There have been various arguments and complaints about the process of optimising your team for B. Bashy teams which take no rerolls, choosing not to apothecary injuries because they make the player better value for TS, and turning turning down +St rolls because of the TS cost. Rather than getting into the philosophy of gaming, I'm going to suggest something that addresses what I see is the core issue. It is, I think, basically the same core issue that applies any time something is described as cheesy and/or powergaming:

The strategies for playing optimally do not align with what players are 'supposed' to be trying to do.

People are supposed to be managing a blood bowl team. Tuning that team to an external and artificial rating scheme is a different task. Currently in B, people need to tune their team for TS, which is much more about understanding TS well than about understanding blood bowl. Ideally, people should be able to play optimally without needing to know anything about the rating system at all, just the ordinary rules of blood bowl. My solution is to align the two tasks of team management and blackbox optimisation as much as possible. So what does that mean?

I propose a new rating scheme that I call the Team Blackbox rating (TB). The goal is to give a rating where matches are as fair as possible, while still ensuring that tuning your team for TB is as close as possible to normal team management. Two other things that are generally worth keeping in mind are maximising the variety of close to optimal strategies (e.g. balancing out the different races), and maximising the opportunity to use skill to your advantage (e.g. not allowing the rating to try and compensate for people's bad skill choices).

The key to this is starting from a base line of TR. TR is at one extreme, as optimising your team for TR aligns perfectly with ordinary team management, and the winnings and handicap systems already require people to do it so it does not involve learning anything new. However it often doesn't lead to fair matches. From this TR base line, I will then give a series of suggested changes, in the order of what I see as most to least desirable.

0. Starting point is TR
1. Exclude unspent cash
2. Players with mng injuries are not included in the rating

Even stopping here would give a reasonable rating system for B.

3. Replace the 1 point per 5 spps with a fixed 2 points per skill (see also #6).
4. Replace ((player cost) / 10k) with the most up to date rating for the player's base cost - e.g. as calculated by TS.
5. Multiply a player's final value by 5/6 for each niggling injury.
6. Give a higher value for the harder to roll skills as per lrb 5 (i.e. 3 for doubles and +Ma, 4 for +Ag and 5 for +St)
7. Reduce the contribution of high fan factor to (fan factor/2).

This is about the point where I'd choose to stop. The suggestions below are ones I would still consider, but currently think on balance aren't worth doing - either because of down sides or because of implementation effort.

8. Cancel +stat skills with a -stat injuries on the same stat.
9. Some rosters simply aren't as good as others in a way that's not bourne out by the value of the players. Where appropriate, an offset that's a function of the race and possibly the TB level may help.
10. Various adjustments to the value of things like wizards, apothecaries are possible.
11. If the scheduler could handle it, an offset for bad matchups in race pairings could be included as well.