What sort of answer are you looking for? * I thought Bernstein was pretty clear on the subject -- he even recommends tools that will help you find the "ideal" allocation:
I'll go read that section of the book again, but wasn't he just saying that by running long historical returns data from multiple asset classes through the optimizer (constraining to avoid corner solutions which just tell you which asset class 'did best' and aren't practical going forward), that is how he came up with the rough percentages for asset allocations? And remember, we are balancing return and risk (variation in returns, or standard deviation or volatility) so there are multiple solutions based on what we want to accomplish (return at or above a certain level, risk at or below a certain level. ) The Index Investor does all this pretty nicely in arrays of allocations and portfolios and currencies, though it costs $25 a year.
As for value tilts, Fama/French lays it all out pretty well (let me know if you need the paper citations), but it boils down to just looking at the historical data for the returns -- value kicks tail on growth over the long run.
Why? It seems intutitive to me (and therefore may be wrong!), but value stocks (high book -to-market ratio) have low prices vis-a-vis the broad market. Usually they have done something bad like missed a quarter on their earnings. Market punishes them. No portfolio mgr wants to be seen owning them. People over-react, and thus creates the opportunity for the buy-and-hold guy with a system.
Caveat: this is all about playing the odds over large numbers of these stocks over long periods of time, thus the need for a fund. It works often, but not always, thus we need to buy a fund of these things and not individual issues -- you dont want the Enrons, you want the Tycos, the fund will at least make sure you have both. And a fund mgr may do even better at sniffing out the rats at Enron, or buying tyco during the 3 months that it is a value play while the index fund might miss the Tyco completely. Why? The index fund is limited by the fact that it is recalibrated (new stocks chosen) just once a year. A new value stock can only enter the index once a year, at which point much of the price movement may have happened. For this reason, value funds are the one place I can kinda-sorta justify an active mgr, (and Windsor II outperforms VG lg cap value index consistently year in and year out).
Everybody knows growth stocks are great companies to own, because everybody loves their products and everyone knows the company is widely respected and successful. SO the price is already high, and returns are constrained. One slipup and the growth stock gets nailed. Own their products, but not their stocks. "Bad companies make good stocks" is talking about this counterintuitive phenomenon. If the value firm does anything right it will appreciate, sometimes a lot. The growth stock has to do everything right just to stay up there. That is why I hate the Cap-Weighting approach, and the S&P500 in particular. Its full of stocks that everybody knows are winners -- growth stocks. I like indexing, but I want to index more counter-intuitive out of favor sectors. (and sectors that don't move in tandem with each other: that is the Modern Portolio Theory part of the story, for some other post.)