Can someone talk a little bit about portable alpha? How does one eliminate market risk with derivatives? How is portable alpha a measure of a manager’s security picking skill?
You can read more about portable alpha here: http://www.allaboutalpha.com/blog/category/portable-alpha/ The way I understand portable alpha is that when constructing a portfolio of assets with a given beta exposure, by adding uncorrelated assets that provide additional return, your alpha becomes portable. Consider having a portfolio consisting of one investment such as an S&P index ETF (your beta exposure). Now in order to get portable alpha you need to add an uncorrelated asset, say a private equity fund (your alpha). Since your PE fund is uncorrelated with your S&P index ETF your portfiolo would now have portable alpha. If you had a portfolio of the S&P index ETF and then you selected some additional stocks for your portfolio, any alpha achieved would not be portable since the investments would be correlated.
Wow… that’s not at all how it was explained to me? Here’s my understanding… determine the beta of my portfolio using mean variance formulas, etc. short out the equivalent amount of broad market exposure using derivatives, etfs, etc. In other words if my portfolio has a beta of 1.2 and is $1 zillion dollars then you would short $1.2 zillion dollars worth of SPY. You are thus left with just alpha. My portfolio has no market exposure. Alpha will be *consistently* positive if I’m a good stock picker.
I’m going to have to agree with virginCFA.
Different definitions of portable alpha? I thought the whole idea of alpha being “portable” is that your alpha can essentially come from anywhere (PE Funds, futures, commodities, classic cars, etc) and your underlying beta exposure is not affected, although not necessarily eliminated.
Here is one of the papers I remember reading about portable alpha. It is from the MS portable alpha team. http://www.euromoney-yearbooks.com/images/143/downloads/MORGANSTANLEY1FINAL.pdf
Basically you have a portfolio that generates alpha. The two possible sources of alpha are security selection and market timing; however, most people believe that security selection is more doable than market timing (even though i think that security selection is actually a kind of micro-market timing exercise). If you remove the possibility of broad market timing, then all alpha is the result of security (and possibly sector and industry) selection. When you have your portfolio together, for a variety of reasons, it might have some correlation with other benchmarks. That is basically a residual beta: residual because you weren’t actively trying to get beta, it’s just there because each of your security picks has some beta in it and it didn’t just cancel out. So what you do is take a short position on that beta’s benchmark index (short, assuming that your portfolio’s beta is positive, which it usually is). That makes it so that your portfolio has net 0 beta exposure to that benchmark and all that’s left is “pure alpha,” which, since it’s uncorrelated to beta, can now be “ported” onto any portfolio that uses the benchmark you just neutralized. You can do this more or less with any benchmark, as long as it turns out that your source of returns doesn’t actually come from trying get exposure to that benchmark or market time it.
So what makes it “portable”? I don’t necessarily believe in portable alpha, but it sure doesn’t come from buying a mess of stocks and then shorting S&P futures. You would walk into a hedge fund and say you have portable alpha as long as they can come up with a gajillion dollars for the zillion dollars of stocks.
This whole idea of portable alpha is clear as mud.
Well, isn’t the main point that your alpha and beta are by definition uncorrelated with each other, and the important thing is to remove all beta exposure so that what’s left is just the alpha component. You can remove it with a future or you can remove it by constraining an optimization to produce zero beta. Then, since your alpha is uncorrelated with beta, the alpha can be “ported” onto another portfolio with any benchmark that you’ve removed from the portfolio. That other portfolio may try to collect other portable alphas which is a good idea as long as the alphas are from uncorrelated models. The portfolio that is trying to import your portable alpha might have 100MM in it, get its benchmark exposure from buying futures, putting down 5%, and using the other 95MM to invest in various portable alpha portfolios that are designed to have zero correlation with its benchmark. The challenge there is 1) to ensure that whatever portable alpha sources you add are not themselves correlated, and 2) to make sure that you aren’t taking on more risk than you think you are taking with the remaining 95% of investable assets. Anyway, not promising that I understand it full, but that’s how I understood it.
bchadwick, i really like your explanation.
So first off “portable alpha” is about 1-part truth and 1-part rhetoric. bchadwick has a good explanation for the alpha part (and, of course, that’s the part that will get you tons of song and dance from the hedge fund managers). The portable part theoreticaally means that you can get this alpha without disturbing your other holdings or committing “any” capital to it. I would say that a long equity portfolio hedged with futures is not portable alpha because the stock position will require capital. Hedge it out with futures contracts and you can probably convince your broker to give you really good leverage but it can’t be overlaid on anyone’s portfolio without causing some cash drain. Of course somone selling this would say that you can probably get 10-1 leverage so it’s mostly portable. Currency overlays are much more portable as you can probably add that to almost any prime brokerage account with no additional capital required. A good FX trading program would almost certainly be portable alpha.
it is porting another alpha source onto a beta not associated with that alpha source. for example, say you have a large cap portfolio and don’t think that you will be able to beat the benchmark, so you want to use alpha from a small cap portfolio and “port” it onto the beta from the large cap portfolio. You buy Russell 1000 futures and sell Russell 2000 futures. The Russell 1000 futures give you the large cap beta and the Russell 2000 futures removes the small cap beta from the small cap portfolio. What is left is the alpha of the SC portfolio and the beta of the large cap market. I think it is used if you think you have a better chance of outperforming in a size, asset or style but want to use that outperformance in a size, asset or style in which you have a lower chance of outperforming (and you probably expect the beta return to be higher in this second asset class). dunno if that makes sense to you.
If you have an equity portfolio but you are a better Fixed Income manager and can create alpha by trading bonds. You can convert you equity holdings into bonds and enter into a index swap where you pay away the fixed income benchmark return and recieve the equity benchmark return. If you created alpha on you fixed income, you have ineffect ported it to your equity portfolio, and earned a net return of your fixed income alpha plus the equity benchmark.
Something definitely sounds strange here. I would certainly like extra expected return with no additional outlay of cash. I’ve recently realized that how much cash you outlay in an investment is actually not terribly material, other than the simple fact that if it requires cash that you don’t have, your options are pretty much 1) borrow cash, or 2) don’t invest. But there are some investments that require essentially no cash outlay, and some option strategies give you a credit upon entry. The real question - whether you have a cash outlay or not - is how much risk you assume, because if your portable alpha manager screws up, you are definitely going to have some cash outlay to cover those losses. If you short a stock, for example, you don’t actually need to lay out any cash, so that sounds great, except if the stock moves against you by X%, you will be laying out X% cash after all. So I’m not sure how portable alpha managers can truly offer you some alpha without asking you to lay out some cash. I just figured that you were finding the cash by using the implied leverage you get from buying a futures contract on whatever benchmark you are being measured against. (BTW, I liked the Fixed Income example)
Some of the issues with portable alpha are ensuring that you have a clear and good understanding of all asset classes being used. Otherwise, the chances of alpha manager “screwing up” is higher and you are going to have to deal with clients asking why bonds are sinking their equity portfolio.
Trevor in T.O. Wrote: ------------------------------------------------------- > If you have an equity portfolio but you are a > better Fixed Income manager and can create alpha > by trading bonds. > > You can convert you equity holdings into bonds and > enter into a index swap where you pay away the > fixed income benchmark return and recieve the > equity benchmark return. > > If you created alpha on you fixed income, you have > ineffect ported it to your equity portfolio, and > earned a net return of your fixed income alpha > plus the equity benchmark. Is there anybody in the world who would let you do this (ans: not with their money).
JoeyDVivre Wrote: > Is there anybody in the world who would let you do > this (ans: not with their money). Investment consultants + pension fund trustees = stupid ideas + large fees
Best description (don’t know about execution) of PA I’ve heard was actually from a local student run investment fund. They take the cash and equitize the entire fund via SP500 futures, then make only equal size pair trades to try and eek out alpha on top of the index return. Obviously this would require some capital to be tied up.
Trevor in TO is completely right, and there are lots of people (institutional clients) who want exactly this product.