Refining for cheap

took

I wonder if the G16 gears follow the same success rate. Considering that they cost 5 instead 1 mirage, I suspect that their success rates are also changed. I haven't done any systematic experiments, but on my G16 weapon the rate seems clearly lower than those in the table. For example, it took me 80 trials to +2 with mirages, while the expected number is 10 according to the table; and I've tried 8 times to +4 with Tienkang and all failed, while the success rate is supposed to be 45%. I know my sample size is too small, so I want to hear from you guys. Any opinion?

WarrenWolfy - Sanctuary wrote: »

Thanks to the help and input from so many here on the forums, many of the questions about refining without Dragon Orbs have been answered. The tl;dr version is as follows:

The chance of success when refining without Dragon Orbs is:

From    To   Mirage  Tienkang   Tisha  Chienkun
 +0 --> +1:    50%      65%     53.5%    100%
 +1 --> +2:    30%      45%     33.5%     25%
 +2 --> +3:    30%      45%     33.5%     10%
 +3 --> +4:    30%      45%     33.5%     4%
 +4 --> +5:    30%      45%     33.5%    1.67%
 +5 --> +6:    30%      45%     33.5%    0.77%
 +6 --> +7:    30%      45%     33.5%    0.47%
 +7 --> +8:    30%      45%     33.5%    0.25%
 +8 --> +9:    25%      40%     28.5%    0.13%
 +9 -> +10:    20%      35%     23.5%    0.07%
+10 -> +11:    12%      27%     15.5%    0.04%
+11 -> +12:     5%      20%      8.5%    0.02%

- Generally speaking, refining armour up to +7 costs, on average, about 1/3 the price of Dragon Orbs, provided you do it correctly are willing and able to accept the risks involved!

- Refining weapons up to +7 costs, on average, about 1/2 the price of Dragon Orbs.

- Dragon Orbs typically become the cheapest way to refine when dealing with refines of +8 or greater.

- Dragon Orb Oceans, which can refine up to +10, are usually the cheapest option for refining +8 and +9 as well.

- Sale prices obviously have a huge impact on refining costs. The best strategy is to wait for a 50%-off sale on Tienkangs and Tishas and then refine up to +7, then wait for a 65%-off sale on Dragon Orbs to refine your gear higher.

- When not on sale, Tishas/Tienkangs can also be purchased for a discount by purchasing DQ items from other players, then exchanging them for Event Gold. The exact amount of savings will vary depending upon server prices.

- Choosing which stone is the most cost effective for each level of refine can be calculated by entering the current market price for Mirages, Gold, Stones, and Orbs into this spreadsheet: Zoho Sheet - Refining Guide

- A more feature-rich, Excel-only version of the same spreadsheet, that is also capable of running simulations of thousands of refining attempts, can be downloaded here: Excel Version - Refining Guide

- Duke Blacke's rumours regarding certain times of day, certain NPCs, certain techniques, certain types of equipment, etc..., having better odds for refining success appear to all be false. For example, the rumour that the Harshlands Officer gives better +1 refining odds to players who belong to the faction that owns the land was tested with several hundred Mirages and no improvement was found.

Good luck, and happy refining! (for cheap!)

(The rest of this post is the original question and theory that I posed to the forum. Please note that my 40% theory was eventually proven false! The correct numbers are now listed above.)

- end of EDIT -

HrunsPanda - Archosaur

As you said, your sample size is too small. Not just a little bit too small, your sample size is insignificant.

It doesnt seem logical that the chances would be changed. 5 instead of 1 or 2 is a simple enough change and is enough to make you want to use orbs for +3 and maybe even +2.

On a side note, a friend of mine actually just refined his first G16 item to +4 with only mirages. It took him 100 mirages.

Fissile - Archosaur

I have refined 5 g16 weapons to +5 without d orbs and I can say from experience that the success rates are consistent with the table in the original post within expected statistical variation.

Also, you may find the following interesting. Note that the widely accepted estimator for the probability of success in a Bernoulli trial (i.e. if there are n successes in m trials then the probability ~ m/n) is biased.
http://arxiv.org/pdf/1105.1486v1.pdf

Fissile - Archosaur

To expand on my above post, let's pretend I had never seen the table from the leading post in this thread. Let us also pretend that Asterelle's refining simulator is the real PWI refining system.

I decide I want to determine what the success probabilities are for refining without any refining aids. My test procedure is to simply refine an item repeatedly until it gets to +5 and stop. At each step I will record what level of refine I attempted and whether it succeeded or failed.

Here are my results from refining just 1 item (a total of 819 refine attempts). I present the number of trials and successes for each refinement level and the 95% confidence interval for the probability of success (i.e. there is a very high chance that the true probability lies somewhere between the Lower and Upper bounds, where my best guess for the true probability is the Mean):

From    To  Successes  Trials   Lower    Mean     Upper
 +0 --> +1:    239      469     46.4%    51.0%    55.5%
 +1 --> +2:     75      239     25.8%    31.5%    37.5%
 +2 --> +3:     31       75     30.9%    41.6%    52.7%
 +3 --> +4:      5       31      7.2%    18.2%    32.8%
 +4 --> +5:      1        5      4.3%    28.6%    64.1%

Thus, even after spending over 4,000 celestones refining a G16 weapon, I still wouldn't have a very good grasp on what the actual success rate is.

For those of you that don't like math, here is a spreadsheet that can do the calculations for you. Note it has a user-defined function (macro) in order to calculate the inverse of the regularized incomplete beta function.
https://docs.google.com/file/d/0B9m7SJi6d6dRMGJsS3pKelRWVTA/edit?usp=sharing

Elenacostel - Heavens Tear

How many people try to get to +8 without orbs? I'm looking for any success stories.

I just used the refining simulator a few times. One of the attempts to +8 cost 98m, using the ideal setup in the spreadsheet according to current market conditions.

Azura - Lost City

Elenacostel - Heavens Tear wrote: »

How many people try to get to +8 without orbs? I'm looking for any success stories.

I just used the refining simulator a few times. One of the attempts to +8 cost 98m, using the ideal setup in the spreadsheet according to current market conditions.

I have refined to +8 a lot of times. I have refined 3 different sets without using orbs.

HrunsPanda - Archosaur

I think you better just use the mathematical facts than asking people. Espescially when it is about chances, peoples percieved experiences (the emotions) dont line up well with the reality.

Just use this sheet
http://public.sheet.zoho.com/public/warren.forrest/pwi-refine-costs
dont forget to multiply the mirage cost with 2 or 5 depending on how many mirages are needed per attempt and you will have to modify the cost of dragon orbs to the correct prices.

I find it interesting BTW to see that many people are very risk averse to refine with mirages and start using orbs very early. (on items that dont cost 5 mirages) while at the same time people buy packs and take gambles worse than you can find in any casino.

Elenacostel - Heavens Tear

HrunsPanda - Archosaur wrote: »

I think you better just use the mathematical facts than asking people. Espescially when it is about chances, peoples percieved experiences (the emotions) dont line up well with the reality.

Just use this sheet
http://public.sheet.zoho.com/public/warren.forrest/pwi-refine-costs
dont forget to multiply the mirage cost with 2 or 5 depending on how many mirages are needed per attempt and you will have to modify the cost of dragon orbs to the correct prices.

Greetings, hero.

I am indeed using the spreadsheet provided in the opening post. That is, in fact, why I am posting on this particular thread instead of making a new topic in this subforum. I apologize for any ambiguity in my previous post regarding the relevancy of my post with regards to the opening post of this thread.

Zsw - Dreamweaver

Idk anyone who goes for +8, but i know several have gotton up to +11 without orbs, from +0.

The__Sun - Dreamweaver

I got my chest and wrists to +9 with tienkang and other armor pieces to +8 took about 79 event gold on average. One +8 was only 600 mirages but the rest was far more

Elenacostel - Heavens Tear

Thanx for the user stories!

Manamana - Harshlands

much appreciated info

Burnout - Harshlands

Was able to tisha 2 armors to +10, 1 to +9, a cape to +9, and 2 rings to +9 using this method.

I think my largest loss was 6000 mirages to get +8, but overall I'd say it's worth it if you can deal with the variance.