I can offer my experience as a new manager with these recommended players.

I started Dec 2. As best as I can remember, the first player I signed (recommended) was a GK. I then signed a ST, followed by an AMC/AMR, followed by an AML/ST. It's possible one of those players was actually purchased from the "Scouted" players section for cash + tokens. Again, can't remember 100%.

After that I was recommended a batch of ST, none of which I signed...

Right around that time I decided to transition my roster to not being full (only 19 or 20 players, after always having the full 22) AND to being more balanced across Midfield and Attacking, specifically.

This is because in my first few weeks of competition I quickly adopted the 4-V-1 formation, and to support that I carried 7-8 Green Dx eligible players, making sure to have some DMC eligible players as well. Basically, my roster was defense heavy. All solid quality players, but I wasn't happy with the match results.I felt there was simply more scoring potential to be had. (Still a great formation against top competition, I don't mean for this to be an opinion on formation)

BUT in the past couple weeks NOW I am often simple running out formations of just 3 DC across the back line. No DL or DR or DMC. As such I started carrying only 5 total Green Dx eligible players. I wanted to commit to a larger presence in the midfield and specifically in the top 6 attacking box / grid. (Again, not lobbying this is correct, just what I did)

And then finally I was recommended my first batch of Defensive players - All were DL eligible... a position that I only had a single player eligible for on my roster. And as you can guess, none of them were worth the tokens and they remained unsigned.

So, this is an incredibly small sample, but here are a few points that might be something:

1. I think the engine logic might operate best (for managers) when you don't have a full roster.

2. I think the engine logic aims to balance out your roster once you do have 1 or 2 roster openings, regardless of your preferred formation.

3. I think the engine simply breaks when you have a full roster + very high quality on average. As we approach the end of this season mine is hovering around 104%, probably very high for ML-2. By "breaks" I mine it will just recommend players that it weights as the most valuable, i.e. those that most impact goal scoring. ST and GK.

Lots of assumptions on my part, but maybe seeing how the system works for a new manager can help determine its logic.