I did a couple of in game tests to check the retraining dynamic.
Those units with 2 silver chevrons but at 10% strength, seem to have high kill counts. When retrained, it is as if they are mixed with non-experienced new troops. I had a levy hoplite 1 silver at 50% strength. I mixed it manually with a newly trained levy hoplite with zero chevrons. It would go from 1 silver to 2 bronze. I kept swapping it until I got back my 41 men 1 silver chevron levy hoplite back. Then I retrained it. I got a 2 bronze chevron levy hoplite on the next turn.
I haven't tested Marvin's proposal, and there is a chance I adulterated the test results by not reloading. All I do know is that there are times when you have a unit at 50% where retraining them will get you a full strength unit with the same experience level. Then there are other times that they decrease. I will have to try that situation out with the use of reload.
The closest interpretation I have for this is that each kill count by each soldier in the unit is counted and averaged. If you get a high enough number, by some kind of formula, you get a new experience level for the unit. If the 41 members of a unit have sufficiently high kill counts that adding in 40 more new members won't change the result of the equation, then that would explain the inconsistent effects we see with retraining of units.
In the end, I prefer to mix my units together, since it combines experience with experience, and is just a lot faster. I can also bring the experience level of a new unit up by giving it some members from a silver unit, without decreasing the silver unit's chevrons. In fact, a couple of times the silver unit's chevrons increased by taking soldiers from it to another unit. That, also, I can only attribute to some kind of underthe hood equation that takes in the average of the kills by the members of the unit, which now happens to contain a higher average than it did before when it had more soldiers in it.
Bookmarks