Act! Addict! Actors! — Predicting Events and Counting Cards
The time I attempted to count every card in the game as well as every event start date for the next 3 years.
Hello! In this part of my A3! x Data series, we’ll be looking at my attempt to compile all of A3’s event information. If you’re not familiar with the A3! game, do check out this series’ prologue for a quick explanation as well as the first part of the series, which goes through compiling and analysing all of the characters’ basic information. The second part of the series goes into teambuilding tactics, and actually the tables I go through here are the root of how I determined each character’s “metaness”, while the third part which analyses past event results can be found here!
Before we begin, do open the full spreadsheet to save your eyesight and view the most up-to-date version of the tables! If you’re in a hurry/not in the mood to see a giant wall off text, scroll to the end for the first 2 weeks of my Sankaku! Spreadsheet! Saturdays! series, which attempts to make my A3! x Data series more Instagram-friendly…
Predicting Event Start Dates
The Event List table is probably the table which received the most revisions over the course of my A3! x Data project. The first version happened when I was trying to see my free-to-play chances at getting more SSRs of my favourite characters… and honestly there weren’t a lot. I realised pretty early on that the Rookies have it much worse, since that had a lot of gacha cards. I’ll go back to this point later on in the card-counting sheet! All the information is already on the Yaycupcake wiki, but I collected them in a table so I could easily sort and count events based on their event SSR character. I collated attributes such as the event name, type, start dates and SSR cards from both the event prizes and gacha.
The second version was made shortly after I realized that the gaps between recent events have been really inconsistent. For example, there was a 10-day break between Conquering Misoshiosa Island and Various Vows of Love, but the breaks before Knights of the Round IV On Stage and Meet Me by the Milky Way only lasted for 4 days. Thus, I came up with a fairly basic model that assumed consistent 4-day breaks between events, with each event lasting for 10 days each (except for Rerelease events which last for 5 days each). This combined with my ongoing panic about gem budgeting resulted in sticking with this scary estimate and creating an Event Calendar which showed my event start projections for 2021 in a prettier format, as well as all the characters’ birthdays. Do note that subsequent updates are only reflected in the Event List, so the Event Calendar is already a little bit out of sync. Oops.
The full spreadsheet shows the events for 2020 too, because I wanted to see if there was a trend in breaks between previous events. The number of days was inconsistent, but the average seemed to be about 6 days of break between events. However, I decided to assume 4 days between events for my projection to simulate the most painful timeline for my quest to hoard gems and to reflect how it felt like recent events were released more quickly than before.
Observations from the projection:
- Good luck to those who have a catch-them-all approach to event SSRs because the calendar looks really painful. Barely any breaks from event hell…
- The assumption of a 4-day break between events means that the breaks tend to fall during weekends, while Mondays to Wednesdays are consistently highlighted as part of event hell. Most events would start on Mondays and end on Wednesdays using this schedule. On the bright side this means that we can go all out for the Friday/Weekend daily practices, but real life commitments exist.
- If this fast-paced timeline continues, we’ll be seeing the next New Year’s event in October compared December for last year’s, but the timeline seems to sync with how the 1st Anniversary event was in October last year. This also means that the event order is sort of accelerated by two months, as seen from how the next Valentine’s Day event might happen in November instead of January. Well… Halloween happened in the summer…
- Caveat: I based the projected event order on JP, but we’ve already seen that EN doesn’t exactly follow this order, especially when it concerns Rerelease events and Episode Promo Campaigns. So don’t be surprised if a Rerelease event suddenly drops and wrecks the timeline… in general these last for 5 days after a 3-day break following the previous event. As for Promo Campaigns, I don’t think they will release Episodes 9 and 10 this year since the rookies are still pretty new, but who knows for next year!
- Another caveat: Since making the prediction model, we’ve seen 5-day breaks and an unusual week-long breaks, bringing back hope of a decent break between events. I’ve never been so glad to be wrong about my data…
The full spreadsheet also has a blank calendar template for 2021! Version 3 of the list colour-coded the spotlight SSRs by attribute to help with meta planning, as well as the projected start and end dates for all future events up to JP! Past event end dates have been added too, and based on my early estimates, it seems as if 2021 JP events will only hit EN in 2024. After I unlocked the “Conditional Formatting” skill while making the card-counting sheet, I modified the event list to automatically grey out cells describing events which have ended in EN, so that it’s easier to locate the most recent event for EN!
Counting Cards
The Event List also includes my first attempt at counting A3’s SSR cards:
Back then, I didn’t know much about Excel formulas so this table’s values were the result of a very long counting session. Not my brightest idea, but the initial numbers were accurate even after applying COUNTIF functions later on, surprisingly. This table was what gave me the impression that the rookies’ SSRs were more gacha-biased, due to the lower average Event/gacha ratio for their SSR cards. Spring and Winter Troupes are noticeably event-biased, while Autumn Troupe really liked the gacha, which was interesting to note especially after considering how Autumn’s popularity is one of the highest in the fanbase! However, do note that popularity isn’t a good indicator of how gacha-biased a character is! I plotted a scatter plot of the Gacha/Event ratio rank (Chikage #1, Sakuya #24) against the popularity rankings from the 2020 A3! EN Playerbase Survey Results (Reddit version). Do note that at the time the popularity poll was made, the rookies were very new to EN and hence unfamiliar to most players.
Points in the chart were coloured by troupe, and we can see that Winter Troupe is generally less popular and more event-biased compared to the other troupes (leaning towards the top-right quadrant). However, the scatter plot’s shape and extremely low R-squared ratio of 0.007 show that it is most likely that there is no linear correlation between gacha bias and character popularity, despite the initial implication that less popular characters are more commonly found as event SSRs, since Liber would want people to spend on their favourites’ gacha cards. Something fairly interesting to note is how the characters with the most even ratios between event and gacha SSRs also have the highest overall number of SSRs, while those with extreme ratios don’t have as many SSRs.
Attempt 2 to count the cards was much more comprehensive: I created a table which counted every single gacha and event card regardless of rarity, all the way up to JP. Please note that I didn’t include cards which don’t fall in these two categories, such as login reward cards for special occasions. Here’s a work-in-progress shot of the Event Reward Cards sheet, the full version can be found here! It was really tedious to make especially when I decided to link every cell to the wiki page for the corresponding card so that Google Sheets would give a preview of the card art, but at least conditional formatting made it such that all I needed to do was enter the rarity and character name before adding the link.
Afterwards, I counted all the cards using COUNTIF formulas so that future entries would be automatically counted too, and because there were way too many cells to count by hand. More conditional formatting and charts finished up the sheets, and then rinse and repeat for the Gacha Cards sheet!
Firstly, I created two tables which summed the cards based on rarity and attribute, then generated charts from them. Here are the overall results:
As the first character we meet in the story and arguably the leader of Mankai, it really shouldn’t be a surprise that Sakuya has the most cards (but most of them are SRs as he doesn’t have that many SSRs)! The rest of the Troupe Leaders (A3ders) appear in the top 6, so they’re no slouches too… Actually, most of the actors have similar card numbers except for the rookies, who are a year behind the others in terms of card numbers. The attribute charts show how certain characters’ cards might be biased towards a particular attribute, such as Action Yuki vs Drama Yuki. Coincidentally, Yuki’s Drama link skills are his weakest, so the low number of Drama Yukis isn’t too concerning and this chart could be good for meta card hunting, which would bring us to the second part of this series!
Here are the separate results for event and gacha cards (sorted by in-game order):
My findings for these charts were already pretty nicely summarised in the pictures at the end of the article, so do check them out! Basically, the separated charts give you a deeper look at which character/attribute combinations are more difficult to find in free-to-play (hence event rewards), such as Comedy Banri cards. In terms of rarity, you’ll need to have some gacha luck if you really want SSR Chikage cards, as he only has 1 event SSR even up to JP, which is a bit problematic since Chikage can form pretty strong link skill builds. In terms of overall card numbers, Banri and Tsumugi appear very frequently as event rewards (notably they are the more popular troupe leaders), while Autumn troupe might possess more gacha cards than average. However, the numbers continue to change with every JP event, and since the numbers are relatively close I won’t declare a decisive winner and let the chart speak for itself!
That’s it for this part of the A3! x Data series, I hope you enjoyed it! As always, feedback and suggestions are really appreciated, more resources can be found here, and check my Instagram out if you want to see the Saturday posts on an actual Saturday… as promised, here is the Instagram-friendly version of this article!
