Sword & Banner gives every soldier in the shield wall a name

The hands do exactly what they did in the strategy games of twenty years ago. You drag a box, you bind a group to a number key, you give a move order. What comes out the other end is a shield wall bracing to absorb a charge, a cavalry wedge going around the side, or a longbow volley dropping onto a formation that has nowhere to be.

The unusual part is underneath the selection box. Every one of those troops is an individual with a name, an experience total and an upgrade path that starts at peasant and ends somewhere like knight. A unit in most real-time strategy games is a cost and a build time — you spend it, you replace it, and the replacement is identical. Here a spearman has been in four battles, carries the kills he made in them, and the reason you are hesitating over an ugly frontal attack is that the front rank is full of veterans you cannot re-buy.

Around the battles is the other half: a sprawling overworld of towns and villages with populations, industries and prices that move with supply and demand. You travel it, you raid settlements, you hunt bandits, you buy low and sell high, and you spend the proceeds recruiting fresh bodies and upgrading the ones who lived. The warband is the save file — it is the only thing that crosses from the map into the fight and back out again.

Battles themselves are generated rather than authored: procedural maps across biomes and seasons, from grassland to snowed-in mountain pass, with weather running while you fight. Rain darkens the ground, snow settles on it. The camera goes all the way in to watch a single sword swing and all the way out to command the whole field, across fifteen-plus unit types and three playable races with their own units and habits — human, goblin and undead.

How To Combine Video Game Genres is the essay this game lives or dies by. Its argument is that a mash-up does not work because two genres are in the same box — it works when each genre has a job to do and both point at the same focus. Sword & Banner is running three at once: a trading and travel layer, a grand-strategy campaign, and a real-time tactical battle. The thing that stops that being three games in a trench coat is the roster. The economy exists to buy soldiers, the battle exists to spend and promote them, and the named veteran is the object both halves are arguing over.

What Makes a Good Combat System? supplies the test for the fighting. Good combat is a stack of live choices: where to stand, who to hit first, when to commit, and what you have left to answer with. Formation combat is unusually pure on that last pair, because a cavalry charge is the most expensive commitment in the genre — it is fast, it is decisive, and it cannot be taken back once the horses are moving. The promise of “rank and flank” with high micro on individual units is a promise that both scales of decision stay live at the same time, which is the hard version of this design.

Procedural Generation Only Works When Something Checks It is the risk. A generator that can produce a bad map will produce one, eventually, on someone else. In a tactics game the terrain is not scenery — it is the ruleset for the fight. A generated field with no flank to take, a ridge that splits a shield wall into two useless halves, or a forest that makes cavalry pointless is not an unlucky map, it is a match where the interesting decisions were removed before you arrived. The rules that reject those layouts are the actual design work behind the feature list.

Sword & Banner is by Bitrot Games, self-published, listed as Strategy and single-player, with Steam achievements, cloud saves and Family Sharing. It is not out yet — the store lists Windows, and a window of Q1 2027.

What the store page is honest about is the lineage. It says Mount & Blade meets Total War with classic RTS controls, in pixel art, and every feature listed is recognisably from one of those three parents. That is a good sign more often than it is a bad one: a developer who can name the shelf their game sits on usually knows which part of it they are trying to improve.

If the appeal is a two-genre pitch where both halves have to be real rather than one being a menu bolted to the other, Monster Fantasy is the sibling here — Monster Hunter on one side, Animal Crossing on the other, and the same question of whether the quiet layer is a genuine game or just the place you go between fights.