I noticed that CPU-teams often have big variance in their performance over seasons. For example at fastticker:

Zwolle
This season probably champion.
Last season: Ends Eredivisie on position 14.

Helmond Sport
This season, fighting against relegation
Last season: Ends in second on Europa League.

Groningen
This season, bottom place
Last season, third place

There hardly seems be be any correlation between last seasons standing and this seasons (for the cpu teams). How comes?

I have noticed that too and I think that comes from the fact we don’t spend too much time improving CPU teams logic. Just to give a few examples:
-CPU teams in general don’t use very effective formations and they have no memory of which ones they’ve used and if it had a good result
-they “change their mind” quite often about which players to keep and sell
-the balance of good players vs promising players is not very good, too many reasonably good players are sold / let go

I usually ask myself what is the priority of enhancing CPU team behaviour vs improving something that matters to human managers and I end up working on something else, but we may be neglecting them too much. I was working on a change to make CPU teams use formations that were currently in use by human managers, but I put that on hold, afraid they would be much stronger. Anyway, what is your view on CPU behaviour improvements?