Best-in-slot (BIS) has been an issue long discussed since the beginning of TFT 5 years ago. Over time, the importance of BIS has been lowered, both as player skill has increased and as developers have tried to lower the power difference between getting good and bad items. There’s no other way about it: sometimes you get screwed by items in game, and those games really suck. Or at least, they used to. Item variance and output has generally been normalized, perhaps not to prevent, but at least to soften the blow of these situations. Instead of going 8th instantly when playing an AD comp with 3 rods, you can claw your way to a 6th!
You know, these aren’t even that bad.
However, this problem has somewhat remained with mana. Mana items are build-defining in a way raw damage has never been. While attack damage (AD) backliners can usually take a litany of assorted items, ability power (AP) backliners have been largely unplayable without some source of mana generation. All of their combat outputs are concentrated in their spell, and without being able to cast it multiple times a fight reliably, they simply cannot keep up with the innate high uptime of AD carries.
Furthermore, mana and damage differ mathematically. Damage comes from many different stats and sources, you can’t overcap on damage, and damage scales linearly. Mana’s place in TFT’s combat systems are at direct odds with each of these properties: mana has rather limited sources, mana only needs to hit specific thresholds, and mana generation is quantized. (Please excuse the gross oversimplification; each of these pillars has exceptions, but in general they tend to hold true for most units.)
At the same time, mana has historically been a space ripe for abuse. Mana lets AP champions participate in the game, but at the same time can easily remove other champions from doing the same. We’ve seen this throughout various sets—Shadow Blue Buff Ryze and LeBlanc, mana printer Sona, A.D.M.I.N., QSS Invokers Taric, Blue Battery and Axiom Arc—the list goes on. Mana has limited sources for good reason: spells are designed to be cast at noticeable intervals, and eliminating that time creates some less-than-stellar game states.
The difference in power between BIS mana generation and none was one of the largest in the game. This set, however, you may have noticed relatively few of these issues! Besides removing the particularly egregious edge case augments and traits, Riot has shifted mana design to lean into classes, changing units and items accordingly. In Set 10, mana generation was flattened across items. In Set 11, back line caster champions have maximum mana values to match their intended cast frequency. This, I argue (and to little dissent, tbh), has increased learnability and decreased the difference between mana items while still keeping their distinct purposes.
Classes are subcategories of unit classes that have similar outputs. Instead of forcing players to learn 16 different units and their precise mana itemization and various jibber jabber, classes allow players—and developers—to treat multiple units similarly. This increases learnability, as well as inter-composition flexibility by reducing opportunity costs for itemization.
The main difficulty with differentiating classes by maximum mana is in choosing the correct buckets. The impact of max mana varies based on class, most noticeably in backline AP casters. These units’ effectiveness depends heavily on multiple casts. They also don’t gain mana from taking damage, so thresholds are more fixed. Deviating from the max mana values above can have some adverse effects. For one, small decreases don’t really matter: mana resets to 0 once a spell is cast, so there’s no effect if you “overcap” on mana generation. Similarly, small increases can be huge nerfs in effectiveness: a change of 60 -> 61 mana is effectively the same change as 60 -> 70 mana, which is a 18% nerf to cast frequency in most cases.
The following graphs simulate mana generation and unit casts, assuming a starting mana of 0 and an attack speed of 0.75. Large caveat: these do not account for the cast times and mana lock, but those numbers are not easily available and I’m basically just hoping they’re negligible (even though they definitely aren’t). I'll link to my code at the end. It's not particularly user friendly, but it'll do. Hopefully.
This class of casters contains Teemo, Kog’Maw, Kindred, and Syndra. 30 is a convenient breakpoint because it makes equal use of Blue Buff, Shojin and Adaptive in making nice and tidy mana generations. The first auto attack generates half the mana, and the second generates the rest.
https://lh7-us.googleusercontent.com/docsz/AD_4nXdgPrgAmnhIIEh6BmG_tynoktkd4Vmgy9E7IZKMiYmR57wG--NU0_J4UcGOoBYYkzd0fWCQQzz4FjLlVgwdjIU6u9nsUmQKVJtPvoHoUNXghxKR2riRKfRkFhBcwNYgN2DB6Cf01Loy9TcqQxuPhH_FBXbS?key=1hCaKsJ66D5IwjiRkgkdsA
This class of casters contains Ahri, Soraka, Zoe, Morgana, and Alune. These champions are in a sweetspot where Adaptive, Blue Buff, and Shojin all work similarly: Shojin and Adaptive get more casts, whereas Blue Buff can increase damage. For any generic AP unit, this is a nice middle ground. I suspect Riot puts units who they want to be particularly flexible with mana items, since BIS matters the least for this class of caster.
https://lh7-us.googleusercontent.com/docsz/AD_4nXcs5iulWZEi9_-UuOsuFsKDgu4UfmNxdQ0S86u8ZJncJXK9DB2AFjSuk7Ll1TWLWb5m4dpfXFwBDTMd43CEDrX0V2iv8jBzwfFq6r9dkT4s3YmH6Gke846PRLnlJkdAva8M6OscTc5-JiJArlTEpY3daBWQ?key=1hCaKsJ66D5IwjiRkgkdsA