Q&A: Roy Niederhoffer

Veteran hedge fund manager and Top Traders Unplugged host Niels Kaastrup-Larsen talks to the founder of RG Niederhoffer Capital Management

roy cut outRoy Niederhoffer’s path to becoming one of the world’s top short-term managers began with an early interest in computers and programming.

Aged 13, after a year of begging, he got a Radio Shack TRS80 – one of the first microcomputers – as a gift, and immediately set out to teach himself to program.

Niederhoffer decided he didn’t want to feed quarters into an arcade machine, but wanted to make his own version of the classic game Space Invaders.

In an early sign of his entrepreneurial spirit, he realised this might be something other people wanted to have as well. So he started a small company.

By the end of his High School years, it had about 30 employees and three other partners.

During the summers, Niederhoffer began working for his brother Victor, the pioneering hedge fund manager, and so was exposed to futures trading in his early teens.

However, Niederhoffer spent his undergraduate years studying neuroscience, notably how the structure of the brain has a tremendous influence on human behavior – an insight he took with him into trading – how to avoid cognitive biases, and then capture the tendencies that other people have.

When Niederhoffer graduated from Harvard in 1987, he had been set to go to Cambridge and into the field of neuroscience. But instead went to work [full-time] for his brother in New York, starting trading fixed income just before the 1987 crash, including some short-term strategies he developed himself.

Whilst at his brother’s firm, Niederhoffer was exposed to the likes of Monroe Trout, Steve Wisdom, and Paul Buethe, now part of Toby Crabel’s operation, and Crabel himself, who was there for awhile.

With an idea for an institutional version of Victor’s shop, in 1992, Niederhoffer left to form RG Niederhoffer Capital Management. He began trading in July 1993.

Since inception Niederhoffer has sought to distinguish his firm by providing not just diversifying, but truly protective returns that benefit a portfolio, to avoid trend-following and capture realized volatility in the short term-space – the very same things that he tells investors today.

Niels Kaastrup-Larsen (NKL): Before we go to your story, I wanted to ask you a question that I sometimes struggle with answering myself.

Imagine that you meet someone that you haven’t met before, and you start talking, and suddenly they ask you: “So Roy, tell me what you do?” How do you respond? How do you explain what you do?

Roy Niederhoffer (RN): The strategy that we employ has a very specific intent which may distinguish it from many other things out there.

We are trying to combine both interesting standalone returns with very consistent downside protection for people’s portfolios in equities, traditional investments overall and also alternatives.

What we try to do is maintain a consistent negative correlation to equities.

In other words, we do better than average when equities are having trouble, and there’s a lot of volatility, typically when portfolios that most people have are having their toughest times.

Our strategy’s actually tuned not to maximize our own Sharpe ratio, not to maximize our own risk-adjusted return, but actually to maximise the risk-adjusted return of our clients.

NKL: What do you think it was that Victor taught you, or what was so special about the environment that you were in at the time that actually has produced so many people who became successful in their own right?

RN: One thing my brother really taught us all was to avoid using charts. He was a tremendous opponent of any sort of charting whatsoever.

My view on why charts are not so helpful is that your brain wants to see patterns that aren’t there.

If you see too many patterns in the data, you make decisions that are not statistically based.

We want to make decisions based on probability rather than on our belief in visual patterns that may or may not actually be predictive.

So one thing that we’ve tried to do is avoid the traditional methods of visual pattern recognition that people have used in technical analysis.

I think one of the things [Victor] really emphasised was taking a very scientific approach to what one was doing. To not just believe things because everyone else believed them.

If you have to prove that something has statistical significance, you immediately have to have a testable hypothesis and a falsifiable hypothesis.

So, number one I would say: a scientific approach to the data, rather than an almost religious approach to the data.

Number two, I think there was a tremendous emphasis on not following the status quo, on not doing the same things everybody else did.

So immediately, you avoid the popular trades, and having to deal with 10 other or 1,000 other people doing exactly the same thing you’re doing at exactly the same time.

Because, typically, providing liquidity to the majority is a very good thing to be doing, you immediately start with a strategy that has a positive expected return.

I think looking at intra-day data was also something that was unusual at the time. It’s certainly a lot easier right now.

NKL: Now Roy, when I look at CTAs and hedge fund strategies in general, I see them positioning themselves to a large degree as a standalone investment.

But when I look at your program and your marketing message, it’s really a story about how you shouldn’t really look at your program in terms of a standalone investment.

You should look at it how it works and how it helps in combination with an existing portfolio of stocks and bonds, for example.
Tell me why you chose this specific approach which is different to how most firms position themselves?

RN: For better or for worse, we have chosen to tune our strategy to maximize our benefit to our clients rather than to ourselves.

That I guess boils down to my belief that people should be paid on the basis of alpha, not on beta.

It’s rare to hear me agreeing with Calpers, but I think their absolutely right that a lot of the hedge fund community gets paid on beta.

You saw what happened in 2008. Even this year, to have what happened in October (the equity selloff) occur [while] the hedge fund industry was 0.87 weekly correlated to the stock market this year and just about that last year.

How much alpha can you possibly produce? You have to produce incredible performance with that kind of correlation and very few people do.

So what we’ve tried to do is maximise the alpha that we provide.

The strategy from the top down is designed to have this negative correlation to equities, to hedge funds, and essentially to any type of portfolio that we see from a client to maximize the amount of alpha that we provide.

It’s been a bit of a quixotic quest, I know, because when everybody else is making money, very often we’ll stand out at the bottom.

But at the same time, if you look at our track record, our best years are 2008, 2000 to 2002, 1998, 1994, some of the selloffs in 1997, 2011, and we had huge performance in January of this year and the first two couple of weeks of October when there was this big sell off.

That’s what we a specialising in and to me that’s what a hedge fund manager is supposed to do.

The reason it’s important is that if you can reduce the size of your portfolio’s drawdown, when you get back on a positive environment you start the next positive cycle at a much higher NAV level.

So the amazing thing is if you take our strategy and the equity markets at a 50/50 mix – just exactly equal concentration – the combined return of us plus the equity is actually higher than the return of either us or equities.

It’s a complete mathematical paradox and the reason is that negatively correlated assets tend to meld together and work in harmony in a way that people are generally not used to seeing because it’s so rare to have positive expectation, negatively correlated assets.

To focus on standalone Sharpe ratio all you’ve got to do is add beta. It improves your Sharpe ratio until the equity market sells off.

To focus on standalone Sharpe ratio all you’ve got to do is add beta. It improves your Sharpe ratio until the equity market sells off

NKL: What about the managed futures industry? Do you see yourself being an outlier or outsider because it’s a little bit confusing I think for some investors because they think of managed futures, and certainly think of trend-following, as also being a protective element in their portfolio?

RN: I’d say our continual focus on maximizing the Sharpe ratio of our clients by providing both positive expectation and negative correlation at the same time makes us unusual.

Short-term strategies, to begin with, will have a much greater relationship to realized volatility than they do to trend.

The ideal environment for trend followers is of course a very quiet market that goes straight in one direction for years.

For a short-term manager, the ideal market could be completely flat over the course of the year but moving up and down 4% every other day. So that’s a very different “ideal” environment.

We’ve always described ourselves as majority contrarian rather than majority momentum, or primarily, or exclusively momentum of various time periods. So I think that’s a difference as well.

Eventually CTAs do manage to provide protection. That is definitely true.

My view on it is that you have to look very closely at the numbers and what they mean.
CTAs were not very protective portfolios during the first part of the decline [from] the highs of 2007 through, say, September of 2008.

At the very end of the year, the trend started to happen in fixed income in a big way. A lot of managers caught up.

Actually the same thing happened in 2000. There were a lot of profits at the very end of the big decline, but as of August, September, October, it was only some of the short-term managers who were up significantly after the stock market had really corrected.

Why is it important? It turns out that a lot of the damage in equity market corrections occurs very rapidly and on a very small number of days.

If you can protect your portfolio on those big days, you actually have a huge benefit to your overall portfolio because whether it’s a big decline in equities or a small one the next cycle begins at a much higher NAV.

So long as that protective strategy is actually earning you money itself and providing a positive return, the combination has an incredibly positive impact on the [portfolio] return. I don’t think too many people look at their portfolios in that way.

A lot of allocation decisions are made [like]an all-star team, you just want the best manager on a standalone basis.

I think in sports we often see that the best teams are not the best combination of the best individual players.

In sports we often see that the best teams are not the best combination of the best individual players

NKL: We’re going to be spending most of our time talking about your largest program, but I do want to offer you just the opportunity just to say: “This is what we do today…”

RN: Sure. We are employing a core strategy that has a number of different timeframes ranging from a few minutes at its shortest to a few weeks at its longest.

Both momentum and counter trend contrarian signals all of which is done systematically.

We have about 60 or 70 individual strategies that we put into, you might call them ‘style buckets’, we call them ‘families’, and the whole thing runs essentially automatically, I like to say, in a way that an aircraft flies.

That’s the way our strategy works, where most of the time the aircraft is on autopilot and does a great job of flying itself.

Every once in a while it’s necessary for the pilot to step in In the most difficult moments, like landing the plane, that’s a very important time for the pilot to be at the controls.

But 99% of your flight is automated. That’s what we do for all of our different programs. It’s the same set of models.

We have tuned our core strategy to have this negative correlation to equities, to hedge funds, to portfolios in general.

That is our Diversified Program and that is the same program that I would have been talking to you about 22 years ago if you were thinking about being one of my first investors.

It has remained our flagship product. We haven’t changed it.

That strategy is supposed to have about a -0.3 to -0.4 correlation to the equity markets.
We decided a few years ago to say “What would happen if we just brought that up to zero?”

Part of the intent of it was that I needed it. Being so negatively correlated in my business, I needed some risk-on.

So we brought it up to zero, and we created something called the Optimal Alpha Program.

This is a program that has in the last couple of years as you can imagine, outperformed.

Then we have another program called iHedge. iHedge is a program that is again reflecting one of my own portfolio needs. I was very concerned about inflation a few years ago, that we’d have rising interest rates and rising commodity prices.

Now obviously, I was completely wrong about that macro call, but I wanted to have some of my own money invested in it so I created this fund called iHedge which shifts the emphasis of our program so we have more buying of commodities and more selling of fixed income.

NKL: As a short term manager, you have other needs and constraints to longer-term managers, so tell me about how your business is set up today, and what the infrastructure looks like in order to handle that?

RN: The infrastructure that we’ve developed has had 22 years to coalesce, so everything we do here is our own code that we’ve written.

Our platform is generally in C++ and everything from our top down allocation tools to decide how much we’re going to do of each piece of our strategy and when we’re going to do it, down to the algos that we do to execute our code, everything is developed here and is run out of this one office.

I do run three shifts. I have to have people “flying the plane” essentially 24/7 while the markets are open.

So we have about two dozen people here, almost all of them are on the investment side.

There’s probably 13 or 14 these days – people who are quants – they’re all programmers, very good programmers.

They’ve all worked the overnight shifts. They’ve all run the strategy – essentially flown the plane.

We don’t have to do very much to actually trade. Signals come up, they’re sent directly to the markets and everything really pretty much operates by itself.

So what most of my people are doing is working on how to make the strategy better; working on risk management; working on the new ideas; making old ideas improved, or perhaps deciding we’re not going to do certain things that we have done for awhile.

A lot of it is offline and very creative. I like to think of it very much the way, say, a scientific research lab would work in a university, where you have a professor who has a lot of suggestions and tools at his disposal, but in reality it’s the grad students who are providing the creative experiments and actually doing the coding and actually running the testing of hypotheses.

It works very much in the way that I had some experience in back in university doing research on neuroscience.

NKL: I know you’ve already talked a little bit about it, but the Diversified Program, tell me a little bit more about, from a top down point of view, why you’ve designed it the way you do? What’s the rationale behind that?

RN: Given that we have focused the mission of the business on providing diversification and protection, I guess it’s not surprising that we apply that same idea to the strategies that we employ.

We’re trying to do a lot of different things: to have them be as different as possible and to allow them to operate effectively in harmony with each other.

So in the design of the program we have tried to think about some interaction effects regarding risk management.

We’ve looked at offsetting of trades and how that is important in reducing costs.

We’ve looked at the tendency of models to perform or not perform at certain times, and also to make sure that when we do this we don’t believe our backtests quite as much as one might if you had a pure quant approach.

We’ve tried to use models that contain interesting ideas at their core that we believe in, that we’re going to stick with, that we think should be robust, that are as elegant.

I don’t want to say “simple”, but they’re not over-parameterised, And to not force our own tinkering and biases onto the approach.

So what I believe we have right now is a group of nine different styles – ranging from minutes to weeks – that complement each other and each of them represent a different way to think about the markets.

NKL: Can you give me an example. What’s a way to think about the markets?

RN: One of the theories that we believe in is that the recent path that a price has taken is highly influential of what happens next: in the mean that one achieves, in the shape of the distribution, in the direction of the path and the tendencies that that path has.

So when we think about what makes our styles of trading different, we’re thinking of how different types of paths, be they strength of momentum, or mean reversion after a market becomes oversold, the paths that get us to those conditions have a certain consistency and that’s what makes a style of model for us.

So it’s a type of effect that we’re capturing in each one of our models that defines each style.

NKL: Now tell me a little bit about your system in the sense that you have this kind of 10 step process from idea generation to actually implementing something in the system.

What’s the thinking about going from the initial stages to hopefully end up with something that you can trade?

RN: I think the goals of our research process- which is designed to create viable, robust strategies that succeed over the long term- is very much the same as the way one would approach any sort of good scientific process. The beginning of a good scientific process is having a good hypothesis.

A good hypothesis is one that is extremely falsifiable, so the classic example is all swans are white. Well, that’s easy to falsify if you come up with a black swan.

That’s what the original black swan idea was, that it’s an easily falsifiable hypothesis.

So we prefer highly statistically significant, very frequent observations – things that occur a lot, rather than things that occur very rarely because we can easily tell when something has stopped working.

So to falsify one of our trading strategies, it has to be statistically significantly different in actual use for a certain time period versus what it previously had done.

The second idea is parsimony. We want to have as simple an idea as we can possibly get away with before it becomes trivial.

This is something we’ve really learned over time. The simpler, the better. It is so easy to fit the data.

The simpler, the better. It is so easy to fit the data

The data is just waiting to have the models just squeezed out of it with enough variables.

So we have really tried to focus on parsimonious ideas – the kind of thing you can just take a couple of sentences and explain, not complicated parameter sets.

The third one is robustness. We are looking for things that are as consistent as possible across markets.

Now this is something that not everybody does. We believe that if what we are finding are things that are true about the way people approach markets, then they should be true in whatever market we try, be it Google, or Soy Beans or dollar/Swiss or Two Year Note.

Ideally what we’re looking for are things that are very, very consistent across asset classes and across all 54, 56 markets that we test them on.

That gives you a large number of observations.

A lot of people don’t do it that way. A lot of people have specific systems, by market, by direction, and that’s okay too.

There are a lot of different solutions to this problem, but that’s how we do it.

NKL: The next topic is risk management. I think it’s such an important part of what we do.
I wanted to ask you, in broad terms, how do you define risk? What is risk to you?

RN: I think volatility is one synonym for risk. I think there’re different aspects of risk that go beyond that.

You can have a very quiet program with liquidity risk, and you have a left tail that doesn’t appear until you actually need liquidity.

I think people learned that in 2008 that just measuring volatility wasn’t enough. As long as you’re in the most liquid markets, and they’re open almost all of the time to 100% of the time, then volatility starts to be a closer measure to what your risk actually is, at least in terms of your static position.

There’s also the question of what you’re going to do. Your risk is not just what you have.

Maybe if there’s an event risk situation, sure, but there’s also what are you going to do. If your strategy is going to hold for an hour and get out, that’s very different from a strategy that’s holding for six months.

So that should affect your estimation of your risk.

There’s co-variance, so it’s not just what’s going to happen, but how confident are you that the risk that you’ve calculated will continue to be that risk if the correlation of say stocks and bonds goes from -0.4, -0.6 in a shock event to +0.4 in a ‘taper tantrum’ or something like that.

So you have to be aware of the limitations in both variance and covariance.

How much can those change? How volatile and how much can the interaction vary of markets to each other?

There’s systemic risk. What do you do if your lines are severed to your executing broker or the exchanges, in our case, and you can no longer execute your strategy?

So I guess the short answer is there are a lot of definitions of risk and it’s really a very, very broad and intricate topic.

Merely a VAR [calculation] is just the first slice through a very complicated situation.

NKL: The next topic I wanted to jump to is related to risk management. It’s a little bit about drawdowns, and in particular I’m interested in what do you learn from going through a drawdown, because drawdowns are so difficult for investors to go through.

But for you as a manager, what do you take away from going through a drawdown, other than the pain?

RN: We certainly take that away. I think the experiences that we’ve had, and we’ve had some significant drawdowns, every one of them has resulted in a far stronger program than we had before.

It’s like annealing of the strategy – a crucible where it forces you to examine every piece of your organization, every piece of your strategy, and say what do I want to be doing to avoid this? Is it appropriate to be making a change right now?

Is this in this mission of the firm, and is this strategy creep or is this a change that would essentially make a material difference to what our clients expect, and just on, and on?

The fact that we’ve survived 22 years and we’re still here and trading and have a very strong operation, I think we’re doing that with a much stronger platform and a much stronger strategy than strategies that have not really been tested with their particular Achilles heel.

NKL: I wanted to touch upon research, but I actually only have one question that I want to pose, and that is: How do you measure the effectiveness of your research?

RN: It would be easy to say if it doesn’t make money or not, but there’s a lot more to it.

Just like in portfolio construction, your best performers are not the only piece of your portfolio. And to think that the smart pieces of your portfolio are only the things that are going up is, I think, naive.

So the effectiveness of our research would be: Do we feel that our new strategies, let’s say, that we’re using in 2005, perform better from 2005 to 2008 than the strategies that we had in 2000.

In that sense, the mix of the portfolios should continue to improve over time.

Unfortunately there’s a huge confounding factor which is the paths of realized volatility – is it a high vol period, a low vol period; is it a very trending period; a period of a lot of intra-day reversal, so there are tremendous confounding factors.

Unfortunately, therefore, my answer has to be it’s a rather qualitative and subjective assessment of the quality of research.

I don’t think pure quantitative results are the whole answer. Obviously that’s a simple answer to it.

NKL: When you look at being a manager in this space what do you think is the biggest challenge that you and maybe the industry faces?

RN: I think, for us the biggest challenge and one that we really try to face every day is continuing to come up with creative, interesting ways of improving the strategy without merely fitting the data more effectively.

That’s, unfortunately or fortunately the very same challenge that I had when I sat down in front of my computer and put the first opening brace in the code that was then going to turn into our whole technological background, in October of 1992.

It’s still there; it is exactly the same challenge that we had at the beginning and it will never change.

The nice part about it is you are constantly tested by the markets, every day is different.

Every year, every month, every single trading day gives you new challenges in ways to improve.

You can learn new things from the markets every single day. And a very lovely thing, or horrible thing about the trading world is that you really are judged objectively by your results.

There are not too many fields where you can say at the end of the day, this is how good I am; this is how bad I am.

The proof of the pudding is in our track record and what we’ve been able to do for people’s portfolios.

NKL: Based on everything that you’ve learned, if you had the chance to go back and speak to your younger self 20 years ago, 25 years ago, what do you think you would have done differently if anything?

RN: I would say, don’t believe your own bullshit. Be humble. The simpler, the better.
These are things I’ve talked about already in this interview.

I think the tendency of quants, the tendency of quant managers is to constantly try to improve and optimise and go in the direction of additional complexity.

I think with a lot of good science and with good trading strategies going in the opposite direction is actually the optimal path.

To simplify and favor robustness over the highest possible optimized results. So that’s what I would tell my younger self.

NKL: I wanted to ask if you could share a fun fact about yourself, something that people don’t usually know about you, and I will throw in the one thing, that I didn’t know, which is that your office was very close to being part of the movie Wall Street 2, so that doesn’t count?

RN: Well, just a little aside about that movie. We have a big video wall that has about 160 screens on it and that contains the displays of all of our strategies and markets and it just gives us a very good visual reference with what’s going on in the world with our own strategy, but I always tell people that this is the result of growing up without a television.

So that’s one little fun fact that I grew up without a television.

I think maybe a more interesting fun fact is that I think most people don’t know though anyone who’s been to one of my parties does know is that I can pretty much play any song on the piano from memory in any key, and sing it.

I think if the hedge fund thing doesn’t work out I’m going to play piano in a piano bar.

This article is an edited version of a conversation between Niels Kaastrup-Larsen and Roy Niederhoffer which first aired on Top Traders Unplugged, a weekly podcast featuring in-depth interviews with the world’s most successful CTA and hedge fund managers.

Visit www.toptradersunplugged.com.

Take a free trial

Experience HFM Global for yourself. Access exclusive news, research, analysis, events and networking opportunities. Tailored to fit your role, your team and your organisation, allowing you to make more informed decisions.