Yes, your formula is correct. It depends to which degree you estimate the number of remaining decks (full, half or quarter deck).
You are right an IC of 0.91 is excellent but you can raise it to almost 0.93. All you have to do is to add your SC of aces to the Running Count of the main count. This method was developed by a poster named Dancer, who posted this technique some years ago over at the board at advantageplayer.com.
(Dead link: http://www.advantageplayer.com/blackjack/forums/bj-main/webbbs.cgi?read=6479)
This way for Single Deck you would take insurance at when RC+SC>=+5.
For DD @>=+9, for 4D@>=18 and for 6D>=+26. You don't need to calculate the TC, just add both counts together.
The indices were calculated with CVData, they differ a little bit from his numbers.
Another way to improve your insurance correlation would be to use the Victor Insurance parameter where you divide your RC/Aces remaining and take insurance when your result equals or exceeds 1.17 for Hi-Opt II.
http://www.blackjackforumonline.com/content/victorinsuranceoriginal.htm
I believe the V.I.P. is a bit more accurate but I found no way to sim it with CVData.
**Edit:typos in the spreadsheet