How Repairs work.

[Retard]

I'm lazy so I don't feel like rewriting this. But if you read it, you can basically get the gyst of why Edern fails so much (Or not so much)

"Thats not how the system works. Lets say you repair a 10 point weapon at 98% edern. That means you have 2 percent chance of breaking a point off your item. So if you have a 10 point weapong, you only need to repair it 5 times for 1 point to break off.

basically, you have a 1/5 chance of breaking a point off which is 20%. Realistically you only have 80% chance of succeeding. With Ferghus, you will statistically have 100% (Probably more like 99%) chance of breaking a point off of a 10+ Durb weapon and you have a 0% chance (Probably 1%) of not breaking it.

You see people get so confused these days. They think 98% of the time they will have a perfect repair with Edern... No, it's 98% of the time 1 point will be safe not all points."

[Retard]

Dont read the last blurb, it doesnt really make sense XD

[Sunureu13]

Uh.. 49/50 is not 80%. if you fully repair a 10 dur weapon 5 times, statistically you'll break one point.

Let's do the math.

10 dur x 5 full repairs = 50.
1 fail out of 50 repairs = 49.
49/50 = 98%.

[Science]

Uh.. 49/50 is not 80%. if you fully repair a 10 dur weapon 5 times, statistically you'll break one point.

Let's do the math.

10 dur x 5 full repairs = 50.
1 fail out of 50 repairs = 49.
49/50 = 98%.

I think he's going by the logic of 2%+9(2%)=20%

So he says it's a 20% chance of failure or something.

[Sunureu13]

That's not how it works though .-.

[Science]

That's not how it works though .-.

I fail at math so I'm just trying to see how he did it.

Edit: I see repairs as individual events, do the previous repair doesn't depend on the other and vice versa, so you have a 98% chance to fail at any given repair.

/sunu's post wins in my eyes

[Chihaya]

/sunu's post wins in my eyes

QFT.

I don't even understand your calculations, personally XD

Put in some signs in thar ;D

[Cathaoir]

I'm not gonna have an argument about this again, but below is how it works.


But also with how the repair system works, its 99% per point, so each point has a 99% chance to succeed, so only a 1% chance to fail. That is per point and it is not added up. So say I repair 20 dura each time, each of those 20 dura points have a 99% chance to succeed. So first point has a 99% chance, the 2nd point has a 99% chance and so on.


Exactly my point though. In Mabi, there is no expected outcome it is all isolated incidences. No matter what you got on the previous point, be it fail or success, the next one will always have that fixed rate. If you are doing something at a 50% rate on Mabi, you can have 10 successes in a row, but the next one will still be a 50% rate of succeeding.

[Nyo]

Use statistics.

(Each "x REPAIRS" is contained; the percentage sum of each equals 100%)

1 REPAIR
1 success = 98%
1 fail = 2%

2 REPAIRS
2 successes = 98% * 98% = 96.04%
1 success and 1 fail, or 1 fail and 1 success = (98% * 2%) + (2% * 98%) = 3.92%
2 fails = 2% * 2% = 0.04%

3 REPAIRS
3 successes = 94.1192%
2 successes, 1 fail = 5.7624%
2 fails, 1 success = 0.1176%
3 fails = 0.0008%

4 REPAIRS
4S = 92.236816%
3S 1F = 7.529536%
2S 2F = 0.230496%
1S 3F = 0.003136%
4F = 0.000016%

10 REPAIRS
10S = 81.7072806887547%
9S1F = 16.674955242603%
1.53137344064721%
0.0833400511916851%
0.00297643039970304%
0.000072892173053952%
0.00000123966280704%
0.00000001445670912%
0.00000000011063808%
0.00000000000050176%
0.000000000000001024%

[Phanterz]

This is why you bless at 98%.

[Phunkie]

This is horribly wrong.

It's (Base Percent)^(Number of times you repair) for getting all the repairs successful.

Say you repair at Nerys (95%) five times (your weapon is at 0/5 durability). Each point is 95% success. But five points is:

0.95^5 = .7737

Basically almost 77.4% of getting all those points repaired without fail.

[Andy-Buddy]

Darn, I had a spreadsheet of this a while back, too. :l

Here it is.

Repair.xls (http://puu.sh/10sM)

Notes: The blessing stat is unconfirmed. leave it on no for an accurate representation.

[MrpewX2]

Blessing halfs the chances of fail, I worked the numbers with ferg

[Phanterz]

Blessing halfs the chances of fail, I worked the numbers with ferg

confirm by popular belief

[Andy-Buddy]

Blessing halfs the chances of fail, I worked the numbers with ferg

This is how I set my calculations in the Excel spreadsheet, so take that how you will.

[Nyo]

Darn, I had a spreadsheet of this a while back, too. :l

Here it is.

Repair.xls (http://puu.sh/10sM)

Notes: The blessing stat is unconfirmed. leave it on no for an accurate representation.

Here is my formula that calculates the chance of failure, appended to your spreadsheet. I didn't touch any portions of what you did, so feel free to clean it up.

http://puu.sh/143I

(wow that is a neat program)

[Grifter]

I said it in that other thread, and I'll say it again: Purrz needs to take a course in basic probability.

B> derep button

[Retard]

Idk maybe i am a retard. I keep re-running my logic in my head and i keep coming to the conclusion that
80% of the time you will have a perfect repair. T-T

Edit: Scratch the above. I think I know now but seriously no need to ask for a Derep Button ;~;

[casshem]

If you're doing 10 points at 98%, you will have a 81.7% chance of no fails. That's .98^10. Let's say you come back to repair 10 points again; now it becomes .98^20; that's a 66.7% chance of no fails on your second trip to the blacksmith. BUT, remember this: once the blacksmith fails, you stop adding to the exponent. So let's do a quick sum up:

You have a 0/10 weapon and you head over to Edern. The chance of Edern successfully repairing your weapon is 81.7%. Let's say he does not fail any points and you're happy your sword is now at 10/10, so you go spam some SMs instead of cutting your wrists.

After a while of spamming SMs, your sword is now 0/10 :T_T: The probability of Edern pulling off another successful FULL repair is .98^20 or 66.7%. Let's say you're at 5/10 going for 6/10 and he FAILS. You're now at 5/9 :T_T: The probability that Edern will not mess up any further now becomes .98^4 or 92.2%.

Wow-...that wasn't a quick sum up. Oh well

[qaccy]

If you're doing 10 points at 98%, you will have a 81.7% chance of no fails. That's .98^10. Let's say you come back to repair 10 points again; now it becomes .98^20; that's a 66.7% chance of no fails on your second trip to the blacksmith. BUT, remember this: once the blacksmith fails, you stop adding to the exponent. So let's do a quick sum up:

You have a 0/10 weapon and you head over to Edern. The chance of Edern successfully repairing your weapon is 81.7%. Let's say he does not fail any points and you're happy your sword is now at 10/10, so you go spam some SMs instead of cutting your wrists.

After a while of spamming SMs, your sword is now 0/10 :T_T: The probability of Edern pulling off another successful FULL repair is .98^20 or 66.7%. Let's say you're at 5/10 going for 6/10 and he FAILS. You're now at 5/9 :T_T: The probability that Edern will not mess up any further now becomes .98^4 or 92.2%.

Wow-...that wasn't a quick sum up. Oh well

Uh...why would the exponent be 20? You're repairing another 10 points, so it's again a .98^10 chance of having no fails. The success rate doesn't get lower simply because you've already full repaired once before...it's the same odds.

However, if you meant it as in figuring out the odds of, say, a full 0/20 repair with no fails, then the exponent would be 20.

[casshem]

It becomes .98^20 because you are going for 20 succeeds in a row. You only stop piling on the successes to the exponent once the successful repairs in a row streak is broken by a fail (as I demonstrated later on in my post). This is why so many people think the Mabi %'s are wrong.

Just to hammer the nail into the coffin:
Repair#(% success @ 98% )[S= success; F= fail] - #1(.98)[S], #2(.98^2)[S], #3(.98^3)[S], #4(.98^4)[S], #5(.98^5)[S], #6(.98^6)[F], #7(.98)[S], #8(.98^2)[S], #9(.98^3)[F], #10(.98)[S]~and so on.

[TheKartheus]

This all only works in probability, not in actuality.

It's like having babies. If you have 4 babies that are all boys, does that mean you have a 6.25% of having another boy child? No, you still have a 50% chance of getting a boy or a girl. The same concept works for this.

[casshem]

Probability is the guess of what will happen and actuality is what actually occurred. Probability is not affected by the last outcome, unless you want one outcome in a consecutive fashion.

[Omegatronic]

past events do not affect the probability of future events unless they eliminate a possible future outcome.

I can flip a coin 100 times and have it come up heads 100 times (past events), and the probability that the next flip will be heads is 50%.

[casshem]

I give up. Y'all's dumb.

[Omegatronic]

I give up. Y'all's dumb.

http://en.wikipedia.org/wiki/Gambler%27s_fallacy

[casshem]

You're not catching the message. You're saying the odds of a single toss is 50% no matter what, which is of course CORRECT. What I'm saying is: what are the odds that you'll get 10 heads OR tails in a row, consecutively, which would be .5^10 or roughly .001%.

When repairing a 10/10 weapon, you want 10 success in a row ideally. At 98%, you have a .98^10 chance, or roughly 82.%, of that happening.

What part do you not comprehend?

[WheeeE]

Oh nu basic probability turned into a huge post :(

[arklian]

http://en.wikipedia.org/wiki/Gambler%27s_fallacy


We can see from the above that, if one flips a fair coin 21 times, then the probability of 21 heads is 1 in 2,097,152. However, the probability of flipping a head after having already flipped 20 heads in a row is simply 1⁄2. This is an application of Bayes' theorem, From your link.


If you're doing 10 points at 98%, you will have a 81.7% chance of no fails. That's .98^10. Let's say you come back to repair 10 points again; now it becomes .98^20; that's a 66.7% chance of no fails on your second trip to the blacksmith.

The place where you went wrong is that you said "let's say you come back to repair 10 points again." This implies that the first 10 already occurred in the past and you shouldn't factor these in.

[ikotomi]

I can't believe people are still arguing over this stuff.


You're not catching the message. You're saying the odds of a single toss is 50% no matter what, which is of course CORRECT. What I'm saying is: what are the odds that you'll get 10 heads OR tails in a row, consecutively, which would be .5^10 or roughly .001%.

When repairing a 10/10 weapon, you want 10 success in a row ideally. At 98%, you have a .98^10 chance, or roughly 82.%, of that happening.

What part do you not comprehend?

The part that is wrong is that if you repair 10 more times after succeeding in 10 repairs, you do not say that the probability of succeeding all 10 times is .98^20. If it were an arbitrary 20 repairs, that would be true, but we already know that the first 10 repairs succeeded, so they do not factor into our probability calculation--there is no longer any randomness in the first 10 repairs.

To formulate the situation pedantically, we want to know specifically the probability that Edern succeeds 20 times conditional on the first 10 time being successes. The actually computation is 1^10*.98^10. This is a completely different question than what is the probability that Edern will succeed the next 20 times.

[casshem]

M'kay.

[Retard]

Ima lock this, we can discuss this in like a debate site er something XD

say your last words to eachother ;~;

[Aiyoku]

So can someone bottom line this? I got bored of reading all the mathematical nonsense I thought I'd never have to see again after finishing high school. Is it better to full repair or point by point it?

[casshem]

Point by point, as long as you bless your stuff if a fail should arise. If you don't, it makes no difference.