This is a follow up on a previous thread identifying performance disparities with Chaos Knights after the model update.
TLDR: Chaos Knights still suffer from contact issues on the charge which hurts the lance variant particularly hard.Shrinking model count to 48 while increasing mass (+200-300) and HP (+27-35 HP per entity) improves both units performance, but especially the lance performance while not upsetting the units current balance.
Additionally the Swords of Chaos are still extremely weak for the price having missed the WS buff. They need a significant WS increase to be worthwhile at their current price. Something like +45 WS would be my starting point, as this would mean per attack they'd be about even with the Knights of the Brazen Throne who come with 8 more entities and a much better self-buff.
Below is the link to the previous thread as well as a more in-depth discussion and documentation of the proposed changes. All tests are 5 seconds after combat as this is where the performance issue is most readily apparent.
Previous thread: https://forums.totalwar.com/discussion/327643/chaos-knight-performance-issues
Key findings were the new models were interfering with Chaos Knights performance on the charge and causing them to deal less damage than their stats suggested as entities were getting stuck on each other and failing to engage in combat.
Proposed solutions were to decrease entity counter, and raise WS, mass, and health per entity to compensate for the reduction, with a note that simply raising WS could be a quick and dirty solution allowing the models engaging . By reducing the number of entities and increasing mass slightly Chaos Knights were able to charge without interfering with themselves increasing their damage on the charge. The other option, simply increasing their WS also results in increased charge damage but would still leave the Knights underperforming their stat card, but not their price point.
CA went with the latter option and the results are largely positive with chaos knights now performing reasonably for their price, but still below the what their stat cards suggest. To save on screenshots, previously chaos knights were struggling to consistently hit above 1000 damage on the charge, after the buff they're now hitting for 1300 and 1800 which is more in line with units like Reiksguard and Gryphon Legion getting around 1500 on average.
This is a positive outcome and if this were the only change noticed I'd consider this a successful fix. However one area of unusual performance I've noted with when comparing Lance Knights and Sword Knights into various targets.
So generally, you see infantry performance is about the same for both of the units on the charge which is not what we'd expect given the targets health pool, nor the relative damage stats of the units (around 1500 damage on average). On the other hand when looking at large we generally see a significant performance increase for the lance knights as one would expect from their roughly 40% greater damage per attack on the charge. In some cases this could produce damage differences around 900 but on average we see around a 400-500 damage gain.
I suspect the issue is again consistency of contact, with the large models of the chaos knights getting caught on each other, something that happens more commonly against infantry but occasionally against large targets (see outliers). Functionally only the first line is consistently making attacks up to 5 seconds, with attacks made in the window between 1 and 6 seconds being very small, sometimes seemingly 0 being made. This likely impacts both variants equally but the lances lose more damage from the bug given they gain more from their CB which is what's being disrupted.
I would suggest that adjusting entity count is a viable solution to this inconsistency by reducing the inconsistency caused by the knight models colliding with each other. Going from 60 to 48 models changes the Chaos Knight formation to allow the back lines to more consistently slide between the front line and engage in combat as seen here.
Potentially a similar solution could be achieved by adjusting the formation or the models but I'm aiming for changing as little as possible while addressing the issue.
When lowering entity count, adding 200 mass, and increasing hp per entity by 20-25% to compensate for entity reduction no consistent damage loss was observed into large targets and damage consistency into infantry was improved. I think this is just because the back lines no longer run into each other when trying to path forward as often.
This is pretty small potatoes as a multiplayer issue where informed players can make decisions based on their needs and use the knight variants accordingly, and is more of a campaign issue where upgrading to lance knights can actually result in performance decreases due to inconsistent performance. If the trade of is supposed to be damage for sustainability the units are still functionally bugged.
For the Swords of Chaos, they need a significant WS buff to compete with other monstrous cavalry units, which is what they functionally are with 24 entities. They were not buffed in terms of WSin 3.0 which has made them look even worse by comparison. I'd recommend 110 WS (+45 WS) as a minimum value for decent performance given the units stats and entity count. Below is a comparison of the damage of the Swords of Chaos with these buffs to the Knights of the Brazen Throne.
As you can see the Knights of the Brazen Throne are massively ahead into infantry and large targets. I'd argue this performance is about what we'd expect from an RoR priced at ~1600 for the Swords of Chaos which is roughly in line with game 3's cavalry RoR balance, that is 200 more gold than the base unit. This is a conservative suggestion however, as the Swords of Chaos are quite bad.