Bob Prince, Co CEO of Bridgewater, on Alpha and Beta in HF Portfolios

>

Bob Prince

Now, if you look back at the history of Bridgewater, for the average market that we trade, we made 1 percent return with 3 percent risk. We’ve had a .3 ratio for each market. Our overall ratio is about 1. The difference between these two is diversification and portfolio structuring. So, two thirds of our performance has come from balancing risks well, and only one third has come from trying to get the bets right. Of course, if you don’t get the bets right, you don’t have anything to balance, so you have to start there, right? Okay. So, that’s point number one. You can get a lot more mileage out of risk reduction than return enhancement.

Number two is don’t believe the numbers. Now, I’m going to make an assertion here. I don’t know if you’ve ever heard this assertion before. My assertion is that correlation is unknowable. So, we talk about portfolio mean variance, about, correlation. My assertion is that correlation is unknowable. I’m talking about the correlation of any two assets – it is unknowable. Now, I’m going to try to prove that to you on a chalkboard.

So, let’s just say that I have stocks and I have bonds. Okay? And the question is, what is the correlation between stocks and bonds? Well, correlation is the way that the returns co-relate. It’s how the returns relate to one another. In order to understand how the returns relate to one another, you have to really understand something about what drives returns. Okay? Well, there is more than one thing typically driving the return of any one asset. Let’s just say I’ve got economic growth here as a factor that drives the return of stocks and bonds. Now, if the economy is strong, that will be good for stocks – generally speaking if the economy is stronger than expected. If the economy is strong, it will generally be bad for bonds. But the economy’s not the only thing that drives their returns. Let’s say I’ve got inflation here as the factor that drives their returns. If inflation falls, that will generally be good for stocks. If inflation falls, that will generally be good for bonds. So, inflation is working the same direction for stocks and bonds. Economic growth is working opposite directions for stocks and bonds.

   
So, now the question is, what is going to be the correlation of stocks to bonds? I don’t think you can possibly know. You can’t know unless you know what kind of environment you’re going to be in. Will I be in an environment that’s dominated by inflation? If I am, they’ll probably have a positive correlation. If I’m in an environment that’s dominated by economic growth, they’ll probably have a negative correlation. But if I knew what kind of environment I was going to be in, I’d just go bet on that. But if I’m trying to passively structure a portfolio that’s balanced, I don’t have any idea what the correlation between stocks and bonds is going to be. And what it was in the past really has no bearing on what it will be in the future. So, now what do you do? How do we do mean variance optimization when you can’t possibly know the correlation of two assets?

So, what I’m saying is you can’t do this. So, what we try to do is we don’t pay any attention to the numbers. We never try to look at the correlation between any two assets. What we try to do is balance our exposure to economic growth and inflation. So, given the structural characteristics of assets, they will perform a certain way given a certain economic environment. And you can look across your assets and start to think about how you’re balanced against the economic environment through your asset holdings.

   
The last point is that you need to understand the fundamental characteristics of the returns that you’re operating with. And the most important characteristic of those returns is the derivation of the return: is the return derived from beta or from alpha? Beta means the return is derived from the risk premium embedded in the asset. Risk premiums over time will be positive in order for the capital system to function. Risk premiums will be positive over a very long period of time, but it’s very easy to buy a risk premium, so it’s not a really good returning source of return. It’s not a very consistent source of return because it’s very attractive. A lot of people buy it and they bid up the prices, maybe has a ratio of .25 return to risk ratio. So, beta is one kind of return. It’s one type or category of return. There are lots of betas.

Alpha is totally different. Alpha is a bet. I’ve got a view. There’s timing involved. I’m long. Now I’m short. Now over time, people might make bets, but they might on average be long. Then on average they have a beta in their return, right? But … so the question is, are you generating your return through beta or through alpha, through risk premiums or through timing of bets? Because the characteristics of those is radically different, and the ability to produce a very high return is inherently limited if you’re basically holding betas because betas tend to be pretty expensive. You’ve only got about a .25 ratio. No matter what the historical numbers are … don’t believe the numbers. No matter what the historical numbers are, the return to risk ratio of a beta is probably not above .3. It may have an option characteristic that makes it look that way, but it’s not above a .3.

   
And they tend to be very highly correlated because betas are all related to the same economic environment, so it’s hard to get a lot of diversification in betas. Therefore, it’s hard to get a really high ratio — high consistent return — from betas. On the other hand, alphas are very uncorrelated, but you never know if you’re going to make or lose money because alpha is a zero sum game. So, it’s entirely a bet of can you bet on and find the right manager who can take money from somebody else. And if you know … if you think you can, you probably should … as a test, you might want to think about who they’re taking money from as a cross check on your process because somebody’s got to take money from somebody else when it comes to alpha. For every winner there’s got to be a loser.

Now, I just want to show you a quick example of the importance of understanding the composition of beta in the returns of a manager that you might hire. One of the things that we did was that we looked at our database of 2,700 hedge fund managers – we ran the calculations for how much beta is approximately in the returns of these various managers. Well, we did that up to the beginning of the crisis in July 2007, quantified how much beta was in every single one of these 2700 managers. And then, we looked at how those managers then performed over the crisis period.

And so, what this work shows is the managers with more beta lost a lot of money, and that those with not very much beta in their portfolio up to July 2007 lost less. Hardly anybody made money. And the observations are right on the line of best fit (though you should also look at their return to risk ratio).

So, the amount of beta in a hedge fund’s portfolio was something like, 96 percent correlated to what their performance has been since July 2007. So, if you just knew that one thing — how much beta is in their portfolio — you would have been able to identify with a 96 percent correlation how they would have done it through the financial crisis. The same thing is happening going forward because if you actually monitor this through the crisis for those managers, that beta hasn’t changed. They still have that beta.

Now, so if their particular beta does well now, they are going to look good. But it’s because that beta is in there. So, are you betting on the manager or are you betting on the market that they’re involved in? It’s absolutely crucial for you to understand that. And if you’re betting on the market they’re involved in, no matter what their historical ratio is, the beta … the ratio of the beta is inherently limited, so something like .25, .3.

So if you look at what Bridgewater does, right now we’re long bonds. There’s risk premium there that we’re earning today by being in a long bond position. But if you look back at what we’ve done historically, we’re long half the time and we’re short half the time. There’s no systematic bias to be long bonds. And if you understood our process, you would know how hard we try to make sure we don’t have that in there, right? So, literally indicator by indicator, market by market, we are approaching it with an expressed purpose of not having beta in our alpha. And we try hard to not let it get in there.

At any point in time, we could be long or short a market. Funds don’t have to always be market neutral. But what I’m referring to is a systematic orientation toward beta. What I’m saying is that the amount of beta that was in those 2700 manager’s returns was measured by a statistic, that a static holding of asset classes over many years was 80 percent correlated to their return, so that they are, over time, largely in a beta position.

I think the question of relative versus absolute is always trying to get at the real question of “what is value added?”. And there’s a lot focus on the industry on the quest for alpha, so to speak. But I think if you were to ask three people in a room of experts what the definition of alpha is, you’d probably get about four answers.

So, the real question is, what are you trying to do in getting value add other than exceed a simple passive benchmark? And, secondly, how is that going to influence people behaviorally?

That tend to lead you in a couple of directions, one in terms of more complexity — I’ll give you an example of that- our equity mandates. Our benchmark not against an overall equity index, but every security selection decision is against a sector in a country. So, you get six major countries, ten S&P sectors. So, we have something called a 60 cell matrix; you can imagine the operational complexity behind that. But every active choice is done against a very specific subsector so that the sector’s taken out.

The question is, though — and it gets back to that behavioral one — how is that going to influence people, and can you even think intuitively about that when you’re in 60 different dimensions? And so, there is something to be said for simplicity because the reality is that the very best tech fund manager in 2002 might have exceeded his peer group by 5 percent, but that meant they were only down 90 percent as opposed to 95 percent. And that’s not going to solve anybody’s problem.

Edited Transcript from The Greenwich Roundtable