Lucky Shops encounter

Introduction

This rare encounter was first introduced as the Rat Town portal in Set 9.5.
Besides when the lucky shops can occur and that they can pull from both active and inactive traits on your board, its internal mechanics have remained a mystery.

After gathering data from both Ranked and controlled test games, we have uncovered some of the system's hidden behaviors. However, it is still not fully understood, and some questions remain unanswered. It's clear that a lot of thought went into designing it!

Disclaimer

• We present only information that we are reasonably sure about. However, there could still be mistakes or errors, so you do not have to believe everything!

• Percentages shown from our data are illustrative only. We were unable to reverse engineer the real odds with any accuracy.

• Some behaviors could be intentional mechanics or bugs.

When do lucky shops occur?

There is currently a 2% chance of getting this encounter. Starting on Stage 2, lucky shops happen once per stage as the first shop of a random non-carousel, non-augment round.

• They occur on the same round for all players.

• A "hack" animation plays across the screen, and the shop cards glow.

Special cases

• If your shop is locked, you won't see the animation at the start of the round.
  Instead you receive one free reroll to reveal the lucky shop, usable only on that turn. The +1 reroll is your only visual indicator that it is a lucky shop round, so don't miss it!
  You cannot manipulate the tailoring by changing your board before you reroll.

• If the lucky shop round happens while you are under the effect of the AFK augment, you will miss out on it. The following augments are disabled in this encounter: Delayed Start, BRB, Rigged Shop, Shop Glitch and Subscription Service.

Application

• If you haven't seen a lucky shop by round 6, it's guaranteed to happen on the PvE round..

• Lucky shops can appear on 3-1 or 4-1, so you could tailor your board on the PvE round before.

• Lucky shops can't appear on 3-2 or 4-2, so you can play any board on the round before.

• Even players in the top 4 might only see 4 lucky shops in a game (Stage 2 to 5), and they may be spaced far apart (e.g. 3-1 then 4-7). So think about how much you want to play around them, after considering the rules in the following sections.

Do lucky shops obey standard shop odds?

Lucky shops use the standard shop odds for level 3/4/5. However, certain rules take priority before rolling the cost of other shop slots. For level 6+, the shop odds are adjusted due to these rules.

• All levels: 1 shop slot is guaranteed to be a unit chosen from your board.

• Level 6/7: 1 shop slot is guaranteed to be a 4-cost.

• Level 8+: 1 shop slot is guaranteed to be a 4-cost, 1 slot is guaranteed to be a 5-cost.

• 4- or 5-cost units can appear naturally from level 5/7 as usual, but they cannot be chosen as the board unit until level 6/8.

• If the chosen board unit is a 4- or 5-cost, it will take the corresponding guaranteed high cost slot.

• The guaranteed slots appear in random positions, not fixed.

• The other shop slots roll cost then unit as usual, meaning that you can still roll other 4- or 5-costs or copies of the board unit.

• You cannot see more than 2 copies of the same unit per lucky shop.

• The table below from our Ranked data shows the observed appearance rates of each unit cost at each level, compared to the standard shop odds. To balance out the guaranteed high-cost slots, the odds of rolling them in the remaining slots seem to be reduced.

Application

• If you want to see a 4- or 5-cost early, you could push faster to level 6 or 8, e.g. on the PvE round in case a lucky shop happens on the next round.
  NOTE: The High End Shopping augment uses your actual level to determine the guaranteed high-cost slots.

• If you get an early 3-cost from Stage 1 or a 2-1 augment, you can play it alone to guarantee upgrading it using the Stage 2 lucky shop.

• If you get an early 4-cost (e.g. from carousel or Expected Unexpectedness), you cannot force it in the lucky shop until level 6.

• If you are rerolling multiple low-cost units (e.g. for Nitro), you can play 2 or 3 unique units to hit up to 5 board unit slots in the same lucky shop.

• On level 10 you can force 2 copies of a 5-cost in the lucky shop if it is the only unique board unit, which could help you reach 3★.

Forcing specific units

The lucky shop behaves a certain way when you have a small number of unique units on the board, which occurs often at low levels.

Here are the rules for 1/2/3-cost units on levels 3/4/5:

• 1 unique unit: 2 slots are guaranteed to be copies of that board unit (even a 3-cost on level 3, which is usually impossible).

• 2 unique units: 2+ slots are guaranteed to be copies from that duo. E.g. Vi+Draven could result in Vi/Vi, Vi/Draven or Draven/Draven.
  The distribution of 2 slots/3 slots/4 slots is roughly 30%/50%/20%.

• 3 unique units: Same as 2 units except the chosen duo is random. If you're very lucky, you can hit 2 pairs and the 3rd unit in the last slot as well!

You can play duplicate units to make your board stronger.
You can mix 1/2/3-costs freely below level 6. Cost-mixing at higher levels or with high-costs has not been tested.
The table below shows the results of playing 1 or 2 unique units of the same cost at each level.
Levels 3/4/5 are as described above.

How are lucky shops tailored to your traits?

Lucky shop units share at least one trait with the board you played in the previous round. They come mostly from board units and active traits, but inactive traits can be rolled too. There are some other possible effects but they are unconfirmed due to low sample sizes.
Occasionally, you will see violations of tailoring rules. They are rare (assume 1%) and the causes are mostly unknown.
Possible causes: bugs, fallback due to invalid unit/trait, intended small chance of random.
Disproven causes: too few traits, no valid units of that cost, influence from bench units.

• There does not seem to be a minimum number of units or traits required to get a full tailored shop.

• After rolling unit cost, it seems like lucky shop units are rolled using weights such as:
  % chance of board unit
  % chance of active trait
  % chance of inactive trait
  % chance of random

• If there is no valid unit of the cost that was rolled for a lucky shop slot (none match your traits, you already have 3★ of all eligible units, or all eligible units are out of the pool), a random unit of that cost will be rolled.

Application

• The system tends to give you more of the units and active traits that you are playing. This can make tailoring hard and more of a gamble when you don't already have the pieces you need. It can also feel bad if your board is upgraded but you can't or don't want to 3★.

• You can tailor for specific 5-costs very reliably. However, for other costs the odds of hitting a specific unit seems capped at around 70% even with perfect tailoring, so expect some RNG.

• If you already have Xayah and want to tailor for Leona (both 4-cost Anima Squad): Theoretically you should play only Anima Squad and Vanguard units, and bench Xayah so that she doesn't get priority from being a board unit.

• Unless you are desperate to tailor for one specific unit like the above example, it is fine to play more active traits because dynamic weighting will reduce the overall chance of rolling inactive trait units. But it does dilute the odds of each active trait and board unit.

Active trait bias

• The guaranteed 5-cost from level 8+ is from an active trait if possible (99% in the Ranked data, 100% in the test data). Other 5-costs are also overwhelmingly likely to be from active traits (97% in the Ranked data, 99% in the test data), so it seems like the intended chance is close to or at 100%.

• 4-costs are also very biased toward active traits (over 90% in the Ranked data), but the test data shows that the real weighting is more modest than for 5-costs (e.g. ~60% for 1 active trait with no 4-cost board units, ~80% for 2 active traits with no 4-cost board units).

• We do not have test data for lower costs without board unit bias, but it seems likely that they have different active trait weighting compared to high costs.

• The table below from our Ranked data shows how often units with at least 1 active trait were chosen for each cost at each level, when inactive traits were also available. Keep in mind that units with zero active traits were usually not played except at lower levels.

Other possible findings (low sample sizes)

• Lucky shop units seem to be chosen based only on their innate traits. If true, then holding an Emblem will not increase a unit's odds of being chosen, and having fewer remaining copies of it in the pool does not decrease its odds.

• The test data suggests that there is dynamic weighting depending on the number of things in that roll category (board unit, active trait, etc).
  E.g. with 1 active trait you may see ~50% active trait units, but with 2 active traits you might have ~80% total chance split between the 2 traits.

• There is some evidence that traits with more units on the board (regardless of breakpoint tier) are weighted more.
  E.g. 3 Syndicate active > 2 Vanguard active, and 2/3 Cypher inactive > 1/2 Bastion inactive.
  
  

Loaded Dice

This was an item that rolled a shop offering five units sharing a trait with the unit that it was used on. Like standard shops, it first rolled the unit cost for each slot (using the odds for the current level), then a unit of that cost. Unit odds within each cost were split equally, e.g. if there were 2 eligible 4-costs then they each had 50% chance to appear if a 4-cost slot was rolled.
Rat Town/Lucky Shops have often been compared to Loaded Dice, and there was speculation that they worked the same way.

• Lucky shops do not tailor all five slots to the same unit, but it is possible that one or more slots are either tailored to the same unit or roll separately to decide which unit to tailor to.

• The only Loaded Dice theory that we could test with our data was if all five slots roll separately to decide which unit they each tailor to. It became clear that this theory did not explain everything, but it could still be part of the foundation in some way.

• Our data unequivocally showed the existence of active trait and board unit biases, which cannot be explained by the way that Loaded Dice worked (all eligible units of the same cost have equal chance of appearing).

Board unit bias

• Board unit bias (on top of the guaranteed board unit slot) is well supported by both the Ranked and test data (~60% to ~80% inclusive of other effects).

• We tested comps where within the same cost, there were board units with one or more active traits and board units with zero active traits. The latter consistently appeared about as often as the former, showing that this effect transcends traits.

• The table below from our Ranked data shows how often board units were chosen for each cost at each level, when both board and non-board units were available at that cost.

Inactive traits

• Even with multiple active traits available, it is possible to roll units from inactive traits.

• The base chance of rolling units from inactive traits is low, but if they are board units then they have much higher odds, similar to board units with active traits.

• The table below from our Ranked data shows how often players were offered a lucky shop unit that was neither a board unit nor from an active trait when both were possible.

TL;DR

The Lucky Shops encounter is a complex system with multiple levers for different player levels and unit costs, and is influenced by board units, active traits and inactive traits. While there is a lot of RNG involved, there is more room for player agency than previously known.

• You can play around the timing of lucky shops a little on Stage 2/3/4.

• Every lucky shop contains at least one unit from your board, and you can force specific board units to appear by playing 3 or fewer units.

• On level 6+ there is 1 guaranteed 4-cost slot, and on level 8+ there is also 1 guaranteed 5-cost slot.

• You cannot be offered more than 2 copies of the same unit per shop.

• The units offered in lucky shops are largely biased towards active traits and board units.

• It is always possible to lowroll or experience bugs.

• We were unable to fully reverse engineer the system, so some questions remain unanswered.

  
  

Frequently Asked Questions

• Are active traits weighted more than inactive traits?
  Yes, especially for 4- and 5-cost units.

• Can I be offered units with no connection to my board?
  Yes, but rarely and usually due to unknown causes.

• Can I lose a lucky shop if I take the AFK augment?
  Yes, if the lucky shop round is on 2-2 or 2-3.

• Does reaching a higher breakpoint make a trait more likely to appear?
  Maybe. At the very least, the units on the board are more likely to appear. There is some evidence that playing more unique units from a trait (regardless of breakpoint) will increase its representation, over and above board unit bias.

• Do lucky shops work like Loaded Dice in any way?
  Maybe. It is plausible that Loaded Dice logic is the foundation of the trait tailoring system, but with added features such as increased odds for active traits and board units.

• Are bench units considered?
  No, only your board from the previous round.

• Do lucky shops use the shop odds of the level above?
  No, but the distribution may look different due to the overriding rules mentioned in the article.

• Is there a minimum number of units or traits required to get a full tailor?
  No, you only need 1 unit/trait as long as traitmates exist for the unit costs that are rolled.

• Can I lose a lucky shop if I lock my previous shop?
  No, unless you don't use your free reroll that round.

• Can I get 5 copies of the same unit (e.g. Zac) in a lucky shop?
  No, there is a limit of 2 copies per unit per shop.

• Can lucky shops exceed the pool?
  No, as far as we can tell.

• Does having fewer copies left in the pool make a unit less likely to appear compared to one with more remaining copies?
  No, as far as we can tell. But if the unit is highly contested then there may not be enough copies to offer to all of the players.