a specific outcome of a specific hit. Which, yes, I suppose it would, but it's totally irrelevant because it already happened and you couldn't have known if you would have gotten a dodge or a parry until the hit happened.


No, this is incorrect, you're completely forgetting that Blade Ward is not up all the time and the only thing we're interested in is its impact WHILE its up. You can't know that the 1% Dodge you just added by upgrading your gear is helping either, but you aren't specifically looking at each individual hit to determine its overall effectiveness. You know that increasing the chance for something to occur is the point, so adding 2% Parry increases your total chance to parry an attack, I mean, duh.

But it isn't as cut and dry as that. The buff only provides parry, is removed when you parry, and depending on the order of the table, would position dodge or miss as more desirable outcomes because the total avoidance increase remains for longer. If you can create a scenario where a roll that would have resulted in a block ends up as a dodge because of the shift of the table, then it does happen, will happen, and will continue to happen. You can also think up a scenario where, because of its effects on the table, Blade Ward saves you by remaining longer and expiring on an important hit. I'm not saying this will happen all the time, I'm just stating that it does happen and can have varying degrees of effectiveness in a fight based entirely on the positioning of the combat table.

I tried to simplify it to show you what I meant. Regardless of whether you know whether these fringe moments happened or not doesn't change that it could happen, it will happen, and it does happen.

As an exercise, why not increase the effects of Blade Ward to 20% instead of 2%, but keep the same limitations. You'll see how the order in the table would effect the buff.
toothdecaykills
 
Posts: 138
Joined: Sat Jul 28, 2007 10:50 pm

Re: Testing the combat table

Postby Lightstrike » Sun May 03, 2009 12:05 pm

You're arguing a lost case toothdecay...

Range dictates probability, not location.. if I have a 6 sided die and say that 1-4 is a parry and 5 and 6 is a dodge, you have a 4/6 or 2/3 chance of rolling a parry; and 2/6 or 1/3 chance of rolling a dodge.

You can split dodge and parry up as much as you want, but the probability will always be the same...

If 1,2,4,5 are parry, and 3 and 6 are dodge... you still have 4/6 outcomes being parry and 2/6 outcomes being a dodge, it makes no difference as you still have the same chance of rolling a parry as before, and the same chance of rolling a dodge as before.


The difference with bladewarding is that it extends a section of the combat table..

However the order still doesn't matter because the outcome is not known.

Let's use your example here...

Blade Ward procs so you have:

Miss 01-10
Parry 11-22
Dodge 23-85
Block 86-100

You gain the extra 2% parry, so parry has its range extended by 2.. however the argument you are putting forward is that if the roll is 21, you parry again where it would be a dodge if bladeward hadn't procced. But it did proc, so this retroactive analysis is worthless.

It's like watching a lottery draw and then saying that you knew you were going to win after the numbers are drawn and you know the result... however unless you cheated, you wouldn't have known what numbers were going to come out, so you couldn't know the outcome until after..

Maybe a bad example there, but the logic is the same, the outcome is undecided, so wherever parry is in the table is a non-issue.

Order would only be in issue in a situation where it were affected by a specific ability that isn't itself. And this is not the case.
Image
Lightstrike
 
Posts: 380
Joined: Tue Aug 05, 2008 11:50 am

Re: Testing the combat table

Postby Salamandra » Sun May 03, 2009 4:55 pm

If you can create a scenario where a roll that would have resulted in a block ends up as a dodge because of the shift of the table


But you wouldn't cause a dodge because of the shift of the table, because just shifting the table makes no difference for the chance of anything within that table occuring. A 20% chance at the beginning of the table or a 20% chance at the end of the table, it makes no difference at all since they're still both 20% chance.
Image
User avatar
Salamandra
 
Posts: 373
Joined: Mon May 12, 2008 5:45 pm

Re: Testing the combat table

Postby Rhiannon » Sun May 03, 2009 7:49 pm

How is this still going on?

If you've a 25% chance to dodge an attack and so not consume the bladewarding buff, and a 20% chance to parry an attack and so consume the bladewarding buff, either way your dodge + parry is 45%, and regardless of how they are ordered in the combat table you have 20% chance to consume the bladewarding buff. The only time ordering in the combat table is relevant is when talking about pushing things off the combat table, which is certainly not the case in this situation.
Rhiannon
 
Posts: 1061
Joined: Mon Mar 03, 2008 8:17 am

Re: Testing the combat table

Postby majiben » Sun May 03, 2009 9:29 pm

It's relevant for the reason that one ordering allows for hot streaks (and more cold streaks) and another doesn't. The average avoidance is the same though. There is a difference in the individual cases even if there isn't in the overall result.
Amirya wrote:some bizarre lovechild of Hawking, Einstein, and Theck
User avatar
majiben
Moderator
 
Posts: 6999
Joined: Fri Aug 22, 2008 4:37 pm
Location: Retired

Re: Testing the combat table

Postby Dorvan » Sun May 03, 2009 9:32 pm

Majiben wrote:It's relevant for the reason that one ordering allows for hot streaks (and more cold streaks) and another doesn't. The average avoidance is the same though. There is a difference in the individual cases even if there isn't in the overall result.


No. There is no "hot streak" or " cold streak" phenomenon. On every single roll, the only thing that determines the likelihood of an outcome is how much space on the roll table it takes up. Unless some numbers are inherently more likely than others, I have no idea what you're arguing.

Why would you get "hot streaks" and "cold streaks"? If you could lay out your reasoning more clearly, perhaps it'd be easier for me to explain why you're wrong. Maybe responding to Lightstrike's post would be helpful, as his die example is a very simple version of the exact same issue.
Image

WHAT WOULD BEST DESCRIBE YOUR PERSONALITY?
Moonlight Sonata Techno Remix
Scriggle - 85 Fire Mage
Fizzmore - 81 Mut Rogue
Adorania - 80 Disc Priest
User avatar
Dorvan
Maintankadonor
 
Posts: 8462
Joined: Wed Apr 04, 2007 3:28 pm

Re: Testing the combat table

Postby majiben » Sun May 03, 2009 10:09 pm

Using the simplified combat table there is a .04% chance of getting 2 rolls in a row in a row in the 51-52 range. That's a hot streak if parry comes before dodge since that results in 2 avoided hits, as opposed to if parry came after dodge, in which case you would have one avoided (blade warding consumed) and then one hit. Now this of course means that parries in the 11-32 range burn the enchant with no benefit, especially the 31-32 as those were previously dodges.
Amirya wrote:some bizarre lovechild of Hawking, Einstein, and Theck
User avatar
majiben
Moderator
 
Posts: 6999
Joined: Fri Aug 22, 2008 4:37 pm
Location: Retired

Re: Testing the combat table

Postby Lightstrike » Sun May 03, 2009 10:13 pm

Yeah, but you could claim this no matter where parry is, If parry was before miss, you'd say the same thing about miss.
If parry were before block then it would be the same thing there too... the only thing that changes is what is displaced during the "hot streak" The only way it would be different is if parry is last on the table and is getting knocked off of it.

The hot streaks and cold streaks are there no matter where parry is.
Image
Lightstrike
 
Posts: 380
Joined: Tue Aug 05, 2008 11:50 am

Re: Testing the combat table

Postby Dorvan » Sun May 03, 2009 10:16 pm

Majiben wrote:Using the simplified combat table there is a .04% chance of getting 2 rolls in a row in a row in the 51-52 range. That's a hot streak if parry comes before dodge since that results in 2 avoided hits, as opposed to if parry came after dodge, in which case you would have one avoided (blade warding consumed) and then one hit. Now this of course means that parries in the 11-32 range burn the enchant with no benefit, especially the 31-32 as those were previously dodges.


Please lay out the entire combat table (there have been several, so I'm not sure which simplified one you're referring to) and redo your calculation over the whole table. The probability of getting two rolls in a row in 51-52 is .04%, that's true, but that's not really important. What matters is the probability of *events*, such as block, parry, dodge, miss, etc. Looking at two points of the table in isolation from the rest is....well.....pointless.
Image

WHAT WOULD BEST DESCRIBE YOUR PERSONALITY?
Moonlight Sonata Techno Remix
Scriggle - 85 Fire Mage
Fizzmore - 81 Mut Rogue
Adorania - 80 Disc Priest
User avatar
Dorvan
Maintankadonor
 
Posts: 8462
Joined: Wed Apr 04, 2007 3:28 pm

Re: Testing the combat table

Postby Lightstrike » Sun May 03, 2009 10:25 pm

Taking what dorvan said a little further, I think the main flaw of your theory is that you do not know the next outcome as it hasn't happened yet. If you have .04% chance of parrying twice and it being a cold streak or whatever, it doesn't matter where in the table parry is as it's place in the table doesn't affect how much room in the table it takes up unless it is at the end which we know not to be true.
Image
Lightstrike
 
Posts: 380
Joined: Tue Aug 05, 2008 11:50 am

Re: Testing the combat table

Postby Selinaria » Sun May 03, 2009 10:59 pm

Lightstrike and Dorvan are correct.

Since the attack table is just a glorified list of probabilities and their chances to occur, the actual ordering does not matter because what really does matter are the raw numbers. A 20% parry chance is still just a 20% chance that every single attack against you will be a parry regardless of the ordering. A proc of blade ward increases the chance that parries occur (while removing hit or block from the table) but is independent of the all previous and future results. Having a 22% chance to parry with the proc is still simply a 22% chance to parry with the buff present, once this parry occurs, we are back to our original 20% chance to parry.

It is truly a case of the saying: Statistics have no memory. There is no hot or cold streak, there is simply a perception of such based on random chance. Avoidance is just a slightly more developed system than pure chance in that there are rare cases where ordering does matter (over 100% avoidance vs. a given mob).
Selinaria was suddenly surrounded by a gang of sharks, a cracked bat as her only protection!
Image
User avatar
Selinaria
 
Posts: 84
Joined: Mon Mar 23, 2009 7:55 am
Location: Burlington,Ontario, Canada

Re: Testing the combat table

Postby Jasari » Mon May 04, 2009 4:11 am

How is this debate still going on?

It's been proven almost a dozen times in this very topic that order doesn't matter at all.
User avatar
Jasari
 
Posts: 2195
Joined: Sun Apr 27, 2008 12:32 am
Location: Boston, MA

Re: Testing the combat table

Postby Argali » Mon May 04, 2009 6:09 am

Technically, order does matter.

When? If you have enough miss+dodge+parry to push block off the combat table.

Which will never happen.

So stop this thread.
Image
User avatar
Argali
 
Posts: 1080
Joined: Sat Jan 26, 2008 7:38 am

Re: Testing the combat table

Postby jere » Tue May 05, 2009 5:59 am

I think the problem you are hitting Majiben is that you are only looking at one side of the dodge region. Let's take your example:

Let's assume an order of miss=>parry=> dodge.

A combat table ordered 1-100
1-10 miss
11-30 parry
31-50 dodge
51-100 hit

Let blade ward's proc extend parry's range by two.

The new combat looks like
1-10 miss
11-32 parry
33-52 dodge
53-100 hit

This means that a combat roll of 51 or 52 is a dodge thanks to blade warding. This increases the value of blade warding as it can provide avoidance without being consumed.


You are right, now 51 and 52 which were once hits become dodges. However, you also lose 31 and 32 to the parry portion of the table (31 and 32 used to be dodges, but now they are parries). So now you have an increased chance to eat up charges in that region, even if you gained a dodge region in 51 and 52. In order to show that order matters, you need to show that it is more likely that you get a roll in the 31-32 region than in the 51-52 region (or vice versa). If both regions are equally likely, then it doesn't matter if dodge comes before or after parry because you have just as much chance of falling in the newly converted dodge region as you do the newly converted parry region (which the dodge region lost to the parry region).

I am pretty sure that all 2 point regions are equally likely in a space of 1-100 (or as close as a pseudo random number generator allows, which could be something to consider if you know what algorithm they use).
Image
User avatar
jere
 
Posts: 2959
Joined: Fri May 11, 2007 5:12 pm

PreviousNext

Return to Advanced Theorycraft and Calculations

Who is online

Users browsing this forum: No registered users and 1 guest

cron

Remove Advertisements

Who is online

In total there is 1 user online :: 0 registered, 0 hidden and 1 guest (based on users active over the past 5 minutes)
Most users ever online was 380 on Tue Oct 14, 2008 6:28 pm

Users browsing this forum: No registered users and 1 guest
?php } else { ?