Tree of Savior Forum

Edo’s Random Stat Rerolling Guide


There is a lot of equipment in the game that have random stats. This guide will tell you about the different types of equipment with random stats, when you should care about them, what they’re used for, and how you might go about changing these stats to suit your needs. This guide won’t discuss what stats you should get, as that differs from build to build.

Random Stat Basics

All random-stat equipment begin as unidentified equipment before acquiring their random stats, but different random-stat equipment are acquired through different means. Each class of equipment also has different ranges and uses, which means that for some equipment it is worth the investment to ‘reroll’ the stats while for others it is not. In this guide I will refer to changing the stats on an item as ‘rerolling’ because every method has some randomness involved, so it is akin to rolling dice. The main reason you’d want to reroll equipment is so that you can make ichors. You can read my guide on ichoring here. Ichoring allows you to take the stats from some equipment and give them to certain legend-grade equipment.

Image: Various places you can identify at. A player store (top left), the appraiser master in Fedimian (top right), the blacksmith in Klaipeda (bottom left), and the blacksmith in Orsha (bottom right).

An unidentified equipment cannot be equipped and has no random stats. To make it usable, you must visit a player appraiser shop, the appraiser master in Fedimian, or either blacksmith in Klaipeda or Orsha and pay a fee to have the equipment identified. When identified, the equipment will acquire a random number of stats, of random groups, of random types, with random values. Supposedly, people have better luck with player shops.

Image: Unidentified equipment (left) and a random-stat equipment after identification (right).

The number of stats that a random equipment has is often referred to as the ‘lines’ of stats that the equipment has. Armor and 1H weapons can have 1 – 4 lines, 2H weapons can have 3 – 6 lines, and trinkets can have 1 – 2 lines. Each line must be a different stat, and players often value having more lines. Each line is preceded by a colored orb indicating the stat group. There are four stat groups representing different types of stats: attack (red), defense (blue), support (purple), and main stats (green).

Equipment Type Minimum Number of Lines Maximum Number of Lines
Trinket 1 2
Armor 1 4
Shield 1 4
1H Weapon 1 4
Subweapon 1 4
2H Weapon 3 6

Equipment Classes

In this section I’ll discuss only the highest-level random-stat equipment. Unidentified equipment can appear very early on in your ToS career, but most of it is not worth your attention and can be passed over. Any equipment that you don’t want can be dismantled at a blacksmith in town and turned into powder that can be used for other things. At best, low-level random-stat equipment can be identified and worn temporarily while leveling, but you certainly should not waste resources rerolling that equipment.

Level 400 Savinose Equipment

The level 400 equipment go by several names. The weapons all are called Legva weapons while the armors are called Akmo (plate), Pilgriste (cloth), and Vymedzai (leather). Berthas, Primus, and Savinose are just descriptors telling you the rarity of the random-stat equipment. Berthas and Primus are rare and unique, respectively, while Savinose is the highest rarity, the legendary equipment.

Level 400 Savinose equipment are semi-endgame equipment crafted from Planium or often given out as rewards in events. Planium is a rare drop from the level 391 – 420 maps. Savi equipment have lower maximal rolls compared to the most recent Dysnai equipment, and thus aren’t really worth rerolling. However, this equipment may be worth acquiring if you are a new player, as legendary equipment have the highest attack/defenses of any equipment in their level tier. When a Savinose is first obtained, it is unidentified and thus requires identification to have stats and be used. If this is your first legendary set, it may be worth rerolling a little just to get an important stat, like CON.

Image: A Savinose equipment. Note that it has a lot of extra stats besides its random stats due to various upgrades.

Level 400 Berthas/Primus Equipment

Berthas and Primus are unlike Savinose in that they are not crafted. Instead, they drop from fields as unidentified equipment. Berthas and Primus Legva equipment are not at all worth rerolling because their maximal stats are far lower than Dysnai equipment.

Image: A level-400 equipment.

Level 430 Berthas/Primus Dysnai Equipment

Berthas and Primus Dysnai equipment are the level 430-tier of random-stat equipment. Unlike previous level tiers, all armors and weapons bear the ‘Dysnai’ name. These were introduced with the Episode 12 update and have a much higher upper bound in stats compared to previous random-stat equipment. The stats on these are so high that rerolling any other equipment frankly looks like a waste of resources. Dysnai equipment are randomly dropped from mobs from the level 421+ maps, and can be obtained either through field farming or by clearing challenge modes. Berthas equipment can randomly roll values between 40% and 100% of the maximum value while Primus can roll values between 80% and 100% of the maximum value.

Image: A level-430 equipment.

Item Rarity Stat Range (Min~Max)%
Magic 10 ~ 100%
Rare 40 ~ 100%
Unique 80 ~ 100%
Legend 80 ~ 100%

Table: The stat ranges described in this developer blog post.

What to Reroll

You might consider rerolling level 400 Savinose equipment a little bit if you are newer and you are in dire need of the stats, but otherwise the only other equipment that really should be rerolled are Dysnai equipment. You’d want to reroll level 400 Savinose quipment to use as-is, but the reason why you’d want to reroll Dysnai equipment to make as an ichor for Skiaclipse Varna equipment.

Types of Rerolling: Re-Identification and Magnifiers

There are two types of rerolling in the game. The first is re-identification, which costs nucle and sierra powder. The other uses magnifiers, which changes the stats of random-stat equipment in many different ways, depending on the magnifier used.


Item Re-identification semi-randomizes the stats based on the existing lines/groups. Re-identification does not change the number of lines, but changes the stat and/or value (though it can reroll into the exact same stat and value). When rerolling the stat, it will reroll within the group. For example, an attack stat will remain an attack stat. The group is indicated by the colored orb next to the stat line. Re-identification changes all lines at the same time.

Image: Talking with the appraiser master to re-identify.

Re-identification is performed at the appraiser master in Fedimian, or at a blacksmith at either Klaipeda or Orsha. It costs some amount of sierra powder and nucle powder, depending on the level and rarity of the equipment being re-identified. The cost goes up with level and rarity . We frequently have re-identification discount events, so consider planning around those!

Image: Re-identification UI for three different equipment. Note the differing costs.

The sierra and nucle powder that you need to re-identify equipment is obtained as a drop from farming field mobs in the 420+ areas or as a drop from challenge modes. They are also obtained when dismantling equipment at a blacksmith. Nucle powder also happens to come from dimensional collapse.


Magnifiers are single-use items that can change the stats of random-stat equipment. Each magnifier affects your equipment in a different way and are obtained in wildly different manners. They are usually put under the TP tab. Though each magnifier can be obtained from normally playing the game, many of them also often show up in the TP shop or in events. All Sandra’s magnifiers can be used only on equipment level 430 or above (Dysnai equipment).

Mysterious Magnifier

Mysterious magnifiers are the most basic and random magnifier. It randomly changes all your stats, and can even change the number of lines. It is one of two magnifiers able to affect the number of lines. Mysterious magnifiers are typically obtained from challenge modes, emergency quests, and Goddess’s Grace events.

Artisan Magnifier

Artisan magnifiers are the same as mysterious magnifiers, except with the option of choosing between the current random stats or changing to the new stats. Artisan magnifiers are typically obtained from challenge modes, emergency quests, and Goddess’s Grace events.

Sandra’s Magnifier

Sandra’s magnifiers change the stat, value, and category of a line. It can affect only existing lines and is unable to add or remove lines. Multiple lines can be selected and rerolled at the same time. Sandra’s magnifiers are typically obtained from emergency quests and Goddess’s Grace events.

Sandra’s Detailed Magnifier

Sandra’s detailed magnifiers change only the value of a single line of your choice. Detailed magnifiers are typically obtained from emergency quests and Goddess’s Grace events.

Sandra’s Perfect Magnifier

Sandra’s perfect magnifier changes the value of all lines to the maximum value. They can only be used on equipment or ichors with 4 or less stat lines. Recipes for perfect magnifiers are obtained from Dimensional Collapse Points and require a large amount of detailed magnifiers.

Sandra’s Ultimate Magnifier

Sandra’s ultimate magnifier changes the value of all lines to the maximum value. They can only be used on equipment or ichors with 6 or less stat lines. Recipes for ultimate magnifiers are obtained from Dimensional Collapse Points and require a large amount of detailed magnifiers.


Rerolling Process Flow

When rerolling equipment players are typically working towards ideal or near-ideal equipment. This is because the process of rerolling can be extremely expensive, and aiming for anything but ideal equipment is likely a massive waste of resources. The process I outline below is designed with this in mind, though some adjustments could be made depending on your goals and resources. Below is a flowchart covering the general idea approach toward rerolling equipment to make ichors.

Choosing an Equipment to Start With

Making amazing random-stat equipment/ichors begins with acquiring a piece of equipment to work on. Depending on your time and resources, you might go farm and try to drop the equipment out in the field or buy it from the market. Also, your luck might put you in a different part of the process; if the equipment has most of the parts you want already, you might skip some of the earlier steps.

An early choice for you to make is to decide whether you want to work with a Primus or a Berthas equipment. This choice affects you initial investment in the process as well as the costs later in the process. Primus are generally more expensive than Berthas because they tend to have higher stat values. This means that if you want generally high stats, primus equipment will require less detailed magnifiers.

If you want perfect stats, you might want to commit to the idea of using a perfect/ultimate magnifier. If you decide to commit to using a perfect/ultimate magnifier, then you should start with Berthas. These magnifiers will grant perfect stats regardless of whether the equipment is a Primus or Berthas, so the generally higher stats of Primus adds no benefit. Since that is the case, go with Berthas since they’re simply cheaper and cost less powder to re-identify.

Setting the Number of Lines

After choosing a piece of equipment, the next goal is to obtain the maximum number of lines. This is done by rerolling using either mysterious or artisan magnifiers. No other magnifier affects the number of lines, so this is the only option. You might be tempted to use these magnifiers to also obtain the appropriate stat groups, but the likelihood of getting the maximum number of lines appears to be so low that you’d quickly wish you just went to the next step.

Image: Rerolling using a mysterious magnifier (left) and rerolling a mysterious magnifier (right).

Setting the Stat Groups

Even though you may have the maximum number of lines, it is likely that some lines are not the ones you are seeking. If that is the case, use Sandra’s magnifiers (the normal blue ones) to change the lines you don’t want. If you have a very specific set of groups that you want, then always keep the correct groups and only reroll the incorrect groups. For instance, if you want 3 green and 1 purple stat but the equipment is at 1 green, 1 purple, and 2 blue, then only change the 2 blue.

Image: Rerolling using a Sandra’s magnifier.

If you have multiple group distributions that you are okay with, then you might consider rerolling already-correct groups, but you risk using more Sandra’s magnifiers.


At this point you take the equipment over to the appraiser master or to a blacksmith to re-identify. Take some nucle powder and sierra powder with you and be ready to re-identify a lot. At this step it may be helpful to have multiple acceptable options for a stat group, especially if you’re looking at a big stat group. The main stats group (green group including CON, STR, INT, etc.) is the smallest group (it has 5 stats) so it is reasonable to aim for exactly one stat spread for this group. But for a bigger group like the support group (purple, with 10 stats), it may be useful to have a handful of combination of stats you’d be interested in. For instance, if you are working with 2 purple stats, you might choose to settle for either evasion/accuracy or evasion/critical rate.

Image: Re-identification UI.

Something you should really consider is stopping on something that is good, but not what you desire. There is a pretty good chance that you’d find a roll that is amazing for someone else but useless for you. In that case, you should consider stopping and seeing if you could sell it to someone and fund a new attempt. Rerolling often can get very expensive, so you should take good luck where you can get it!

Optimizing Stat Values

Once you have a good stat spread, the next goal is to get good stat values. This means using detailed magnifiers on any stats you consider too low. Alternatively, this could also mean using a perfect or ultimate magnifier if you’re looking for perfect stat values.

Image: Changing stat values using a detailed magnifier.


Once you get the stats and values you want, it’s time to ichor! You can read my ichoring guide here. This effectively ‘locks’ your stats for the most part, as the only way to change the stats of an ichor would be through a perfect or ultimate magnifier. Those are not easy to obtain, so be sure you’ll be happy about ichor before making it!

Appendix I: Stat Table

This is the table of maximum stats according to this news announcement. The maximum value differs depending on the stat and equipment that it is on.

Stat/Part Gloves/Bottom/ Top/Shoes Shield/Sub/One-handed Weapon Weapon Accessory Two-handed Weapon
Attack against Cloth Armored Target 530 993 497 1489
Attack against Leather Armored Target 530 993 497 1489
Attack against Plate Armored Target 530 993 497 1489
Attack against Small-type Target 530 993 497 1489
Attack against Medium-type Target 530 993 497 1489
Attack against Large-type Target 530 993 497 1489
Attack against Ghost Armored Target 530 993 497 1489
Attack against Plant-type Target 530 993 497 1489
Attack against Beast-type Target 530 993 497 1489
Attack against Devil-type Target 530 993 497 1489
Attack against Mutant-type Target 530 993 497 1489
Attack against Insect-type Target 530 993 497 1489
Additional Damage 794 1489 745 2233
Attack Offset against Medium-type Target 530 993 497 1489
Attack Offset against Cloth Armored Target 530 993 497 1489
Attack Offset against Leather Armored Target 530 993 497 1489
Attack Offset against Plate Armored Target 530 993 497 1489
Critical Resistance 265 497 249 745
Add. Damage Resistance 794 1489 745 2233
STR 80 149 75 224
DEX 80 149 75 224
CON 80 149 75 224
INT 80 149 75 224
SPR 80 149 75 224
Stamina 19 35 18 53
Looting Chance 80 149 75 224
Critical Rate 265 497 249 745
Block Penetration 265 497 249 745
Block 265 497 249 745
Accuracy 265 497 249 745
Evasion 265 497 249 745
HP Recovery 265 497 249 745
SP Recovery 265 497 249 745
Max SP 530 994 498 1490

Appendix II: Combinatorics, Probability, and Some Case Studies

In this section I’ll go over some basics of the math involved in calculating reroll odds and then look at some specific cases in order to help support the rerolling process flow that I presented earlier. In order to try to do some of these calculations, I make some assumptions about how IMC coded the probability of certain events because I wasn’t able to find anything conclusive. Generally the assumption is that IMC went with the simplest way of doing things.


Combinatorics is an area of math mainly concerned with counting. We are concerned with combinatorics because we want to enumerate all the different combinations of possible stats in order to calculate the chances of finding the ones that we want. I’ll begin with counting some simple things and generalize that to counting any combination of stats.

Before we begin, let’s talk about some abbreviations that I am going to use to quickly describe an equipment. When talking about the stat groups on an equipment, I will use a number and a single letter to describe the number of lines belonging to a certain group. Each group is then separated by a slash. So a top with 2 greens, and a purple would be a 2G/1P. If I had undesirable stat groups, I’d indicate that using an X. So if I wanted a 3G/1P but instead have a 4G, I could also describe it as being a 3G/1X. The table below summarizes the groups, the number of unique stats in that group, their colors, and their abbreviations.

Stat Group Number of Stats in the Group Color Abbreviation
Main Stats 5 Green G
Attack 13 Red R
Defense 16 Blue B
Support 10 Purple P
Undesirable n/a n/a X

Table: The stat groups and abbreviations.

Let’s begin by looking at the smallest group, the main stat group. Looking at the table in Appendix I, we find that the main stat group has 5 stats: CON, INT, STR, SPR, and DEX. If we’re trying to re-identify a 1G equipment, there are 5 green stats to choose from. So obviously there are 5 possible rolls.

Let’s instead consider a 2G equipment. We go across each line and choose the stats for each line, one at a time. The first line has all the main stats available to it, so there 5 options. Since one stat has been taken by the first line, the second line will then have 4 options. Choosing stats in this fashion, there are 20 different permutations of two main stats. In combinatorics we can represent this as P(n,r), where n is the size of the set and r is the size of the subset. So for our 2G equip, it’d be P(5,2) = 20. Explicitly, the permutation function is:


Unfortunately, permutations are over-counting. When counting permutations, order matters. So a CON/STR roll and STR/CON roll are counted as being different, though we shouldn’t care about the order in which we get our stats as long as we get them. So what we do is divide by the number of different ways we can organize a given number of stats. For two stats, there are two way they can be organized. So if we’re picking 2 stats out of a group of 5, there are 10 combinations of stats. We can represent these combinations as C(n,r). For our example, that’d be C(5,2) = 10. The equation for this is:


So far I’ve been using n! to denote the factorial of a positive integer n. That is the product of all positive integers less than or equal to n:


You can quickly calculate the number of ways you can arrange something using factorials. For instance, if you had a piece of equipment with CON/STR/DEX, there are 3! = 6 different ways of arranging those stats. These are:


Now that we have the basic methods for counting combinations of stats, we can mix and match these to quickly count much bigger sets. For instance, if we wanted to know how many different combinations of 2G/2P stats there are, we can find out how many combinations of 2G stats exist, how many combinations of 2P stats exist, and find the product of those numbers. That comes out to:


We can extend this to any number or combination of stats. With the ability to enumerate how many different stat combinations might come up, we need to make a few assumptions in order to calculate the chances of getting what we want.

Probability and Assumptions

To calculate the chances of desirable rolls, we need to know the chances of individual events. Because datamining hasn’t found exactly how IMC decides random rolls when items are created, we will assume that different stats are equally likely across all groups. This is the simplest case, and I suspect that the typical player wouldn’t be able to tell the difference between doing it this way as opposed to another way (it’d require a sample size far beyond most people’s means). A reasonable alternative is to imagine that the game first rolls for the stat group then rolls a second time to choose a stat within the group, but that can be explored independently.

To expand upon what I mean by “different stats are equally likely across all groups,” let us start by imagining I have a piece of equipment with a single stat, and that stat is undesirable. I decide to use a Sandra’s magnifier on it (a terrible idea). Since there are no other stats than the one stat that is being rerolled, the game can choose from the entire pool of 34 stats. The assumption is applied here: each of the 34 stats will be equally likely to come up, so if I am looking for one stat then there is a 1/34 or 2.94% chance of finding the stat that I want.

One implication of this assumption is that if I were just looking for a particular stat group, then bigger stat groups are more likely. With 5 different stats, the main stat group is only 14.7% of all the possible stats while the attack stats nearly triple them with 13 stats, 38.2% of all stats.

The most important implication of assuming that each stat is equally likely is that when re-identifying, all unique combinations of stats are also equally likely. Take our earlier example of a 2G/2P equipment. There are 450 different combination of stats for 2G/2P equipment. That means if we want exactly one combination of stats, there is a 1/450 chance of getting it when re-identifying. With this in mind, we can go from just counting how many possibilities there are to estimating the likelihood of finding what we want.

Case Study: Re-Identifying a 3G/1P

Imagine I have a 3G/1P equipment that I want to re-identify to get CON/STR/DEX/critical rate. How many attempts at re-identification do you expect me to make to get the stats I want? What is the expected cost of attempting this?

We first calculate the number of combinations available for this scenario. That is choosing 3 main stats from a pool of 5 and choosing 1 support stat from a pool of 10. That is:


There are 100 different combinations of stats that can come from re-identifying a 3G/1P. That means there is a 1/100 chance of arriving at the stat spread that I seek. One reroll costs 880 nucle powder and 63 sierra powder, so I should expect to need to spend 88k nucle powder and 6.3k sierra powder, on average, to get the stats that I want.

Now imagine I got the main stats that I wanted (CON/STR/DEX) but did not roll the support stat that I wanted (critical rate). Should I use a Sandra’s magnifier, or should I continue to re-identify to get the stat that I want?

If the first three stats are fixed, there are 31 available options for the 4th stat. We can simply say that there is a 1/31 chance of choosing the stat that I want. If that is the case, then it’d on average require 31 Sandra’s magnifiers to obtain that desired stat. At 4m silver per magnifier, the expected cost of this approach is 124m silver.

When rerolling or re-identifying, we consider each attempt an independent random event. That means that even though I may have spent some money rerolling and I almost got what I wanted, I’d still expect to need 100 re-identification attempts to obtain the stats I want. I essentially pretend like all the previous attempts never happened. If I price nucle powder at 400 silver and sierra powder at 100 silver, that makes re-identification have an estimated cost of 35.8m silver. Re-identification appears to be significantly cheaper than using Sandra’s magnifiers to correct the last stat.


curse of every OCD player, especially seeing that 1 imperfect lines

Some errors, intentionally ignoring the minor stuff like capitalisation rules and grammar:

“A player store (top right), the appraiser master in Fedimian (top right)”

" Magic 10 ~ 100%" -> does not exist among lvl430+ gear and all those values in the spreadsheet do not apply to lower equips

“…or as one of the rewards from the ore pouches obtained from guild hangouts” -> no longer available

Also a comment regarding probabilites:

I am working now on the last piece of an 6-ichor-set with a strict specific set of 4 stats (2G/1R/1P on all items).
By doing so I noticed that without exception getting the specific R-roll was by far the hardest to get.
The way it went everytime was getting the right 2G and 1P was not easy but doable.
Then working on the last specific 1R usually with Sandras - I happen to get more G quite often, suspiciously often.
But that should not be the case when all stats had the same prob and 2 different G are already taken.
Talking about that with some mates I had to agree with some that you most likely get a 1/4 chance of a group followed up by an extra roll within the group.
Take this with a grain of salt but that would be my observation after more than 200 Sandra rolls to get 1 last specific stat.


Just wow, wish I have had this kind of guide weeks back, had to learn by literally asking around

Great guide, honestly. Since ichoring is a pain. There were great points I learned especially with the perfect magnifiers…too bad I spent hundred thousands nucle already

1 Like

Thanks for pointing out those mistakes. I went back and changed most of those things. I’m going to keep the range for magic random stats, even though those don’t exist for 430 equipment. IMC provided that table and I think it’s probably coded into the game, if those were to be added at some point.

I was working on this guide quite a while back but things kept happening and it got delayed. Sad times.


For 430 equipment, the only difference is Berthas (Rare): 20% - 100% to compensate from having no “Magic” tier. You could split the table in two (level 400/level 430), although I don’t think anyone would reroll anything other than 430 anymore.

Nice guide :smiley:

This is the way I got my rolls. I’ve just got to farm/buy enough sandras to correct one stat on two separate pieces and then I’ll have (more or less) perfect rolls on all of my gear.

but the chances on getting the 4th desired stat are higher if it is one from a smaller pool like e.g. the green one. Let’s say I have 3 lines I like, 2 green ones and one purple.
If I roll with a sandra magnifier, once it gets into the green pool,there are only 3 options left, each of them having 33% chance to appear.
of course this could be wrong if every stat had the same chance to appear regardless of group, but the item description specifically states that it changes the group, meaning the group is determined first.

Regardless of how high the chance for each group is, the pool from which can be drawn is already limited by your other stats, so you have a statistically higher chance to get the stat you want by using magnifiers(1 in 31) instead of rerolling the complete equipment(6 in 600), which can easily result in a complete disaster once you already have the stat you need from within a larger pool like for example critical rate.

So in my opinion it is more safe to use magnifiers if the stat you need is from within a smaller pool and if you already have the stat/stats from larger pools set (crit rate/evasion/block for example, or the medium target attack offsetting, which is a great mean in PVE to further decrease the damage taken) than rerolling possibly hundreds of times to get that one single perfect equip setup on 4 stats.

Sure, if you have 4 green stats, it’s better to just go to the identifier to re-identify your equipment, but once you dip into the territory of purple and blue stats on your item, you need either a hell of good luck or enough resources to try over and over again. Given that you also need powders for Ark upgrades, I’d rather sell the powder and buy magnifiers to try once in a while for the one stat that is still not ideal one than hoping for the perfect roll on all stats.

You didn’t explore this possibility at all in your calculations:
resetting stats on equip till 4 green stats, then rerolling for powder till you get all greens you want and then replacing the last one with magnifiers.

Yes, the 1/100 chance sounds tempting, but once you realize that there are only 6 viable outcomes in a pool of 600 possible items, and that the 1% chance only applies reliably on average within a pool of 10.000 tries, I wouldn’t want to risk that item on a mere 1% chance to get 100 perfect items in 10.000 tries.
By the way, this only applies in one session of re-identification, which most people don’t realize, since RNG via random number generator works differently than normal chance calculation.
So if you do not do these 10.000 tries within one session, you have either higher or lower chances on achieving the setup you want, since the random numbers are changing in the background,thus making the moment of time you try to achieve the rolls more impactful if you only attempt a few times, as it’s more likely to always re-identify at times that give you a bad outcome the fewer tries to accumulate materials for.

I just personally farmed and stocked on nucle and did it the other way around… (rerolling with discount) and on top of my head at a rough estimate i would get my desired str con crit eva every 100k nucle (with dc)

The description says, “… change its stat, group, and value…” which doesn’t necessarily imply that the group is determined first. It just says that all three of these properties of the stat line can be changed, which is still true when I say that all stats are rolled with equal likelihood. I also must remind you that this is an assumption and not a fact - I don’t think there is conclusive evidence regarding whether it works one way or another and I wouldn’t take IMC’s item description as conclusive evidence since their translations aren’t the best.

Personally I’d be confident in how the stat is rerolled if either someone datamined and actually could pull the code that is used to do the rerolling or if someone went out of their way to do use Sandras to reroll the same item the same way hundreds of times (I’d have to do a bit of thinking to decide what is an appropriate minimal sample size).


I don’t explore the idea of using Sandras to get four greens, re-identifying, then using Sandras because it seems like an awfully roundabout way of converting the groups, especially if one assumes that each unique stat is equally likely to be seen. If you were rerolling a 4X into a 3G/1P, you’re looking at a pool of 46376 different stat combinations with a distribution of possible results being something like:

G P X count %
0 0 4 C(19,4) = 3879 8.4
0 1 3 C(10,1) x C(19,3) = 9690 20.9
0 2 2 C(10,2) x C(19,2) = 7695 16.6
0 3 1 C(10,3) x C(19,1) = 2280 4.9
0 4 0 C(10,4) = 210 0.45
1 0 3 C(5,1) x C(19,3) = 4845 10.4
1 1 2 C(5,1) x C(10,1) x C(19,2) = 8550 18.4
1 2 1 C(5,1) x C(10,2) x C(19,1) = 4275 9.2
1 3 0 C(5,1) x C(10,3) = 600 1.3
2 0 2 C(5,2) x C(19,2) = 1710 3.7
2 1 1 C(5,2) x C(10,1) x C(19,1) = 1900 4.1
2 2 0 C(5,2) x C(10,2) = 450 0.97
3 0 1 C(5,3) x C(19,1) = 190 0.41
3 1 0 C(5,3) x C(10,1) = 100 0.22
4 0 0 C(5,4) = 5 0.011

Again, we’re assuming that each stat is equally likely, so the percentages should be interpreted with that in mind. The biggest caveat of this analysis is that this is just on the first roll. So we cannot say that it’d take 46376/100 = 463 Sandras to get from 4X to 3G/1P because that’s a ridiculous number. That’s disregarding gradual progress towards the goal, like rolling a 4X into 1G/3X before eventually rolling into 3G/1P. The pool gets narrower as we start getting the stat groups that are wanted.

What we might infer is that getting something like 4G is probably really unlikely in general, so asking to roll into that and then asking to roll out of it seems like it’d use up a lot of Sandras. If you happen to get it, that’s one thing. But if the plan is to get a rare stat distribution and then roll out of it, then it’d probably not be a good use of resources. Once again, this is assuming that all stats are equally likely.


Let’s play with the assumption that all stat categories are equally likely and revisit the 3G/1P case study that I put in the last section of the guide. If I have 3G that I want and a 4th stat that I don’t, what happens? Well, let me first note that this case encapsulates having 4G with one green stat that you don’t want (so something like CON/STR/DEX and an INT that you don’t want).

If your goal is to get crit rate, there are 10 utility stats to choose from. You also need to roll the group you want, out of four groups. So using a Sandra’s magnifier on the fourth stat, when you want a specific utility stat, gives you the following chances:

1/4 * 1/10 = 1/40 = 2.5%

This is actually worse than the 1/31 = 3.2% that I had found assuming the other way of rerolling. So if rerolling did the group first, then it’s worse to use Sandra’s for larger stat groups while being better for smaller stat groups.

I just want to point out that for 6-stat items and stat spreads where your goal is to have a lot of reds/purples (at least 2) it appears that it may be more favorable to start with an excess of reds/purples, re-identify while you have an excess of reds/purples to get most of the reds/purples you want. Then you go back and tweak the remaining lines using Sandra’s Magnifiers.

The reason you’d want a different approach is that the pool of reds/purples is so huge that it is extremely unlikely to get exactly what you want. So then you want to have extra red and purples lines to increase your odds of getting exactly what you want. 6-line equipment, on the other hand, is a matter of just having too many possibilities. Rolling and getting 4/6 or 5/6 of the stats you want is already extremely difficult. At that point you might want to lock the good luck you’ve had so far and not touch those stats by using Sandras.

1 Like

hey with free sandra event, i got 2x int con eva crit and 2x str con eva crit and str con dex eva crit +random ( i need more sandra for red stat) , this fediman pot event can abused with many alt account and trade weapon for free sandra , red line have huge pool but many of them is good like beast, medium, insect, cloth, devil after have this stat just stop rerolling with sandra , just sell that weapon/ichor and buy what ichor u need with silver,

  1. Sandra is good if free
  2. dont buy sandra blue for gamble , it is a trap

Hair acc is more difficult to roll for AoE attack and movement stat, roll 200 times still suck