An opinion coming at it from a different point of view;
Running the numbers, backtesting, analyzing and re-analyzing, is fine and the math makes people feel like it is scientific.... and to some extent it is.
I believe that one can 'exceed the precision of the model' ... and many do it all the time. IMO, I just need to get close, because there are way too many factors that we don't and can't control. ... like the economy, markets, war, terrorism, Katrina, ...etc., that will affect the stock and bond markets which will affect our finely tuned AA portfolios.
What the math should do is 'get you close' to the answer that allows you to 'sleep at night'. Something that you read and re-read a number of times and come to the conclusion that 'yeah ... this sounds right'.
Backtesting all of the different AA's tells you what has happened, not what will happen. ' ... past performance is not an indicator of future results ...'
My approach;
1) pick the AA that makes sense, in my case (pension/no pension, age, amt of assets, how long I think I will need this to last, ...etc.) 60/40 Equites/Income
2) Figure out allocation of the 60 part and the 40 part that gives me comfort (growth) and income to provide the funds to 'live on'
3) check it every year or so and get the AA back in line
4) go off and live life ... and stop my former obsessive ways