Not fully thought through, but it's so simple that maybe it doesn't need to be thought through too much.
Get rig of individual coaches altogether. Replace it with a coaching budget that can range from $0 to $20m. Actually, it would be replaced by four separate coaching budgets, each of which covers your players at all levels (ML through RL):
Hitting (ranges from $0 to $5m)
Pitching (ranges from $0 to $5m)
Fielding (ranges from $0 to $5m)
Base Running (ranges from $0 to $5m)
For each category, a $2m budget gives you average development (of younger players) or average regression (of older players). Go below $2m, and you risk slight regression of ratings in that category (or accelerated regression for older players). $0 gives you a lot of regression. Go above $2m, and you get above average development for younger players (slower regression for older players), with $5m being maximum improvement / minimal regression.
Why do this?
1) Nobody likes coach hiring. Many people hate it.
2) The current system is built on smoke and mirrors: there is no objective way to quantify how good "good" coaches are, or how bad "bad" coaches are, other than anecdotal observations. This gives you more of a "hands on" approach to knowing what your budget is buying you.
3) You can tailor your budget to your specific needs. If you have a fully developed veteran pitching staff and weak pitching prospects in your system, and a solid crop of young hitting prospects, you can go $3m on pitching and $4m or $5m on hitting.
Looking at my three worlds, a typical coaching budget is $10 or $11m. Going "average" in the four coaching buckets would cost $12m, so that's low impact (on average) for most people. But seeing that you have more control over the impact of coaching, owners now might choose to invest more in coaches than they did previously, which takes away budget from some other category. More decisions means a little more strategy.
Another possibility: similar to the +/- $4m maximum change in other budgets per season (the 4 scouting budgets, medical and training), you can limit the amount of change in the coaching budgets from season to season to $1m per season, thus preventing huge swings from season to season. This conforms to the "planning for future seasons" aspect of the game (actually, it introduces it to the coaching budget).
Is this realistic? Not at all. It's a deviation from "real life". But it might be an improvement to the game.
Thoughts and feedback welcome.