all personal opinions, except # of draft rounds should be relative to duplicate layers of MiL system (# of R leagues, for example, not # of teams) and the rules restricting who is eligible at each level. provid emore than enough is better than providing just enough.
try to implement things one at a time when you deviate from default. the more related things are the better off you are changing 1 at a time.
aging/development is accurate at default... group-think in forums says differently. i trust their research over some 'feelings' any day of the week and twice on sunday. if you want them in the mlb at a younger age, you increase development etc.. fit settings to wants.
scouting -- info is power, more info means better decisions.. .applies to frequency and accuracy settings.
whether you want ratings relative to mlb or position is a bit of opinion. what's not opinion is that making things relative clouds the picture and the ratigns mean somethign slightly differetn every time you look at them -- objectively more difficult for you to learn / predict etc based on that info. (small differences, but still differences)
there are benefits too making ratings relative... learning curve probably isn't as steep for mil promotions, for example, but still not as precise nor as good as you can get with a good memory and experience while using more consistent/precise info.
RP look alot better with relative to position... this is the one thing i do use, now... only because they changed how it worked in last couple years, lol. begrudgingly, i might add! wihtout it, there's a very few well-rated rp and the rest are 30-50 or lower out of 80. not much inbetween.
when it comes to alot of these settings, thnk about how it affects teh AI and the game world more than how it affects you. you will adapt much more easily than the ai.
increasing inaccuracy, for example, will spread success out more. it's making "luck" a larger portion of the equation for success. this is somethign that affects you the same as the AI. (it probably hurts the ai more than you when you lower accuracy, but maybe you want to see talent later in the draft... i find "low" works well.. very low was just too much randomness in the draft for me to have fun.
TCR - 100% opinion, but it's going to have correlated outcomes, nonetheless.
Traditional / sabermetric -- will affect coaching sliders and lineup choices. with lineup choices, i think the biggest differnce is who bats 2nd... although other differences do exist.
finance tip that may work well -- resulting contracts are all about supply and demand... not just of players but of cash supply too. a salary cap, even if slightly above what any team can afford, may help rein in exorbitant salaries. i think the AI more tightly controls % of Hard Cap per player than if none exists.
e.g. i hav ebeen using ~250M, which only 4 teams can reach in my league, and rarely come near. ~30M is about is high as a yearly salary goes. in past i had very similar finance settings and 35-40 was a typical high-water mark. the only difference being a hard cap, now ... and it doesn't affect team's max spending at all. Anyway, this may be a very easy way to control a poorly setup financial enviroment... otherwise, stick to default settings if oyu don't like economics
*Always use a max cash on hand cap.. defualt 10M works fine or extensions will get out of hand expensive.
bottom line:
don't let what a feeling about a setting trump what actually happens in game. another reason to try to change 1 or few things at a time...more clearly see what results from that change.
if you get frustrated by it, try to arbitrarily pick out a set and stick iwth it... after a while look back on it and plan out 1-2 changes for next game world... don't bite off too much, don't overthink it... it's mostly opinion, so find somethign you like.