Evaluating Unit Cost-Effectiveness in Kings of War: A Comprehensive Analysis Using Elo Ratings for the CoK 2025 Updates

Introduction

This post updates my unit cost-effectiveness analysis for the Clash of Kings 2025 rules, using Elo ratings from 1.25 million combat simulations. Background on the method is here. CoK 2025 is new, and Elliot is still fixing errors in the Mantic Companion. Those changes may affect the results, so please let me know if you spot an error.

Methodology

Data Collection and Preparation

I compiled unit data from CoK 2025, including faction, unit name, type, size, and combat statistics. I excluded units marked ‘I’ (Individual) because their special rules and buffs make them difficult to compare through direct combat simulations. For missing critical values, I either used defaults or omitted incomplete entries.

Elo Rating System Implementation

Each unit was assigned an initial Elo rating of 1500. The Elo rating system, traditionally used in competitive games, was adapted to evaluate unit performance based on simulated combat outcomes. The rating updates were governed by a dynamic K-factor, which decreased linearly from 100 to 1 over the course of the simulations, allowing the system to stabilize as more data was collected.

Combat Simulations

A custom-built combat simulator executed 1.25 million battles between randomly paired units. The selection process balanced between random pairings and matches between units with similar Elo ratings, promoting a comprehensive evaluation across the performance spectrum. Each simulation consisted of multiple rounds, with outcomes recorded to update the respective units’ Elo ratings.

Cost-Effectiveness Analysis

Post-simulation, the cost-effectiveness of each unit was assessed using two primary metrics:

1. Elo per Point: This metric was calculated by dividing a unit’s final Elo rating by its point cost, providing a direct measure of performance relative to cost.
2. Elo Difference: A linear regression model predicted expected Elo ratings based on unit cost and size. In other words, different base sizes for units have different performance expectations, and this metric attempts to correct for that by comparing units versus their footprint size’s expected performance. The Elo Difference was determined by subtracting the predicted Elo from the actual Elo, highlighting units that outperformed or underperformed expectations.

These metrics were standardized and combined, with a weighting of 60% for Elo per Point and 40% for Elo Difference, to formulate a Composite Cost-Effectiveness score for each unit.

Results

Top Cost-Effective Units

The analysis identified the top cost-effective units across various types and sizes. I plan to get into more detail in future posts, but here’s a quick preview. Below is the updated summary table of the top cost-effective units by Type and Size:

Cavalry

Chariot

Heavy Infantry

Infantry

Large Cavalry

Large Infantry

Monster

Monstrous Infantry

Swarm

Titan

War Engine

Visualization

The violin plot below visualizes the distribution of cost-effectiveness scores across different unit types. It provides insights into the variability and central tendencies within each category, aiding in comparative analysis:

The chart shows which unit types provide higher average value for their points and which have more variable results. War engines, chariots, and monstrous infantry span a wide range; cavalry, monsters, and titans have tighter results.

Conclusion

Combining Elo ratings and points cost gives a framework for comparing unit efficiency. Players can use the results as one input for list building, and I can update the analysis when rules change.

I plan to look at specific armies next. Let me know which unit types or factions you would like to see in a future breakdown.