i don't think the coach ratings have a huge effect -- especially in the majors because development ends by late 20's. i could be wrong about it's effect on the win column, too. i've not done enough looking into it, so this is just a "best guess" based on a sample that doesn't give a high % of confidence.
the following applies more to the majors, than the minors:
when running the same season over 2 times i made the same coaches all 200-rated and all 0-rated coaches each run, and i got fairly similar results (* no other factors changed). so, nothing too drastic is happening since range of results overlap to some extent. Also, the difference between 2 potential coaches for any particular rating is nowhere near a 200 spread. pragmatically it's likely a very negligible difference relative to the ratings of available choices.
maybe well-rated coaches will be a big deal if large portions of your team are poor at handling critics/success/failure blah blah blah... this was a very good team all-around in the test -- that's what i want to maximize though... playoff teams. just like my team is constructed mostly for the playoffs and not the regular season.
_______________________
minors:
i want coaches that enhance development -- all other factors are tie-breakers. so, controlling managers that can handle young'ens, and pc/hc that are good at teaching skills, etc..
results don't affect development - law of independent results. affirming the consequent is a logical fallcy - pretty confident they adhere to this in the code. in fact, elevated frequency of anything to an extreme degree may actually help develop the related skill even if it hurts the team's performance. that is, if game repitition is coded as a factor. not a problem in the minors... development > wins
However, i think if they are on the active roster and getting in games fairly regularly, that's all you need to maximize development as far as gametime is concerned. i don't even think you have to play more than 50/50, but i always have minor leagues starting potential over ability in the settings. so the top prospects in the league are gauranteed to get plenty of playing time. *a few slip through due to scouting innaccuracis, but that's the exception definitely not the norm.
______________________________end of minors section
bit off-topic, but majors vs minors should be a completely different thought process, so i'll throw it out there.
for ML managers, i focus on game strategies and coach style (conventional et. al.). after that, apply other stuff that you value to decide between coaches with similar strategies.
hitting and pitching coaches are a bit tougher to evaluate given the information. i'm less sure as to the best strategy for these guys. i just pick the focus that matches how i construct a team: power pitcher focus and neutral or power hitting focuses.
post note:
we can see "teach baserunning" in the coach editor now. i'm like the other comment above, i just cheat with coaches... i very rarely convert former players. i just shortlist the ones with strategies i like, then compare the numbers in the editor after that as a tie-breaker. this makes picking a hc/pc very easy
lately i've been just making a manager named "Coach Surname," then II, III, IV etc... i pop them into my minors ~5years before i retire the preceding iteration - i make sure their rep grows by managing decent minor league teams - the players are league level or borderline next tier quality. so nothing too crazy or time-consuming. i like seeing how their careers compare and slightly adjust strategy to see if i can figure out something better in a systematic way without putting extra effort into it. same with owners -- "Owner Surname," II, III, IV etc.