Risk Management Lessons in Leveraged Commodity Futures Trading
Risk Management Lessons in
Practical issues involved in applying a disciplined risk management methodology to futures trading.
This article will show how to apply methodologies derived from both conventional asset
management and hedge fund management to futures trading. It will also discuss some of the
risk management issues, which are unique to leveraged futures trading. By HILARY TILL.
The Most Important Element of an Advisors (CTAs) have had losses in from a programme’s benchmark.
Investment Process excess of 40%, which have been • Make assumptions about the
THE KEY TO a successful invest- acceptable to their clients since these expected returns, volatility, and
ment programme is not in discovering investment programmes sometimes correlation of the active bets.
proprietary investment strategies – a produce 100%+ annual returns. • Construct the client’s portfolio so
diligent literature search will turn up a Investors know upfront the sort of that the client’s return and risk tar-
great number of strategies. Instead, swings in profits and losses to expect gets will be achieved if one’s sta-
the most important element of an from such managers. tistical assumptions are correct.
investment process is how one imple- If an alternative investment pro- • Continually monitor the portfolio’s
ments the programme’s portfolio con- gramme were advertised as an equity actual return and risk performance
struction and risk management, so diversifier, then clients would expect for adherence to the established
that one can have both smooth per- that the programme should not do too targets.
formance and stay in business during poorly in the face of a large equity
dramatic market moves. decline. Risk management is designed into
The parameters of a programme’s investment process. The conventional
Risk Management Policies are a risk management policy should direct- asset manager approach to risk man-
Product Design Issue ly flow from the return, risk, and cor- agement is a useful first step in
In derivatives trading, one has a lot relation expectations of the pro- designing a risk management pro-
of flexibility in designing an invest- gramme’s client base. When attempt- gramme for leveraged futures trading.
One still needs to add several layers
> Risk management may be the most important of risk management to this approach
because of the unique statistical prop-
element of an investment process < erties of commodity futures contracts
and because of the different way
futures products are marketed.
ment programme. Futures trading ing to adhere to these top-level param- A futures product typically does not
requires a relatively small amount of eters, the actual implementation of a have a benchmark, so the convention-
margin. Trade sizing is mainly a matter programme’s risk management policy al asset manager approach of translat-
of how much risk one wants to will rely heavily on the particular ing a client’s guidelines into risk and
assume. An investor is not very con- assumptions about the statistical return targets with respect to an index
strained by the amount of initial capi- properties of futures prices. does not directly apply. Instead, one
tal committed to trading. needs to determine what the accept-
With the use of options, one can also Standard Risk Management able total-return-to-total-risk trade-off
be very particular about the risks that Methodology from Conventional is for a client. Given the ability to
the investor wishes to hedge away by Asset Managers is a Useful leverage, a number of CTAs offer 1-
paying option premia. Starting Point times, 2-times, and 3-times versions of
We believe that the leverage level The way that risk management is the same programme. In other words,
chosen for a programme and which applied at conventional asset man- a client can directly choose the lever-
risks are hedged are product design agers is typically as follows: age level for their investment based
issues. One needs to determine How • Translate the client’s guidelines on their ability to tolerate losses of a
will the programme be marketed, and into return and risk targets with given magnitude.
what will the client’s expectations be? respect to an index or benchmark. The second step in a conventional
A number of top Commodity Trading • Determine the active bets away asset manager approach to risk man-
September 2002 * Commodities Now 1
agement consists of making assump- this approach alone is obviously inad- able to get a complete idea of the
tions about expected returns, risks, equate for a commodity portfolio, range of possible outcomes.
and correlations of active bets. It is at which consists of instruments that If one is relying on historical data to
this point that the unique behaviour of have a tendency towards extreme pos- find pockets of predictability in the
commodity prices creates extra steps itive skewness. futures markets, then examining
in a risk management programme. While this measure is useful, it has to worst-case outcomes can also serve
be used jointly with other measures another purpose. If the loss on a par-
Risk Management Rules Flow from and actions. The measure is useful ticular commodity futures trade
an Understanding of Price since one wants to ensure that under exceeds the historical worst case, this
Behaviour normal conditions, a commodity posi- can be an indication of a new regime
Research from the 1970s showed tion has not been sized too large that that is not reflected in the data. This
that diversified portfolios of equities one cannot sustain the random fluctu- would trigger an exit from a systemat-
have returns that appear to be sym- ations in profits and losses that would ic trade since one no longer has a han-
metrically distributed. It is a different be expected to occur, even without a dle on the worst-case scenario.
matter for commodity prices. dramatic event occurring.
Deaton and Laroque  note the Sizing a trade based on its volatility Deep Out-of-the-Money Options
following about the empirical behav- is especially important the longer the In a systematic investment pro-
iour of the prices of a number of com- frequency of predictability is. For gramme based on historical data, one
modities: example, if a trade’s predictability is at can make determinations about the
• “Commodity prices are extremely quarterly intervals, the trade has to be expected return of an investment.
volatile.” sized to withstand the daily fluctua- One result is that an investor can
• There exist “rare but violent explo- tions in profits and losses. decide to give up a small fraction of
sions in prices.” this expected return in order to hedge
• In normal times, there is a “high Scenario Testing against catastrophic risk. An investor
degree of price autocorrelation.” The following recommendation can can do so with deep out-of-the-money
• “In spite of volatility, prices tend to only be implemented if a programme’s options.
revert to their mean or to a …
• “There is substantial positive skew- > We would recommend examining the worst
ness” in the price distributions; and,
There is “substantial kurtosis with
performance ... over the entire time horizon <
tails much thicker than those of the
normal distribution.” investment process is systematic and This choice is especially advisable
based on historical data. for commodity futures positions that
Commodity prices tend to exhibit Using long-term data, one can require physical delivery at maturity.
positive skewness for the following directly examine the worst perform- This means that contracts can be peri-
reason. During times of ample sup- ance of a commodity trade under sim- odically squeezed to quite unpre-
plies, there are two variables that can ilar circumstances in the past. In prac- dictably high levels.
adjust to equilibrate supply and tice, we have found that such a meas-
demand: more inventories can be held ure will sometimes be larger than a Diversification & Concentration
and the price can decrease. But, if Value-at-Risk measure based on Risk
there are inadequate inventories, only recent volatility. A commodity investment manager
the price can respond to equilibrate We would recommend examining the can potentially set up dampened risk
supply and demand, given that in the worst performance of a futures trade portfolios of commodity investments,
short run, new supplies of physical over the entire time horizon of the which are very nearly uncorrelated
commodities cannot be instantly trade rather than looking at what its with each other. (see Figure 1.).
mined, grown, and/or drilled. worst performance was over a period Now for all types of leveraged invest-
of say, three days. We believe that ing, a key risk management concern is
Value at Risk markets are ‘learning systems’. During inadvertent concentration risk.
If a portfolio of instruments is nor- a price shock, if a similar event In our own work, we have found that
mally distributed, one can come up occurred in the past, market partici- seemingly unrelated commodity mar-
with the 95% confidence interval for pants know what the magnitude of the kets can become temporarily highly
the portfolio’s change in monthly price move was during the past event. correlated. This becomes problematic
value by multiplying the portfolio’s So an entire, dramatic price move may if a commodity manager is designing
recent monthly volatility by two (or occur in a shortened time-frame com- their portfolio so that only a certain
1.96, to be exact.) The portfolio’s pared to the past. amount of risk is allocated per strate-
volatility is calculated from the recent In practice, if a market only has limit- gy. The portfolio manager may be
volatilities and correlations of the ed historical data, it would be prudent inadvertently doubling up on risk if
portfolio’s instruments. This is the to scale down the size of a position in two strategies are unexpectedly
standard Value-at-Risk approach. Now, such a market since one may not be correlated.
2 Commodities Now * September 2002
the form of either a Eurodollar futures
Fig 1. Portfolio Volatility vs Number of Strategies
contract overlay or purchases of out-
14 14 of-the-money fixed-income calls. This
recommendation is similar to that of
13 13 the fund-of-funds manager noted
above, whose portfolios were at risk to
11 11 Obviously one would prefer to layer
on natural hedges, which themselves
10 10 have positive expected value. We have
found that this is sometimes possible
in a diversified futures programme.
8 8 For example, in the fall there tends to
be a number of statistically significant
7 7 commodity trades that have a long
bias. Also, at the same time there are
1 2 3 4 5 6 7 a number of statistically significant
Number of Strategies long fixed income trades. By carefully
Source: Till, Hilary, “Passive Strategies in the Commodity Futures Markets,” Derivatives Quarterly, Fall 2000, p. 54. combining these trades, the fixed
income trades operate as a natural
Understanding the Fundamental include an interest rate overlay in their hedge to the event risk taken on with
Drivers of a Strategy fund. The interest-rate overlay con- the long commodity trades.
The antidote for this problem is two- sists of going long Eurodollar futures, The hedge fund world also provides
fold. One is to understand what the which do well when short-term inter- other risk management solutions that
key factors are which drive a strate- est rates are cut. The Federal Reserve are applicable to futures investments.
gy’s performance, and the other is to Board’s response to liquidity shocks One concern for a fund-of-funds is
use short-term recent data in calculat- during the last 15 years has been to that its group of funds is inadvertently
ing correlations. If two trades have cut short-term interest rates, so a exposed to some event risk like an
common drivers, then it can be
assumed that their respective per-
formances will be similar. Recent data > ... how one designs and carries out a risk
can frequently capture the time-vary- management policy is key to an investment
ing nature of correlations that long-
term data average out. programme’s viability <
Extraordinary Stress Testing Eurodollar overlay could plausibly off- emerging markets shock. This issue is
As discussed above, risk manage- set losses in portfolios consisting of compounded by the fact that a hedge
ment policies flow from product design arbitrage strategies. fund investor is not allowed to see
decisions. Futures products are typical- This type of macro hedging is very what a hedge fund is investing in
ly marketed as equity investment applicable to commodity futures because this is considered proprietary
diversifiers. Therefore, one job of risk investments as well. A number of com- information by a hedge fund.
management is to attempt to ensure modity futures strategies have a long One risk management software
that a futures investment will not be commodity bias since they rely on tak- provider, Measurisk, solves this prob-
correlated to the equity market during ing on inventory risk that commercial lem by confidentially collecting hedge
periods of dramatic equity losses. participants wish to lay off. One conse- fund portfolios and directly determin-
This extra risk management step is quence is that these strategies are at ing their sensitivity to past financial
unique to alternative investments, risk to sharp shocks to business confi- shocks. For example, if one held a par-
again, because of the way they are dence. And during sharp shocks to ticular fund-of-funds portfolio during
marketed. For example, funds of hedge business confidence as occurred in the October 1987, one could see how that
funds are also marketed as equity aftermath of September 11th 2001, the portfolio would have performed during
diversifiers, so this is also a particular stock market performs quite poorly. the stock market crash. This scenario
area of concern for such funds. Since As noted before, the Greenspan Fed test gives an indication of sensitivity
fund-of-funds typically include a lot of has responded to financial shocks by to such a crash.
arbitrage strategies, which in turn rely cutting interest rates, which has For a commodity futures portfolio,
on the ability to leverage, fund-of- resulted in the stock market stabilis- we believe that it is prudent to exam-
funds are at risk to liquidity shocks. ing. As long as this type of policy con- ine how the portfolio would have per-
And the equity markets typically also tinues, one way to hedge a portfolio formed during various well-defined
do poorly during liquidity shocks. that has exposure to shocks to busi- stock market declines, given that such
One solution advanced by a promi- ness confidence is to include a fixed investments are marketed as equity
nent fund-of-fund manager is to income hedge. The hedge could take portfolio diversifiers. Also, various
September 2002 * Commodities Now 3
crises have shown that the only thing
Worst-Case Loss Worst-Case Loss
that goes up during such times is
Strategy Value-At-Risk During Normal Times During Eventful Period
Deferred Reverse Soybean Crush Spread 2.78% -1.09% -1.42%
If a portfolio shows sensitivity to cer-
Long Deferred Natural Gas Outright 0.66% -0.18% -0.39%
tain extreme events when the stock
Short Deferred Wheat Spread 0.56% -0.80% -0.19%
market has declined, this does not nec-
Long Deferred Gasoline Outright 2.16% -0.94% -0.95%
essarily mean that the portfolio should
Long Deferred Gasoline vs. Heating Oil Spread 2.15% -1.04% -2.22%
be sized or constructed differently. It
Long Deferred Hog Spread 0.90% -1.21% -0.65%
may mean that a macro portfolio
hedge would be in order such as pur-
Portfolio 3.01% -2.05% -2.90%
chasing out-of-the-money Eurodollar
call options, as noted above. Incremental Contribution to Incremental Contribution to
Strategy Portfolio Value-At-Risk* Worst-Case Portfolio Event Risk*
Useful Risk Management Reports in Deferred Reverse Soybean Crush Spread 0.08% -0.24%
Futures Trading Long Deferred Natural Gas Outright 0.17% 0.19%
On a per-strategy basis, it is useful to Short Deferred Wheat Spread 0.04% 0.02%
examine each strategy: Long Deferred Gasoline Outright 0.33% 0.81%
Long Deferred Gasoline vs. Heating Oil Spread 0.93% 2.04%
• Value-at-Risk based on recent
volatilities and correlations. Long Deferred Hog Spread 0.07% -0.19%
• Worst-case loss during normal times. * A positive contribution means that the strategy adds to risk: a negative contributions means the strategy reduces risk.
• Worst-case loss during well-defined Source: Premia Capital Management, LLC
• Incremental contribution to portfolio in history since one is then in unchart- In designing a risk management
Value-at-Risk. ed territory. framework, a leveraged futures trader
• Incremental contribution to ‘Worst- The chart above gives examples of a can use as a starting point the frame-
Case Portfolio Event Risk’. futures portfolio with the recommend- work provided by conventional asset
ed measures displayed. Note for exam- managers and also by fund or hedge
The latter two measures give an indi- ple, the properties of the soybean fund managers.
cation if the strategy is a risk reducer crush spread. It is a portfolio event-risk We conclude by noting that how one
or risk enhancer. reducer, but it also adds to the volatili- designs and carries out a risk manage-
On a portfolio-wide basis, it is useful ty of the portfolio. ment policy is key to an investment
to examine the portfolio’s: An incremental-contribution-to-risk programme’s viability, especially in
• Value-at-Risk based on recent measure based solely on recent volatil- leveraged commodity futures trading s
volatilities and correlations. ities and correlations does not give Bibliography
• Worst-case loss during normal times. complete enough information about 1. Deaton, Angus and Guy LaRoque, “On the
• Worst-case loss during well-defined whether a trade is a risk reducer or risk Behaviour of Commodity Prices.” Review of
eventful periods. enhancer. Economic Studies (1992) 59, pp. 1-23.
2. Till, Hilary, “Passive Strategies in the
Commodity Futures Markets."’Derivatives
Each measure should be compared to Conclusion Quarterly, Fall 2000, pp. 49-54.
some limit, which has been determined Our view is that there are a number
based on the design of the futures prod- of commodity derivatives strategies,
uct. So for example, if clients expect the which earn returns due to assuming
programme to lose no more than say risk positions in a risk-adverse financial
7% from peak-to-trough, then the three world. The returns are not necessarily
portfolio measures should be con- due to inefficiencies in the market-
strained to not exceed 7%. If the prod- place.
uct should not perform too poorly dur- There is a very important active com-
ing financial shocks, then the worst-case ponent to an investment programme
loss during well-defined eventful periods that earns a return due to bearing risk.
should be constrained to a relatively It is the investment programme’s risk
small number. If that worst-case loss management methodology and policy.
exceeds the limit, then one can devise An investment manager must decide
HILARY TILL is Principal with
macro portfolio hedges accordingly. how much to leverage the strategy and
Premia Capital Management, LLC
Now obviously the danger with these whether to give up any returns by
53 W. Jackson Blvd.
recommended approaches is that one hedging out some strategy’s extreme
is relying on historical data for guid- risks. That investment manager must
Chicago, IL 60604, USA.
ance since completely unprecedented also continually monitor the risk expo-
T: + 1 312 583 1137
events do happen. That is why we rec- sures in his or her portfolio and make
F: + 1 312 583 1139
ommend exiting any futures trades in sure that those exposures adhere to
which the losses exceed those known pre-defined limits.
4 Commodities Now * September 2002