MachineGhost wrote:
Markowitz has nothing to do with prediction. What I meant to infer was you could over-exposure yourself to risk by over-concentrating in certain stocks when equal-weighting. Equal-weight means equal $, not equal risk. Mean-variance optimization would ideally reduce the weight of those in accordance to their increasing correlations
I know Markowitz did not attempt prediction with his seminal work in the 1950's and 1960's. But when you use his theories that explain past behavior, combined with statistics derived from that past behavior, in order to structure your portfolio in anticipation of future behavior, I don't know what you can call it other than prediction.
As for trying to equal-weight risk, you first have to define the risk of concern. Markowitz and all that I've read regarding subsequent application of his theories in creating modern portfolio theory, have assumed that risk is equal to price volatility. That risk does not concern me. Further, I've seen enough evidence in historical numbers that I know that attempting to predict future price volatility based on previous volatility is of minimal if any use at least over any term of interest. So I don't see any reason to bother with it. Instead of being worried about price volatility, I treat such as a bonus opportunity to be harvested if convenient, or ignored if not. I was able to make a great harvest in late 2008 and early 2009. Maybe I'll get lucky and it will happen again.
The risk that concerns me is the risk to income, as caused by a company cutting their dividend. One might attempt to balance the that risk by structuring the portfolio so as to equalize the contribution to income of each holding, otherwise known as equal income weighting.
I tried that. I don't do it any more. It was too hard for the little benefit it provided. It made it hard to know where to put additional investment, when volatility could be harvested, etc. so I went to equal capital weighting. 2008 convinced me that my alternate approach to managing income risk is much more effective. That method is to pay attention to the companies involved, and the macro events which uniquely impact them, and to individually assess whether each dividend is at risk.
Since 2005 or so I've made only three significant mistakes in that assessment, including one where I was correct but allowed myself to be convinced not to act on my judgment. That was GE, I sold 1/2 my holding but kept the other half because of public pronouncement by the CEO that the dividend was safe. Shortly afterwards they cut their dividend. Another was BP. I thought they had enough cash flow to cover any reasonable scenario. I still do. They even announced the dividend but political pressure from the U.S. caused them to cancel before paying. The final was WaMu. They had enough capital to have qualified for a bailout like many other players their size and situation, but I failed to account for the influence JP Morgan Chase has on the Fed, combined with WaMu previously spurning JPM's buyout offer.
Still in spite of those mistakes, there has not been a single year since 2002 that my dividend income has not increased. Losing GE and WaMu both in a single year was harsh, but increases from other companies and redeploying capital (like from selling BAC long before the dividend cut) into the stellar bargains available at the time more than made up for them (always helps to have a bit of good luck like losing my job in 2008).
