If you want to keep lancer template and cataphract mixin, you can mix and match any way you want. A cataphract is heavily armored by definition, that is one difference. What you should not neglect in my opinion is: even though both have strong charging capability i.e. lancers, kontos is a longer weapon with more reach than xyston, so you might want to reflect that in increased range.

I started mine mainly with some artwork. I plan to make it unbalanced by design for single player. I guess it is a relaxing hobby of sorts. But i will surely share any idea from my mod which may contribute .    I am a sw project manager, i like to simplify and keep design modular where possible, hence the suggestion. What i like about mixins is: you can keep common/major unit types, so unit types appearing for most civs and are significantly different in templates, and add small differences in mixins. But this is just my preference.   I hear you. The game should aim to incentivize historically accurate armies, and small changes in top to reward insight. Ideas i have but havent put to test to see the effects: 1) keep citizen soldiers we want to incentivize as they are, and make the citizen soldiers we want to downprioritize another category like allies or dependents, which keep their stats and cost but lose resource gathering. In a sense, they are unprofessional soldiers we recruit only if we need them tactically. This may be a drastic change but i wanted to consider and try. 2) or introduce allies/dependents, keep their resource gathering, but put some kind of limit to their number based on a set ratio between number of allies/dependents allowed per citizen soldier. In fact, such a ratio-based limit can be introduced for mercenaries and champions as well, with some civs having a civ bonuses or techs relaxing the given ratios. For example Carthaginians being allowed more mercenaries than other civs. I know limits are not popular in the game but a Spartan army with no hoplites is a travesty... 3) the mildest approach would be to introduce unit-specific generic upgrades, and allow each civ to have only the upgrades of units we want to incentivize. So in your example, a Spartan player will have good hoplites but poor javelineers, so will build javelineers only when they are crucial tactically.

It's just chat gpt summary so not sure how accurate it is: Companion cav: armed with the xyston (long thrusting spear). Sele cataphract: Wielded a kontos, a two‑handed long lance. So i guess in someways we are in accurate calling companion cavalry lancers... although they are still considered "Elite heavy cavalry"  And here its even considered a 1 handed weapon, even though they still didnt use a shield... this is where it gets so "confusing" and tricky. How to differentiate each! Feature Companion Cavalry Seleucid Cataphracts Armor Medium-heavy Very heavy (rider + horse) Weapon Xyston (1‑handed) Kontos (2‑handed) Mobility High Low–medium Shock power Strong Extremely strong Tactical role Breakthrough + maneuver Frontal hammer Best terrain Open but flexible Wide, flat plains

Chandragupta II

I find it funny that you propose to use a Fibonacci sequence, but then you apply it to so few points that you might have as well used a linear one The problem with Elo decay is that it is quite arbitrary, because it’s not part of the fundamental model, and it breaks the zero-sum game representation (although I guess most here won’t care about the meaning of the math, but deep down it's all about probabilities). You could also consider instead the Glicko rating system, which is seen as an improved Elo rating system, and what incorporates is not decay, but makes one’s Glicko rating change more drastically the more inactive the player is. It has been implemented on many games online. If what you want is for completely inactive players to just lose their ranking, then something that I think could be considered is a mix of the new and old FIFA/Coca-Cola World Ranking methods. The new method is Elo-based, and uses an importance coefficient depending on the context of the match (and it’s zero-sum game if asymmetrical points cases are removed, like in penalty shootouts), while in the old method old games were weighted less, but not in a subtle way and rankings would unexpectedly jump. I think that by using the new method and modifying the importance coefficient such that it slowly decreases with time (older games becoming gradually less important) what you want could be achieved way more elegantly and mathematically sound.