Even if everything in NYAN.A goes to zero I will transfer all the assets in NYAN.B and in CPA to NYAN.A to keep it afloat.
If that fails (nuclear war or whatever) I'll just buy back shares with my own money. NYAN.A is the safest and strongest asset you can buy on the GLBSE today.
Having three nominally different funds is maybe more confusing than useful, when holdings and dividends are in fact more or less common.
Maybe you should (should have?) put it all together, deciding simply that NAV of NYAN.A is 1 insured, NAV of NYAN.B is the lesser between 1 and what's left, and NAV of NYAN.C is anything left.
Then manage it as you would with one portfolio to get the maximum return, regardless of what between A,B,C the funds come from. So that NYAN.C would be a high risk-high profit bet only on your ability to manage the fund, and not, as it is currently, also on the craziest possible assets available. This would have probably avoided losing so much on Pirate, for instance; and now on OBSI.HRPT i guess.
That calculation for NAV values works for .A, but the NAV isn't what's important for returns in the case of .B, just the weekly returns (The NAV of .A doesn't matter either as it's insured). NAV could be 0.02 as long as it made 100% weekly (this doesn't even begin to approach reality, just some hyperbole). Usagi does currently use NAV during sales to determine price but they're slightly uncoupled from returns.
At current NYAN's site lists about 3678 BTC worth of assets for 6972 Total shares (A:1606 B:1983 C:3383). So the overall with the above method .A would be worth 1BTC@1%/wk, .B would be worth 1@2%/wk, and .C would be worth ~0.0263 according to their respective NAV's (again, all this based on the above calculation, valuations and share count based on http://www.tsukino.ca/cpa/nyan/).
For reference, prices are around (at time of writing of course) .A:0.938 .B:0.97 .C:0.09.
As for simply maximizing returns, well people that buy into .A want a low risk investment, .B cares less but doesn't expect their investment to go into anything that looks too sketchy, and .C just wants maximum return. But that just means keeping share counts in line with the relative risk that's desired overall.
(This post seems long and dense even after re-reading, sorry)