Risk, Volatility and Leverage: A Talk with Trader Perry Kaufman
By way of introduction, Perry Kaufman has thirty-five years of experience in financial engineering and hedge funds. Beginning as a “rocket scientist” in the Aerospace Industry, where he worked on the navigation and control systems for Gemini, Mr. Kaufman headed systematic trading for Transworld Oil, Ltd (Bermuda) from 1981 through 1992, and was a Principal of Drapeau Advisors, Inc. and later Man-Drapeau Research (Singapore) from 1994 until it was sold to ED&F Man in November 1998. Mr. Kaufman specializes in the development of fully systematic, short-term trading programs in derivatives and equities as well as asset allocation and leverage overlays. He is the author of New Trading Systems and Methods
(Wiley, 2005), A Short Course in Technical Trading
(Wiley, 2003), and other books and articles.
I spoke with Mr. Kaufman by telephone in early October 2008. What follows is an edited transcript of our conversation.
David Penn: What has been the most remarkable characteristic that you’ve seen in the markets over the past year?
Perry Kaufman: I think that volatility is clearly the key to this past period. But also there’ve been very strong trends. So, just by focusing on those two things, this is a technician’s dream market, in particular for trend followers.
Penn: That’s interesting to hear. I remember in one of our earlier conversations you suggested that early on in your trading career, trend following seemed to be both very popular and a very successful way of trading. You added, however, that this approach had become more challenging for traders in recent years.
Kaufman: Trend following falls into a much bigger category. It really can’t be pigeon-holed with a single statement. For example, many years ago when I first started, very, very fast trends worked like a dream because the market was mostly made up of commercial traders. Commercial traders often had the same opinion.
When I first started working with a major oil company, a three-day trend was considered to be highly successful. Now, a three-day trend is ridiculous.
Penn: Yes.
Kaufman: What’s happened over the years is that the more participants that come into the market, the noisier the market gets. I know that sounds counterintuitive, but it’s a different type of noise than type that helps diversification. This noise is more about many people making different decisions at the same time.
This increases the volatility to a point, but mostly it increases the noise. This is what has made it impossible to make money using short-term trends. It now takes too long to identify when the trend starts and when it ends, and that reduces the part of the trend that you can extract as a profit.
So there’s been a development over the past 15 years or so where the ability to make money in trend following means that you need to go further and further out. In other words, much slower trends. Many of the successful traders recently have been macro trend traders, those traders that try to get into trends that are so long that they pick up major economic trends.
This long-term trading window always works – except that, as I’ve just said, because there’s so much more noise in the short term, profits can be much smaller. Added to that is more risk due to volatility, so what was once a brilliant long-term trade is now just a good trade.
Penn: If I remember correctly, your speciality has been not just in developing your own trading systems, but in developing short-term trading systems. Is that correct?
Kaufman: I did that for years. Absolutely yes. But during the 1990s, I was successful in what is called opening-range breakout. With an opening range breakout, you let the market establish an early range in the first half hour, hour, or hour-and-a-half, and then when prices breaks out, you go that direction. That worked really, really well for maybe 15, 20 years, from 1981 to 1999. Now it doesn’t appear to work at all.
So the market changes, and unfortunately we have to change, and the systems that we use, you know, aren’t always there for us.
Penn: In terms of that switch then, since ’99, has that been what you have been most focused on?
Kaufman: There are still ways to make money with short-term breakouts because there is a lot of volatility in the short term. You need volatility to make money on short-term breakouts, but the patterns aren’t quite the same.
But what short term breakouts really capitalize on is trader psychology. The markets still respond to breakouts at particular levels, with clusters of people coming into the markets and doing all the same things. This causes breakouts that certainly can be profitable.
The real problem is in the middle term, the traders that are trying to make money from, say a 10- to a 40-day moving average or a trend calculation period of 10 to 40 days. It’s so erratic. There’s just no consistency. And so I keep far away from that area.
That leaves the very short term, which means under three days, and the very long term, which would be over 60 days. Anything in between is just impossible for a trend follower. So, you have strictly trader behavior and psychology in the short term. On the long end you have economic trends, and then the middle I have not a clue what’s going on! (Laughs)
Penn: What happens when you reach that period, when you have one approach that’s starting to not work as well and need to start a new approach that might work better? How does that transition take place from a systems developer perspective?
Kaufman: Oh, it’s not easy (Laughs). What happened towards the end of ’99 was that I was doing short-term trading in the one- to three-day range, with a lot of the profits came from the German bund and the other interest rate markets, which really declined in volatility to exceptionally low levels.
All of a sudden it was almost impossible even to break even on a trade, let alone carry the rest of the portfolio. So I moved out the timeframe so that instead of holding one to three days, I was holding three to five days. This was because when you hold longer a position longer, you have an opportunity for a bigger move and, of course, a bigger loss.
Now, I still find that there are certain patterns like divergence that are still extremely good and have not been ruined by anything. By divergence, I mean a situation where the momentum in the market is declining as prices are advancing, for example. The prices are going up, but they’re going up slower and slower, and it’s a very good indication that at some point the market is going to roll over.
Another thing traders can do is adapt to the market. If the short-term calculation periods don’t let you make money trend following, then change to mean reversion strategies. There’s so much noise in the short term that mean reversion tends to be a more successful strategy than trend following.
Mean reversion trading, unfortunately, has risk characteristics that are unacceptable to a lot of people and so they shy away from it. But you can adjust your portfolio and in that way you can adjust your risk levels and trade smaller positions to accommodate the risk that goes with mean reversion trading.
Penn: You mentioned risk. This is another topic that you spend a lot of time thinking about and working on.
Kaufman: I adhere to the philosophy that you can’t eliminate risk. I remember years ago when I was doing very well for a client and the client said, “You know, I would be willing to accept lower profits if you would eliminate the risk.” He obviously was unclear of the concept.
You can’t eliminate risk. The case of Long Term Capital Management is still instructive. They managed to engineer out as much risk as possible, and they had a fabulous profile for a long time of just going up. But the fact is you can push the risk around and you can change the shape of it, but the same risk is always there. And when it surfaces, it’s going to be gigantic.
That’s because if you’ve constricted risk all the way up, there’s going to be a situation that is just explosive.
Penn: I’d like to talk just a little bit about some of the different types of risk and some of the different types of approaches traders can take to deal with them.
For example, you’ve talked about hedging strategies, diversification, and deleveraging. Could we briefly touch on those three aspects as they relate to trading?
Kaufman: Sure. Actually, there’s one other approach that just came to mind: If you have a choice, creating a system that is out of the market more often is good.
In other words, if you think about the fact that a price shock is inevitable, that you can’t predict it and that it’s probably going to be one that hurts you – assuming that you are holding a position that many other people are holding because the market always is a shock against the common position…
I did a bit of study that showed that the longer you’re holding a trade, based on how slowly you trade, the more likely the price shock will hurt you.
The short-term traders get in and out. These traders will be hurt by big price shocks 50% of the time. The price shocks will hurt longer-term traders about 70% of the time and that’s because they’re holding the same position as most other traders.
So you can create a system that has an out-of-the-market state, as opposed to a trading system that, for example, is always long or short. If you can make the same amount of money by being out of the market 40% of the time, then you’ve reduced your chance of being hurt by a price shock by 40%.
Believe me, it doesn’t sound as though it’s, you know, rocket science, but it’s the best you can do. If you can’t predict the price shock, then at least if you have less exposure, then you’re better off.
Of course, if you create a system that’s only in the market one day out of 365, it’s not going to be a very impressive system to start with (Laughter), so we’re not talking about extremes. We’re just talking about a rational type of strategy that is just not in the market all the time.
Penn: Interesting, I’ve seen similar results with trading systems that have a minimal amount of exposure. You don’t catch every move. But you catch enough to make money. And you avoid a lot of pain in those price shocks.
Kaufman: Yes.
Let’s discuss hedge strategies next. I think there has been some news on market neutral trading strategies recently that did show that people or hedge funds trading market neutral strategies were not hurt in any way — and may even have been profitable – during this period, this recent period.
One basic hedging strategy, of course, might be where you take two oil service companies or two computer companies that are competitors. You trade one long and one short. That will be neutral in terms of market direction.
So if the market would crash and both stocks crash, you should have offsetting profits and losses – providing you adjusted for volatility, which is essential for any market neutral trading. The risk of your total long positions must equal the risk of your total short positions; otherwise, you’re not really market neutral.
So you would take these opposite positions in related markets, which is called traditional stat arb trading and has proved over the years to be very good. Unfortunately, it’s been exploited a lot lately. You really have to look for specific situations that work. And the opportunities increase when volatility increases, as things go out of line more easily and there’s more money to be made.
In a low-volatility market, these kinds of arbitrage trades between similar companies within sectors or industry groups just fall apart. I mean, they just don’t make enough to offset the cost.
Penn: Interesting.
There were the other two aspects of risk management: diversification and control over leverage…
Kaufman: There are certain basic principles of portfolio construction and naturally one of them has to do with leverage.
Normally you want to establish what your risk level is and try to maintain that level in your portfolio. This means you need to rebalance the portfolio fairly often and you may need to take different sized positions based on the volatility of different markets at the time.
Essentially, in a portfolio, you’re going to want to equalize the risk of all of the things you’re trading because by equalizing the risk, you maximize the diversification. And so you equalize the risk and then you adjust that risk to a certain volatility level. The standard in the industry is somewhere around 12%.
So what you do is you save the final daily changes, percentage changes. You find the standard deviation of that, you figure out what would you have to gear it up or down in order to get it to an annualized volatility of 12%.
Annualizing is simple — you just multiply the standard deviation of the daily changes by the square root of 252 to get the annualized volatility.
Then, to get that to 12% you divide .12 by the annualized standard deviation. You get a factor that is bigger than 1 if your equity stream was less volatile than 12% and smaller than 1 if it was more volatile. Now you go back and multiply each daily return by that factor and you have adjusted to 12% vol.
It’s a very standard way of doing things — market diversification and system diversification — and as a technical trader, I use both of those.
Penn: Does it vary between technical traders — system traders — and other types of trading?
Kaufman: Technical traders have more latitude because they’re able to trade different ways. Discretionary traders tend to have one style. But a systematic trader should be able to create both a trend-following system and, say, a mean-reversion system.
This allows for a mean-reversion system attacking the short end of the calculation period, and the trend-following system on the longer end. You should be able to mix those together. And that kind of diversification is extremely valuable.
So you have the systematic diversification and market diversification. With market diversification you don’t need to be too sophisticated. I mean, you don’t have to understand Markowitz to figure it out (Laughs). You just need to pick a variety of markets that aren’t related.
You just simply group your various markets, into sectors, volatility-adjust all of the markets within the sector, come up with one sector NAV – which is, of course, net asset value – adjust that to your target volatility of, say, 12%.
So now you have multiple sectors. Even though they have different markets in them – one may have 10 markets, one may have 2 markets – you want to try to weight sectors as equally as possible.
Of course, weighting depends on liquidity. But if you could weight them equally, you would then have the most diversification.
So you equally weighed the markets within a sector and create a sector NAV. You have a bunch of these sectors now all with their own NAV, and all adjusted to 12% volatility, and then you equally weigh the sectors into a final portfolio NAV, which would then adjust to whatever your target volatility is.
It’s a very simple method of building and managing a properly diversified portfolio. But you have to keep adjusting to make sure that the portfolio doesn’t get higher than 12 volatility. Within markets, especially the recent ones with extreme volatility, you really have to scale down your trading size, maintain the same risk level.
Penn: What about the other end of the spectrum? When volatility is very low?
Kaufman: It’s not quite clear that when a market goes down to very low volatility that you want to increase at the same rate.
Traditional trend-following systems don’t perform particularly well in low-volatility markets because prices tend to drift. And so you need to find that out for yourself. Mean reversion might or might not work. If the volatility’s too low, then there may not be enough of a move to capture profits at all set costs.
So low volatility is not the greatest environment for anybody to trade, really, not just systematic trading. Because of that, I think it’s a mistake to just keep increasing the size of a position as the volatility goes down. I think at some point you just have to say: you know, the volatility’s too low. I’m just going to hold my position size at the same level.
Penn: Would you say that some of the bigger mistakes that we see in the market happen as much because of over leveraging as due to a failure to make the necessary volatility adjustments in a portfolio as volatility changes?
Kaufman: I think the over leveraging goes back to a Long Term Capital Management type of situation where you extract what you perceive is all the risk by hedging certain types of volatility. You hedge a bunch of things out, and you create what appears to be a systematic return that has extremely low volatility. And because of that low volatility, you leverage it up to get as much return as possible out of it.
Penn: Yes.
Kaufman: And then, of course, it turns out you’re wrong. You haven’t removed this risk. You can make believe you’re removing it, you can remove all the risk you can see in the past, but you can’t remove the inevitable price shock.
Penn: Right.
Kaufman: That’s generally what people do. They engineer a system with relatively low risk and then they believe it (Laughs). For example, if their system winds up with 3% volatility annualized because they’ve managed to make it really, really smooth – which should scare them because in real life there aren’t any systems that perform that way, but if they were getting like a 4 — 5% return with a 3% risk, then they’re going to multiply it all by 4 because now the idea is that we can handle a 12% risk, so the 5% return now becomes 20% return. That’s pretty good (Laughs).
The problem is that your 3% risk originally was fallacious. And so your 3% risk really turned out to be 8% risk. Which means that your 12% risk turns out to be around 30%.
What’s happened is they fooled themselves into thinking that because they could engineer the risk out, they could leverage this thing higher. That’s what Long Term Capital Management did. But Long Term Capital also decided that certain market moves that occurred in 1990 and 1991 weren’t going to happen again.
So they removed that period from the data and were able to engineer arbitrage that looked like low risk because they fiddled with the data. And, of course, something else happened to the Russian ruble in 1998, that effectively was the same thing.
Penn: To hear you explain it, it just sounds crazy.
Kaufman: It’s just unrealistic. One of the ways you know that you’ve done something wrong is when the kurtosis of the system is too high, which means that you just captured all of the profits and none of the losses. There are a number of statistical measurements you could use to decide that you’re really crazy (Laughter).
And kurtosis is really handy. I mean, kurtosis is normally three, and you’d expect a profitable system to have six or seven kurtosis. I’ve seen systems with 30.
Penn: Off the charts.
Kaufman: It’s just not possible. It just means you’ve over-engineered the whole thing. You just can’t have all profits and no losses. But people fool themselves. And the worse thing is they’re willing to put up their money.
Penn: And that of others.
Kaufman: And that of others. Exactly. On the idea that they’ve really engineered out the risk.
You can’t engineer out the risk. You can only change its shape.
Penn: That sounds like an excellent point to leave our readers with. Thank you for a fascinating conversation.
Kaufman: Thank you. You are welcome.