Tuesday, October 6, 2009

Not hedge fund clones, but half-brothers

Mebane Faber has a great post on the topic of hedge fund cloning. He is correct in saying that you invest in hedge funds as pure alpha vehicles (hence the high fees), not because of their correlation characteristics.

Consider the following thought experiment. If I showed you a fair roulette wheel where the house has no advantage and I told you that I was going to bet a controlled amount once a day on a spin and invest the remainder in T-Bills. Such an investment would be uncorrelated to virtually any asset class that you can think of and would have highly diversifying characteristics, but would have no alpha.

Would you invest in such a fund? Would you also pay 2 and 20 for that privilege?

If hedge funds had no alpha, then that’s what you would be doing.

Do portfolio holdings tell the whole story?
Faber goes on to survey hedge fund replication techniques and finds them wanting. He goes on to extol the results of his AlphaClone service.

While AlphaClone’s real-time out-of-sample returns appear to be impressive, does that get an investor all the way there in hedge fund replication?

As I understand it, AlphaClone tracks the holdings of top investment managers through their 13F and other filings. However, there are a number of limitations to this approach, even if we assume that we can identify a smart investor universe:

  • The information may be dated, especially if the fund is a high turnover fund.
  • Limited disclosure: Many filings disclose only the long portion of the portfolio and does not show the short portfolio.
  • Less attention to weights: Some of this analysis focus on the names of the holdings only, without paying attention to the weight. For example, Julian Robertson reportedly entered into a yield steepener trade. How big a bet is it? Does it represent 0.1% of his portfolio or 10%? Focusing on picks ignores all the other parts of portfolio management (see my previous comments here, here and here) and Dash of Insight also has a timely post on the importance of weighting differences between ETFs. I don’t know how much AlphaClone pays attention to the weightings in the portfolio but a lack of attention to weighting can prove to be a problem. One solution is to reverse engineer a manager’s macro factor exposures like I have here.

A half-brother...
In short, I believe that approaches like AlphaClone is the next step forward in hedge fund replications, but it doesn’t get you all the way there. What you get is more like a half-brother rather than a true clone.

At this point in time, the only way to get the true exposure to these funds is to buy them, rather than to try to clone them. Whether the difference between AlphaClone like approaches and directly buying the funds is worth the 2 and 20 fees is up to you.


market folly said...

Hey Cam, good post highlighting some of the flaws of hedge fund portfolio replication. Meb's post was indeed a great one and I just thought I would interject a little bit on some of your points.

I've used Alphaclone extensively since I like to track hedge funds and once inside, you actually have the ability to put each position equal weight or do "manager weighted." This takes the overall value of the position as determined in the 13F and compares it to the overall amount of assets listed in the 13F so that you can see which long position is their relatively largest stake.

However, this obviously has limitations too as it is limited to US Equity positions and does not include international markets. For instance, pulling up a 13F on Greenlight would show their largest position to be xyz, when in reality their largest stake is an international company abc.

So while the limitations are certainly there, you can do a makeshift weighting by using the assets reported on the 13F to create a "match the manager's weighting" clone.

Alphaclone has 14 day free trial right now if you want to check it out and just poke around for a few weeks to better examine hedge fund portfolio replication. It's definitely not perfect but certainly a step in the right direction. Unfortunately, unless we see pure 100% transparency in the industry, it will be impossible to truly clone their portfolios as the disclosure of short positions would be essential to highlight synthetic positions, boxed shorts, etc.

Here's one of my posts on a portfolio I created with Alphaclone if you had not seen it: http://www.marketfolly.com/2009/04/market-follys-custom-hedge-fund.html

Keep up the great work as always!


keithpiccirillo said...

The plurality of investor's don't need to get "all the way there" and Faber's body of work appears to give serious bang for the buck, albeit it does have a short term track record using long term information.
The recent Ivy League performance is disconcerting.
Ironically, there was an article yesterday, that said at market lows all the market timers reappear, and argued it's time for buy and hold again. I dunno.
As an aside, fifteen years ago I sold annuities and one of the products was a market timing service that switched funds between money market accounts and equities with 3 switches a year the average.
The results were not good. Look how far we've progressed.
Both of your sites have taught me much about the dangers of modeling/false assumptions, and LTCM type events.
I regularly read the Goldman Sach's releases, which can somewhat help corroborate who was where at a point in time, and trends are underrated.
The skewness against the retail investor will never change, so I bet on proven high alpha long term jockeys in various forms to get my value at risk.
Thanks to both of you, distant cousins.