Going Rogue: Monster Train

As Nightwell Games gears up to announce our upcoming roguelike title, we’ve been playing every turn-based tactical roguelike we can get our hands on, trying to break down what works, what doesn’t work, and what underlying principles roguelike designers can take away.

Step one is admitting you have a problem

In this ongoing column (monthly, or however often I manage to successfully wrangle the Nightwell Beasts), I’ll be profiling various tactical roguelikes — popular and obscure, successful and unsuccessful — and evaluating how they measure up. For our first column, to give you a sense of what the pillars mean and how they work, I’ll be profiling a familiar title: Monster Train (2020).

The Pillars

We’ve identified four core pillars that the best roguelikes typically share: Simple math, deterministic outcomes, tradeoffs, and intermediate geometry.

• Simple math: Roguelikes work best when health and damage values are as small as possible, and when the number of modifiers and other special effects is kept to a minimum. The player should always be able to quickly mentally calculate the outcome of a given move. This allows the player to be certain that they are making the optimal choice for the situation.
• Deterministic outcomes: Roguelikes should generally not have an RNG, or if they do, it should precede player choice. In math terms, a game may have input randomness, but should not have output randomness. In a deckbuilder, for example, the order the cards come up can be subject to input randomness, but the outcome when you play a card should always be the same. This creates puzzle-like combat scenarios where the player can precisely engineer a winning strategy.
• Tradeoffs: Every move should have a downside, whether it’s time, mana, deck space, or opportunity cost. The best move should always be situational. There should never be a move that you would only not use because you don’t have access to it. Tradeoffs create a variety of ever-shifting strategies as the player balances risk and reward.
• Intermediate geometry: The best roguelikes feature an element of positioning, but not too much positioning. The large open arenas used in other TTRPGs are excessive — there’s a large amount of unused space, players waste turns moving across the arena, and there are often many spaces that are functionally equivalent in terms of game outcomes. On the other hand, games with no geometry, where it’s simply the player idling on one side of the screen and the enemy idling on the other side, are leaving out an element that can add a fascinating layer of strategy.

The Game

In Monster Train, you join the war on Hell on the side of Hell. You must protect the train carrying the last remnant of hellfire by deploying a variety of (mostly cute) demon characters and spells to fight off waves of angels attempting to storm the train. Runs clock in at a svelte half hour — a good lunch-break length.

While choice to stack the levels vertically instead of, as one might expect for a train, horizontally is a bit odd, the theming gives the game a distinctive look-and-feel that sets it apart within the crowded field of turn-based fantasy deckbuilders. You choose from a variety of factions with distinct powers, such as the Umbra, which devour smaller units to grow stronger, and the Melted Remnant, whose units despawn after a few turns. There are just enough hints of lore to make the world feel robust. Think Snowpiercer if it were populated entirely by Dark-type Pokemon. The soundtrack is also a total banger.

Also, the final boss is named Seraph and if you defeat him you get the Sans Seraph achievement, so there’s that.

But how about the design?

Simple Math: C

This first pillar is actually where Monster Train measures up the worst. While there are weak enemies with one health and one damage, bosses quickly rocket to 1000 health and above. Units often end up dealing oddball amounts of damage, like 37x3. A winning strategy requires complex stacks of multipliers, cumulative damage, buffs, and debuffs that’s functionally impossible to calculate mentally, especially in the final wave, when the two sides trade attacks until one side is wiped out.

Happily, a user-friendly interface makes up for a lot of the problem. Damage adjustments are automatically calculated, and you can mouse over to see a preview of the results of a move. You can’t, however, see a breakdown of the contributing factors, so it’s easy accidentally make suboptimal moves and have to puzzle out what just happened(“Oh right, my champion empowers my units with Rage and the enemies have Thorns, and that’s why the morsel in the back died instead of being eaten even though it didn’t get hit”). And since you can only see the outcome of the current move, it’s especially difficult to strategize what will happen to a unit as it traverses the entire train.

But completely avoiding this kind of unwieldy complexity is not necessarily possible or even desirable, because it goes hand in hand with one of the most fun strategies in roguelikes: The runaway victory. In these games, simply collecting the strongest units isn’t mathematically enough to win — you will always fall short. You must figure out some kind of exploit that multiplies damage in some exponential way. These victories are difficult but massively satisfying when you can pull them off, and in particular, they’re great because you can’t just fumble through them — they force you to thoroughly understand the game’s systems and make a plan.

Deterministic Outcomes: A

Virtually all deckbuilders have deterministic damage, so this one is a gimme. Units deal static amounts of damage, typically to a deterministically-chosen enemy. The only exception is some spells which target a random enemy and a few which spawn a random unit, but that’s rarely a make-or-break part of the strategy.

As with other deckbuilders, input randomization comes in the form of draw order, and this can and does swing the battle in huge ways. This leads into the question of tradeoffs.

Tradeoffs: B

A common issue with deckbuilders is that it’s very easy to use deck randomization to smooth over uneven tradeoffs. As is typical, in Monster Train you start with a weak set of cards and gradually gain stronger ones while upgrading the ones you have, so it’s common to have a mix of good and bad cards, like a 2-damage spell and a 12-damage spell, at the same time; the only reason you’d ever play the weaker one is because you didn’t draw the stronger one.

The most interesting tradeoffs happen during the deckbuilding itself as you choose whether to focus on spells or units, whether to concentrate your best upgrades on a few ultrapowerful units or spread them out, and whether to buff your weak cards or purge them entirely.

And with a well-optimized deck, interesting tradeoffs do arise. Pick off the glass cannons or focus fire on the tanks? Rush the enemies at the bottom of the train or hold the line at the top? And, most crucially, in the hugely challenging boss battles: Deal with the mooks (and risk being unable to take the full-health boss) or wear down the boss (and risk being overrun by mooks in the meantime)?

These tradeoffs are mostly created by the pillar where Monster Train is the strongest of all: Geometry.

Intermediate Geometry: A+

Monster Train is strong in most areas of roguelike design, but geometry it blows out of the water. The train has three levels, not counting the pyre room; within a level, units on both sides are organized in a row, which has limited space. Enemy units enter on the bottom floor, exchange attacks with the allied units targeting whichever unit is currently in front, and then move up to the next level.

The geometry, therefore, functionally has three dimensions: Horizontal, vertical, and time. To hit an enemy, an allied unit usually needs to intersect it in all three dimensions. A lot of the strategy revolves around placing units so that they intercept enemies enough times to take them out before they reach the pyre room; a common failure state is having an army that deals enough damage mathematically but not in the right place and time.

The key to the brilliance here is that the three axes all function differently. In a game with a grid map, “one square up” and “one square to the left” are functionally the same, but here, “place the unit one level up” versus “place the unit adjacent” can have wildly varying outcomes. A choice like “move all the units from level 3 to level 2” can be a game-winning move. This kind of strategic thinking is why I lean so hard on geometry as a design pillar.

Overall grade: B+

And that’s, to be clear, a real solid score that other roguelikes could aspire to. There’s no grade inflation at the Nightwell Forge.

Metaprogression: Mostly Game

The question I’m most interested in with regards to metaprogression is whether it’s structured more like a game (objectives set by the game itself and organized in a linear progression, with each challenge unlocking the next one) or more like a toy (assorted challenges and bonuses that the player can combine in any arrangement to create their own objectives).

This choice is mostly personal preference, with pros and cons to both. Game-like metaprogression usually keeps players hooked for longer, but can become grindy and repetitive, which risks losing player goodwill. Toy-like metaprogression has a lighter hand and avoids the grind, but players may lose interest sooner and a certain number will never tinker around with it at all. Game-like metaprogression appeals to externally-motivated players. Toy-like metaprogression appeals to internally-motivated players.

For the most part, Monster Train sticks closely to standard game-like metaprogression. There are a whopping 25 Covenant ranks (difficulty levels), which must be beaten consecutively. As you play each clan, they upgrade, unlocking new cards and abilities. However, there’s also a robust list of Mutators that allow you to customize a run in entertaining, and often wildly imbalanced, ways, such as “get a random Champion from a different clan” or “give all your units an extra upgrade slot.” These runs don’t count towards leaderboards or unlocks, but give players a fun option for unstructured play.

And that’s Monster Train! Overall, this is a solid, well-thought-out game with tight mechanics. While the gameplay can get somewhat repetitive after a lot of replays, especially the seemingly-endless Covenant ranks, the fact that it’s gotten to “a lot of replays” status in my library is in itself a ringing endorsement.

I like to end by giving each game an honorary award highlighting something it did particularly well. Given that I still have that music stuck in my head, this one’s easy.

Honorary Award: Best Music

Join us next month for another analysis — or, if you’re a roguelike dev, send your game to ravenmaster@nightwellgames.com for a chance to be featured!

More Going Rogue

Tetra Tactics
Mortal Glory
Iris and the Giant
Fhtagn Simulator
Meteorfall: Krumit’s Quest