I went to an intriguing site, finportfolio.com, and set up a test portfolio with four stocks: MSFT, MCK, IDXC and HCBK. (I originally had eight, but the site did not recognize the four smallest.) I bought some hypothetical shares in each without thinking too much about how to divide up my hypothetical funds.
I then clicked to have finportfolio.com “optimize” the portfolio. I had read that by using sophisticated techniques, it could re-weight a portfolio to increase its expected performance without increasing its risk.
Finportfolio.com chugged around for a few seconds, performing the kind of calculations that would have taken Mathematicus himself 400 years to complete with slate and chalk, and concluded that I could lift my expected return from 14.83% to 37.43%, without adding risk.
I could achieve this dramatically better mix, it told me, by dropping two of the four holdings altogether and cutting back the third from 25.68% to 13.30%. That way, MSFT would be 86.70% of this small but optimally balanced portfolio.
|Current Weight||Optimal Weight|
My first thought was that this is really remarkable technology. To think that we now have Modern Portfolio Theory models at our fingertips!
My second thought was that this is a little scary.
Because it appears so authoritative — displayed to two decimal places, no less — one might easily jump at the chance to put its results into action.
But can it really make sense? Will the volatility of a portfolio containing MSFT alone (well, virtually alone) really be no greater than that of a more diversified portfolio?
In an even more stunning display of what we can now do instantly and for free that Mathematicus would have labored a lifetime over — and that would have defeated his slower-witted half brother Arithmeticus altogether — a visitor to finportfolio.com gets to click on the horizontal axis of the “Efficient Frontier” graph (not shown here) that accompanies the optimization table. Click to the left or right, and you adjust the level of acceptable volatility . . . and see the optimal portfolio weightings change instantly as you do.
Basically, the more volatility I am willing to accept, finportfolio.com tells me, the more MSFT should dominate the portfolio. If I am willing to make it 100% MSFT instead of “just” 86.70%, the expected return rises yet a hair more, from 37.43% to 37.74%.
Or, if I want the absolute lowest possibly volatility with this combination of four stocks, the model says I should weight them this way:
|Current Weight||Optimal Weight|
Notice that now the volatility figure drops to 26.94% — less risky than the 32.25% in my current mix — and that this reduced volatility comes at the sacrifice of some expected return. With this mix, I should not expect the phenomenal 37% annual return of the prior mix. The model says I’d nonetheless be improving dramatically on my current mix, raising my expected return from 14.83% to 25.20%
But how does it know?
Of course, it doesn’t know. What it does is use assumptions based on the past performance of these stocks to do some very sophisticated, brilliant, and Nobel-prize winning calculations to arrive at its expected results.
But it still doesn’t know.
If it did, it could simply optimize (for example) the S&P 500 index, weighting not by market cap, as S&P does, but by Modern Portfolio Theory black-box magic, and come up with a Nobel-prize-winning index fund that significantly outperforms the index with no more risk.
You can be sure people have thought of this and tried it. Where is it? And why wouldn’t most pension fund managers and bank trust department managers, who have long had access to these computer models, not have generally beaten the market as a result?
Years ago I got to co-host a PBS Documentary called “Beyond Wall Street” that devoted a half hour to each of eight amazing money managers (all living and working somewhere other than New York City). One of them was Barr Rosenberg, outside San Francisco, a brilliant pioneer in much of this quantitative analysis, and who had the kind of computer software, hardware, and global data network that would could turn the average banana republic into a junior superpower. And do you know what? Some years he had done very well, other years not so well — just like you or me. Overall, I was fully persuaded he was smarter and more capable than I’ll ever be . . . but I was not persuaded he could, even with all that fire power, beat the market by much, if at all.
Finportfolio.com is fun to play with, and I happen to think MSFT stock may do quite well going forward. (Hope so: I own some LEAPS.) But that sort of judgment may be nearly as unreliable coming from a black box as from a human brain.
Tomorrow: I Give a Young Genius $20,000 to Beat the Market