One of the most interesting games I have had the pleasure of playing is a small game called Atlas Reactor. It’s a 4v4 team turn-based tactics game where players play as various characters with different skills. However, what makes it interesting is the fact that all moves are chosen then executed simultaneously. As a result, the game is based on players having to predict what the other players are going to do. Let’s look at how the game is designed to give players interesting ways to “out-predict” each other and engage in mindgames and explore how it can be applied to create prediction battles in other games.

Sequencing Actions

Each turn of Atlas Reactor starts with all the players choosing their action for the turn and where to move (players can also choose to forgo their action to move farther). After everyone has chosen, game executes those moves. The execution of moves is divided into 4 phases: Prep, Dash, Blast, and Move. These phases serve to sequence moves and execute them in specific orders. First is the Prep phase, where buffs, shields, and traps are applied. Then comes the dash phase, where all movement actions are executed. After is the Blast phase, where all attack actions are applied. Finally, in the Move phase, players move to where they chose.

The order was chosen to allow players to gain advantages when they properly predict what the opponents are going to do. By having the Prep phase first, it allows players to shield players they think are going to take damage or place traps where you think opponents are going to move through, either with a normal move or a movement action. Then having the Dash phase before Blast phase allows the design of movement actions that can be used to dodge attacks. Having the Move phase last allows players to see where the enemies end up and start the turn knowing where the enemies will be for the Blast Phase, unless they use a movement action to dodge attacks. This is important, because having the Move phase happen before attacks would require the players to predict where the enemies might move to, which would be nigh impossible. By having the Move phase be last, it limits the necessary predictions players have to make to a more manageable level while still retaining depth and complexity. In addition, this 4 phase system creates opportunities for counterplay, punish, and limiting of options, which are important concepts we will cover later.

The important takeaways from this, is that for predictive gameplay to work, a game needs to limit what a player has to predict to avoid overwhelming them. There also needs to be reasons for choosing to do one action over the others, so a player can try to logic through what the opponent might think. For example, in fighting games such as Street Fighter, predicting the moves of your opponent is an important part of gameplay. Even though you can move in many directions in Street Fighter and other fighting games, players understand that the opponent has to move towards them to attack. This limits what players need to predict from predicting the opponent’s actual movement to simply how they are going to approach for the attack. At the same time, the player can reason what the opponent might do in certain situations. If the opponent is backed into a corner, the player might be wary of the opponent trying to jump over them and attack from behind. On the other hand, if the game has too little to predict or lacks reasons to choose one action over another, it would get stale and devolve into luck. Consider the classic game Rock, Paper, Scissors. The goal is to predict the opponent’s choice and choose the one that beats it. However, you only have one thing to try and predict, with only three options. On top of that, there isn’t a good reason to choose rock over paper, or paper over scissors. As a result, most games of RPS come down to luck and pure guessing.

Counterplay and Punish

Another important concept is counterplay. Almost every action in a predictive game should have counterplay. What this means is that there needs to be some other action a player can take to beat that action. In Atlas Reactor, every skill has counterplay. Most attack skills can be beaten by using a movement skill to dodge it or using a shield to block the damage. Players can focus on another enemy instead of a shielded one, use the unstoppable buff to stop knockback, or not move to avoid tripping a trap. All of these provide opportunities for players to outplay their opponent by predicting their moves. In a predictive game, you want to give the player many ways to gain an advantage through correct predictions.

Related to counterplay is the idea of punishes. In contrast to counterplay, where a correct prediction can nullify the opponent’s move, a punish disadvantages the enemy for being too predictable. One skill in Atlas Reactor that showcases this is a skill called “The Big One”. The skill drops a giant marker on the ground the turn a player uses it, but drops the next turn. The center of the blast deals a lot of damage compared to other skills. If a player correctly predicts that the enemy will move into a commonly used location and uses this skill on it, the skill punishes the opponent by dealing massive damage. The purpose of counterplay and punish is to reward players for correct predictions. Counterplay does this by negating the opponent’s move, which could have used a lot of resources, while punishes directly harm the opponent in some way. If the game does not reward correct predictions with advantages, players will be less inclined to try to predict.

What Are Your Options?

The final concept to cover is a player’s options. A player’s options are what they can feasibly do at some point in time. A player’s options change depending on their position, resources, and other variables. Above, we talked about limiting what a player has to predict. This means limiting their available options. In Atlas Reactor, you know that all players can only move during the Dash or Move phase. Movement actions activate during the Dash phase, which allows players to use them to dodge attacks, but they have a cooldown. This means that, when a player has recently used a movement action, they have less options as they cannot use that movement action. This demonstrates an important mechanic for predictive games, which is the limiting of options. The player needs to have ways to forcibly limit other player’s options. By limiting a player’s options, other players gain an advantage. The less options a player has, the more easily other players can predict their move. Therefore, a move that forcibly limits another players options is great for two things: punish, and helping a player make easier or even guaranteed predictions. In fact, a common strategy in Atlas Reactor is to try to limit the opponent’s options by burning their movement actions early. By doing so, opponents can only move during the Move phase, severely limiting their defensive options. This also means that the opponents can no longer dodge any attack, ensuring that the players’ attacks will hit. While this mechanic has the downside of having a negative feedback loop, it allows for skill expression and helps close games while keeping the potential for a comeback.

In conclusion, prediction-based games like Atlas Reactor rely on several core concepts and mechanics to make the game engaging and keep players interested in trying to “out-predict” other players. These include limiting necessary predictions within a range, having counterplay and punish, and having ways to limit other players’ options. They allow for skill expression by rewarding players for correctly predicting the opponent’s moves which is what makes these kind of games fun. Just remember to keep these principles in mind if you want to create a game or mechanic that relies on mindgames and prediction.