hit
counter

Balatro & Auto Chess

Video Games

A composite screenshot of two games. On the left, Teamfight Tactics with a grid of characters. On the right, Balatro with several playing cards and jokers.

Like many other game designers with a systems bent, I have spent copious enjoyable hours this year with the breakout indie hit Balatro. It’s usually described as a “poker roguelike”, which is a perfect logline for onboarding curious new players. Framed this way, it makes sense that it often invites comparison to Slay The Spire, another hugely popular deckbuilding roguelike.

It’s therefore ironic that the developer’s primary inspiration was not poker, but rather a Cantonese card game called Big Two. Furthermore, LocalThunk has stated that he drew initial inspiration from the score-chasing roulette game Luck Be A Landlord, but otherwise intentionally avoided playing other roguelikes during development. This often makes Balatro feel mechanically distinct from other roguelikes, since it does not always converge on the consensus “best practices” in the genre.

Seeing the game this way led me to an offbeat hypothesis. In returning to first principles on a deckbuilding roguelike, I believe Balatro’s design had a sort of convergent evolution towards a different game genre. A genre where players also seek synergies while drafting an evolving build, banking funds is rewarded with interest, risk mitigation is a fundamental skill, and the winner must survive multiple structured rounds with escalating stakes.

That genre is “auto chess”, which exploded in 2019 with the release of the popular mod Dota Auto Chess1. While there are many thriving games in the genre today, the one I’m most familiar with is Teamfight Tactics (TFT), which uses the game engine and characters from League of Legends. Riot’s president of esports claims it is “the number one strategy game in the world” in terms of popularity. For the sake of this comparison I will use it as representative of the wider auto chess genre.

Structurally, both Teamfight Tactics and Balatro always progress through an ordered structure of stages (antes) and rounds (blinds). The games are divided between the core phase (champions autobattling or playing poker hands) and the deckbuilding phase (drafting champions or buying jokers). The player’s performance in the core phase generates the currency for the deckbuilding phase. Both games also feature special rounds scheduled at the end of each stage (a carousel draft or a boss blind).

Note that, unlike Balatro, modern roguelikes do not usually follow this kind of regular sequence. Most have instead structurally converged on Slay The Spire’s randomly generated branching adventure map. Furthermore, it’s uncommon for deckbuilders to use currency for drafting; most favor the simpler “choose one from three options” mechanic.

More conspicuously, Balatro shares interest as a mechanic with auto chess: players receive additional currency at the end of each round for each $5 they’re sitting on, like interest in a bank account. This is an exceedingly rare mechanic in video games more broadly, even in the strategy genre. It suggests to me that the designers of both games had a similar need to incentivize conservative purchasing decisions. Otherwise, the player’s existential drive to win each round would discourage them from sitting on unproductive money. The interest mechanic rewards players who just barely win each round, leaving the remainder of their unspent resources to grow.

In terms of drafting, champions in auto chess and jokers in Balatro provide specific roles within a build. TFT gives buffs for stacking several champions of the same class and origin, and generally rewards balancing combat roles (tank, damage, healer). Balatro players can seek to maximize their score engine by assembling +Chips, +Mult, and ×Mult rewards across multiple jokers, ideally with significant overlap in their trigger conditions.

A tactical element of both games is the power curve of certain champions and jokers. To survive the early rounds, a player needs to draft for things that provide immediate value and tempo. Cheap $1 and $2 champions carry the early rounds in TFT. Similarly, jokers like Ice Cream and Popcorn provide immediate value, but decrease in potency after each round. As the game progresses to later stages, players need to pivot their build towards scaling power2. Managing this transition is a major skill element in both games.

Another shared strategic element is the ability of players to viably target “high” or “low” builds. For instance, an obvious strategy in Balatro is to build around high-value hands such as Flush and Straight. However, there are also competitive strategies for building around the lowest hand possible: High Card. Similarly, there are team compositions in TFT that seek to level up stacks of $2 champions, and others that count on turning out a $5 hypercarry.

Having a broad range of meta builds gives skillful players the opportunity to adjust their strategy on the fly based on the luck of the draft. Additionally, both games have mechanics that disincentivize relying on the same build every run regardless of circumstance. In TFT, all players are drafting from the same pool of champions. If another player is attempting the same team composition, those champions will get snatched up and become harder to find. Skillful TFT players can use this information to their advantage, pivoting their build towards undervalued champions.

In Balatro, the boss blinds create a similar dynamic. For instance, in the first week of the game’s release, the most popular strategy was to build a single-suit Flush deck. However, savvy players noticed that this choice was risky; there are four different boss blinds that each debuff a particular suit, which hard counters this build3. The philosophy that “a burnt hand is the best teacher” coaxes players to mitigate risk and integrate a secondary strategy into their drafting.

Screenshot of Balatro. The joker card Pareidolia is hovered, showing the text: All cards are considered face cards. The boss blind is The Mark: All face cards are drawn face down.

I have read and listened to several developer interviews about Balatro, and LocalThunk has been very open about his process and design values. I firmly believe that the similarities to auto chess that I’ve established here cannot be attributed to direct inspiration. Rather, I believe there are certain game design patterns that emerge organically when seeking to craft a particular kind of experience. Perhaps we’re all following these ley lines of game design, unwittingly tracing some underlying truth of mathematics and systems.

1. Another connection: Dota Auto Chess was also inspired by a Chinese board game: Mahjong.
2. Balatro even plays with this as a difficulty element: adding “Eternal” stickers to jokers so they cannot be sold is a significant challenge.
3. LocalThunk discusses this design choice in detail in an interview on the podcast “Eggplant: The Secret Lives of Games” around the 20 minute mark.

→ 2 CommentsTags: ·  ·  ·  ·  ·  ·  · 

2 Responses to “Balatro & Auto Chess”

  1. Random_Phobosis Says:
    July 8th, 2024 at 8:02 am

    I think it’s a cool observation!

    In the broader sense, I believe both autochess and build-focused roguelikes are subtypes of engine builder supergenre.
    I see defining quality of engine builder supergenre is to have, instead of a continuous action-feedback loop, two distinct phases: first the player builds the engine (which requires a lot of decisions), then the player observes the engine working (which requires little to no decisions). These phases then repeat indefinitely for the player to modify, fine-tune and upgrade their engine.

    Curiously, the genre of second phase can theoretically be swapped from card game to battler to puzzle to racing game without compromising the game structure, as long as the “engine running” phase respects the choices made in “engine building” phase. It’s the “engine building” part that seems to do the heavy lifting, structure-wise.

    Other examples of this supergenre games would be Diablo-likes, tower defense games, some RPGs and JRPGs, and also actual engine builder board games (deckbuilders, tableau builders and such).

  2. Cull Key Says:
    July 30th, 2024 at 11:49 am

    I love this post and the general idea that game design would converge on similar mechanics. One counterpoint however is that Localthunk has mentioned that he’s worked on an auto battler a few years ago(https://twitter.com/LocalThunk/status/1793770530035884461) which leads me to believe that he was aware of the mechanics in the genre.

Leave a Reply


Warning: Undefined variable $user_ID in /home/gangles/public_html/wp-content/themes/gangles/comments.php on line 129

© 2007-2024 Matthew Gallant. Powered by Wordpress. Privacy Policy.