Wednesday, May 27, 2009

LUCK VERSUS FORECAST ABILITY

In Luck Versus Forecast Ability: Determinants of Trader Performance in Futures Markets
(1991) utilizing the same data set and statistical procedures he used in 1987 (above),
Michael Hartzmark builds on the former study by testing for two types of trader forecasting
abilities: (1) consistent forecasting of price direction, and (2) “big hit” ability, where
the trader is able to predict direction plus magnitude, holding the largest positions when
the largest returns are anticipated.
For this study, Hartzmark removes the less active large speculators from the data,
which amount to about half the sample. Unlike the earlier study, the remaining largest
of the large speculators show positive returns. Even though a large number of traders
appear to exhibit significantly superior forecasting ability, Hartzmark concludes that the
statistics are highly supportive of the premise that profits were due to luck, not forecasting
ability.

He focuses on two questions that emerge from his inquiry. First, it is not clear why
there is a massive bunching of traders with no ability—although it suggests that many
individuals use very similar trading strategies or information sources. Second, it is not
clear why a subset of large speculative traders should earn significantly positive returns.
If the performance of all traders, whether large or small, is due to luck, one would not
expect the large traders to consistently perform any better than the small traders. Yet
both in this study and in a number of earlier studies cited by Hartzmark, small traders
are the big losers and large traders are the big winners.

KEYNES’S THEORY OF NORMAL BACKWARDATION

It is essential for any commodity market participant, with longer than a day trader’s
horizon, to understand the concepts of contango and backwardation. Fortunately the
concepts are not as difficult to master as their names. All other factors being equal (and
they frequently are), it costs more to buy 5,000 bushels of corn for delivery one year from
now, than for immediate delivery. Why? Whoever owns the corn for the intervening year
must pay loan interest and storage fees and other costs. These are lumped together as
“cost to carry” or “carrying costs.” If in June for example, the cash price for corn is $2.00
per bushel and the cost to carry is $0.05 per month, then normal (or, in the vernacular,
contango) pricing would imply the following forward prices: July $2.05, August $2.10,
September $2.15, and so on.

Keynes put forward a theory, beginning in the 1920s, that contango was abnormal,
and that the normal relationship is “backwardation,” which is, as it sounds, just the opposite
of contango. (These nineteenth-century words are sufficiently archaic to assure
their continued use in economic circles, so get used to them.) Keynes’s reasoning (not
surprising, given his speculative moonlighting) was that if a farmer wished to lock in a
price for corn before it was even planted he could not ask more than the current price,
but must expect less. In Keynes’s view, the ownership of the corn passed to the speculative
buyer at the time of the futures transaction, and because the speculator is also
risk adverse, he would insist on receiving a risk premium commensurate with the risk
of a price decline prior to delivery. This premium was paid in the form of a discount
off the current spot (cash) price, and would be proportionate to the time until delivery.
Keynes’s “normal backwardation” would value corn currently priced at $2.00 in June, as
worth $1.95 for July delivery, $1.90 for August, $1.85 for September, and so on (again
assuming an arbitrary $0.05 per month cost to carry).

From a trading standpoint, backwardation provides a rollover profit for longs, because
they pay a relatively lower price each time they roll the position to a further distant
(cheaper) delivery. As time passes, assuming a flat price trend, the long’s investment still
rises to equalize with the higher spot price. This can be an important part of commodity
returns, particularly for a trader with a very long-term horizon. Prices in contango, on
the other hand, drop during the holding period, guaranteeing a rollover loss (which can
only be offset through successful market forecasting).

Whether Keynes’s view was normal or backward has been argued ever since, without
resolution. You would think the markets would be the final arbiter, but markets are fickle.
Often the difference in prices between different deliveries will reflect the cost to carry.
Often not. The difference may be called “risk premia” or “convenience yield,” depending
on the presenter’s predisposition. If risk premia is used, suspect you are dealing with a
Keynesian; if “convenience yield,” he may be predisposed to search for other factors. In
any case, the Commitments of Traders report has become a bountiful source of data
for inquiry. (Earlier studies into trader profitability relied on small samples of brokerage
house customer statements.)

One other definition might come in handy: “hedging pressure.” This is typically a
Keynesian concept because it implies that hedgers pay a premium for price insurance;
thus, excess hedging demand moves prices. In the original Keynes premise, this would
be downward price pressure, because he assumed that producers would far outnumber
users hedging in futures. This is not always so, making the hedging pressure angle one of
the more interesting.

Though sufficient for our purposes, this is hardly a comprehensive discussion of an
essential element in a trader’s fundamental knowledge base. For example, seasonal crops
are typically found in contango, but can move into backwardation (sometimes called premium)
when nearby supply is restricted. Thus the two terms “demand” and “premium”
bull market both denote backwardation. (Current demand causes nearby prices to move
to a premium over later deliveries.)
Seasonality in production and consumption can add several more steps in the calculation
of convenience yields for seasonal commodities, like heating oil, and perishable
commodities, like livestock. These price series may show a combination of positive and
negative convenience yields.

Then there are the markets where backwardation is the normal state, but not for the
reasons advanced by Keynes. Bonds pay interest, creating a negative cost to carry that
short-sellers need to consider. In the same vein, difference in currency yields affect cost
to carry, moving higher yielding currencies to premium.

CAN SPECULATORS FORECAST PRICES?

In one of the earliest studies attempting to appraise the relative ability of different trader
types in forecasting prices (Can Speculators Forecast Prices? 1957), H. S. Houthakker
assumed that the more accurately price changes were forecasted, the less they would
fluctuate, and the easier it would be for affected parties to adjust.

In 1957, Houthakker conducted a study to understand the role of speculators by
measuring their ability to forecast prices in the commodity futures markets. Using end-ofmonth
positions from the Commitments of Traders annual reports, and average monthly
prices for three agricultural commodities between 1937 and 1952, Houthakker estimatesthe profitability of large hedgers, large speculators, and small traders. To estimate profits
and losses, he assumes that the market positions existing at the end of each month were
opened at the average price during that month, and subsequently closed at the average
price for the month following. Commission and slippage costs are ignored.

In all three commodities, Houthakker calculates that large hedgers were net losers,
while large speculators enjoyed net gains for the period. (Note that I will be using
the term “large speculators,” for these older studies, since this category was, until
the 1980s, composed primarily of individual traders and not commodity funds or
pools.) Houthakker finds that small traders were profitable in cotton, but they were
net losers in both corn and wheat. In addition to being profitable overall, large speculators
were consistently profitable. They made a net profit in cotton for every year
observed. Although in corn and wheat there were a few losing years, none showed a
large loss.

Results were analyzed at a more granular level in order to isolate profits attributable
to forecasting skill, from any resulting from backwardation (Keynes’s Theory).
Houthakker finds that large speculators show definite evidence of forecasting skill, both
in the long and the short run. The experience of the small trader category indicates that
they do quite well when they stick to the long side, where sustained periods of backwardation
assure them profits. He concludes that small traders show no evidence of ability
in forecasting short-run price movements, though.