Home Videos Games Podcastle


Teamfight Tactics, Auto Chess Mobile and Dota Underlords are in a state of initial public release.

Autobattlers are essentially a randomized set collection games where each round you pick a hero with tags from a random hero pool, you do set collection on the tags for various combat bonuses and then each rounds you either fight creeps or other players via autobattle. Each time you lose to a player you lose points, if you lose enough points you are eliminated from the game.

I’ve played a few games against against normal bots in Dota Underlords, I’ve done fairly well (I’ve consistently either gotten 1st or 2nd against bots) but I still feel the game is too RNG heavy for my tastes. Even if I am bad at DOTA, with certain heroes even I can win against normal AI bots 90+ percent of the time.

I like the pacing of the Autobattlers more than Rougelike Deckbuilders (Slay the Spire, Monster Slayers, and Dream Quest et al.) but overall the pacing still feels slow to me.

I am super interested this forums thoughts on Autobattlers! There’s like multiple conflicting tier lists on the internet for Dota Underlords and I don’t know which one is best yet.

1 Like

It’s a really interesting concept. There have been similar things going as far back to the original PlayStation and N64 era with Robo Pit, Carnage Heart, even Ogre Battle (“Liberation!”) and not to mention the (very silly) recent-ish indie Ultimate Epic Battle Simulator.

I like how these emphasize strategy over tactics, since the actual play is taken out of your hands once the game begins. You set something up, and then you get to watch the show (eat popcorn, take on some side-bets about who’s going to buy the next soda, etc.)

1 Like

Thanks @MinuteWalt! Looking at the larger context of Autobattlers really helps me clarify my thoughts on them more. I also learned from that these Autobattlers owe a lot of mechanical inspiration from the Pokemon Defense Warcraft 3 mod.

I like a good strategy game where you don’t necessarily control the units during battle. Conquest of Elysium 4 is my current favorite one. I might upload a video of me playing COE 4 soonish.

But these newer Auttobattlers are distinct from those games via a specific mechanical soup of "Small Randomized Buy" , “Limited Space”, “Interest”, “Set Collection: Evolution”, and "Set Collection: Bonuses"

The goal of these newer Autobattlers is to build a team powerful heroes over a series of rounds. Win enough of your battles and you win the game. I’m going to analyze each of the mechanics isolation and also analyze the specific composition of these mechanics to explain my dissatisfaction with the newer Autobattler format.

"Small Randomized Buy" . Every round you buy heroes from a small randomized pool of heroes. You can spend resources to re-roll the pool.

The best example of other games using this mechanics are Ascension deck builder style buys and tight drafting games like 7 Wonders. This mechanic at least two main advantages when used in physical board games. 1. Reduces the setup time of the game compared to separate piles of cards (Dominion style buys). I really value the “Pacing” property in board games. Any mechanic that speeds up play or set-up I feel is super valuable in board games. Autobattlers are digital games so they do not benefit from the property as much as their physical counter-parts. 2. Allows the designer to have potentially a large number of potential options for the player, while not overloading them with information at any one time. This property increases the accessibility of the game while also keep the potential complexity of the game high. Autobattlers definitely benefit from this where they are perceived as being much more accessible than the games they are hosted in (DOTA and LOL).

"Limited Space" - In Autobattlers you are constrained by not only the cost to buy the heroes but also both your army and reserve have space for a limited amount of heroes.

The most similar feel of this mechanic to me is in tight drafting games like “7 Wonders”. The mechanic both forces players to commit to a strategy (can’t buy ALL the things) and manages complexity. If a game does not have good use of Limited Space then the game can feel bloated and a the cards unimpactful (Terraforming Mars). Both of these properties are used to good effect in isolation in Autobattlers.

"Interest" A significant portion of your resources come from the interest mechanic wheres players earn interest on unspent resources.

This mechanic is also used in various Tower Defence games such as Element TD. This mechanic shifts the question of “Can I win the current battle?” to “Can I win the current battle with minimal resources?”. It also allow for high level strategy where players can trade power in the early game for power in the late game. Again, in isolation the interest mechanic works for Autobattlers wheres its more interesting to try fights between hero teams with the proper amount of strength versus maximum strength.

"Set Collection: Evolution" - Your heroes in auto chess can evolve into more powerful versions if you collect multiple copies of them.

This might be similar to the “free card” mechanic in 7 Wonders. This mechanic rewards players for committing to a specific hero. Evolved heroes are significantly more powerful than their unevolved counterparts. Level 3 units in particular can be battle defining. Again, in isolation this a a fine mechanic for Autobattlers to have.

"Set Collection: Bonus" - Your hero team gains bonuses for collecting specific types of heroes.

This mechanic rewards players for committing to a specific types of heroes. These bonuses are large and again potentially battle defining. Again, in isolation this a a fine mechanic for Autobattlers to have.

I dislike the composition of all of these mechanics with the specific constants used by the Autobattlers. Basically, “Set Collection: Bonus”, “Set Bonus: Evolution” and “Limited Space” all work in tandem to make players desire very specific heroes, but "Small Randomized Buy" severely limits whats the players can purchase. The cost to buy heroes that don’t immediately synergize with your setup is expensive in both space and resources due to “Limited Space” and “Interest” mechanics. This makes your strategy feel very brittle and inflexible and at the mercy of RNG. Also the “Interest” mechanic feels unsatisfying in the late game because the early cleverness will not always translate to late game rewards as you want very specific heroes and even with rerolls you feel at the mercy of the "Small Randomized Buy".

1 Like

Yes! They are different from those early games.

The way that they play makes them a unique genre in the field. Since you brought it up, I’ve been wanting to dive into these more to find out what makes them tick, to examine why they are appealing. It’s been a long time since I played those older games (design a robot and its AI, or design a team, and just let the scene play out), but I feel the autobattler is much more structured, has a tighter rule-set, and you can possibly figure out the outcome as soon as the match starts. It’s the choice and placement of “resources” that determine the final outcome, while the ruleset just does the logic to complete the game.

The outcome is determined at the beginning unless there is RNG to shake things up during a match (Auto Chess really illustrated this, both where it failed and where it succeeded. The designer deserves props for making this out of DOTA 2). There’s so much more going on than a typical deck-builder, too.

@Quoc, you are a game designer/developer, and this is a new sub-genre that I’m just getting into now (specifically because you brought it here, I probably wouldn’t have noticed it for months, otherwise). Your insights have been very illuminating, I hope we can help you out as well.

(Game design is an art form and a philosophy, and we all have found a place here where we can talk about it openly, and I’m happy you found us geeks at SU&SD.)

You are the de-facto expert in this new genre here on the forums, @Quoc, which three games would you suggest for someone dipping their toes into the autobattler? (Keep it to three. We are all amazing, true, but still mostly have no money).

1 Like

I’ve played Dota Underlords first as it lets you play against bots. I’ll probably try Teamfight Tactics in a few days.

I’ve made a video of a Underlords playthrough https://www.youtube.com/watch?v=O7xSByVrglM

1 Like

CoE4 is good. It works, I think, in part because battle-level simulations are always made too complex for the accompanying AI to handle. By shifting the focus elsewhere, the AI opponent is able to compete*. In other words, unlike Total War games for example, you can’t pull off “amazing against the odds” victories in every. single. battle, which is great.

I haven’t played any of the other games you’ve mentioned.

  • (well, kind of. You still have to give it massive production buffs.)

@Benkyo I also appreciate the janky randomness of the enemy stacks in CoE4 . Its super fun to try and figure out puzzle of “Can my stack of random soldiers fight TWO MOOSE or ONE CRAB”?

I’d also would really like to appreciate Dominions but I don’t want to script 100 mages individually.


Yeah, Dominions is a micromanagement mess.

1 Like