Fundamental indicators can be useful trading tools. Those that are issued monthly and are relatively standardized in construction and application are the most appealing. We previously examined the relationship between home sales and lumber prices, and it was found that a reasonable correlation exists (see “Exploiting the homes sales/lumber link,” March 2010). Based on that, we could use a linear regression equation to use home sales data to project the next monthly closing price of lumber with confidence.
To continue this search, we can turn to another fundamental indicator: the Institute for Supply Management’s Purchasing Managers’ Index, or ISM PMI. The ISM publishes this index monthly. It is a national survey of 300 purchasing managers at major industrial manufacturing companies. The ISM calculates nine different sub-indexes. These include new orders, production, employment, supplier deliveries, inventories, prices, new export orders, imports and order backlogs. The ISM PMI is composed of the five major indicators: new orders, inventory levels, production, supplier deliveries and employment.
The ISM PMI is centered at 50. More than 50 indicates the purchasing managers are expanding purchasing, less than 50 indicates they are reducing purchasing. Critical levels are accepted to be beneath 43 and above 57. It is widely believed this is a good fundamental indicator for the economy, combining a form of consumer confidence and demand-side economics.
Theoretically, as managers expand purchasing -- that is, increase demand for goods -- the prices of those goods rise. Also, companies expand purchasing when managers feel good about the economy, primarily prior to and during growth phases. Conversely, when managers feel pessimistic, they hold back on orders, thereby shrinking demand and ushering in contraction.
The ISM PMI data start in 1948. There were 740 entries through December 2009. The mean of the data set is 52.557 with a standard deviation just over 7.00. Of the 740 readings, 494 (66.7%) exceeded the benchmark of 50. Interestingly, managers tend to be predominantly optimistic. Of course, the economy has spent most of its time in growth mode since 1948; there were only 11 recessions during that 61-year period.
Does the PMI index help us as traders in any fashion? To answer this question, we can test PMI cycles against the S&P 500 index. The PMI will dictate buys and sells in the stock index and, if the index is useful, we can expect to realize a net profit and have more winners than losers in these hypothetical trades. We will confine testing to the most recent 40 years. Prior to 1970, the S&P 500 was at a low level, volume was almost non-existent and technology was not a factor in the markets. As technology increases, and information dissemination becomes faster, wider and cheaper, purchasing managers should become even better at their business.
To perform the test, we’ll look for troughs and peaks in the PMI (see “PMI highs and lows,” below). In general, we should see troughs less than 43 and peaks more than 57. However, we’ll permit a trough to occur under 48 given the knowledge of the perennial optimism of the purchasing managers.
Here’s the trading rule: When the PMI bottoms on a monthly basis, we buy the S&P 500 index. When it peaks, we sell the long and reverse to a short.
As shown in “Trade report” (below), the simple rule results in a total of 16 closed trades between November 1970 and December 2008. Of the listed 16 trades, 11 were profitable (68.75%) for a cumulative net gain of 1,022.11 S&P points. The average trade shows a profit of 63.88 points. At $500 per S&P point, and allowing $250 for slippage and commissions on each trade, this is a net gain of $507,055 over the 40 years. The open trade is a buy on Dec. 31, 2008, at 903.25, still open as of April 30, 2010 at 1,183.40, for an open profit of 280.15 S&P points. We judge the 68.75% accuracy rate with an average of 64.13 S&P points to be a success.
It should be noted this test is not concerned with drawdown. Clearly, the S&P short in May 2004 was premature, and a trader would have suffered through a significant and lengthy drawdown as the market climbed to its monthly closing peak at 1,549.39 in October 2007. Nevertheless, the position trader would have been ultimately rewarded with a 217.43 net gain by December 2008.
A matter of timing
The success of the basic system raised the question of timing and whether the PMI is a lagging or leading indicator for the S&P 500. Obviously, in 2004, it led the S&P 500 significantly. However, we should test to see if this is predominantly the case.
We’ll impose a three-month waiting period to trade the S&P 500. For example, for the bottom in January 1991, we would wait until the close in April 1991 to cover the existing short and go long. Doing so reduced the gross profit points from 1,026.15 to 946.15 and reduced the win/loss ratio to 10/6 or 62.5%. Moving forward, a six-month time lag reduced the gross profits to 919.06, but returned the win/loss ratio to 11/5. Finally, imposing a full one-year time lag reduced the gross profit to 670.06 S&P points and dropped the win-loss ratio to 50%.
Backing up three months -- that is, the October 1994 peak that caused a sale at the end of July 1994 -- caused a significant reduction in the gross profits, reducing them to 451.96 S&P points. The win/loss ratio was 10/6, or 62.5%.
In these tests, we started at the same S&P level, 87.2 in November 1970, to reduce any variability with the beginning point.
It also may be seen from “Trade report” that all of the losses are on the short side, as is the largest profitable trade. This may be attributed to the bias of the stock market from 1970 to 2000 to move upward almost exclusively. The Internet and the huge increases in trading volume and volatility may have changed all of this permanently, exacerbating the effects of economic factors on both the long and short sides. The 1,074 S&P gross points were realized in the five trades since June 1998, and all five of them were profitable.
These tests indicate that ISM PMI often leads market moves and we can note that anecdotally it would be logical that it would be a leading indicator because it is based on purchasing managers’ economic outlook, which can be flawed but is probably better than the outlook of your average economist.
Lumber is a highly cyclical commodity and heavily influenced by economic factors. As such, it follows that the same relationship with the PMI may be evident with lumber. The lumber price data set begins in January 1993. Because the PMI is long as of January 2001, we immediately begin this trade as a buy in January 1993. “A look at lumber” (below) reflects the results using nearest future closing prices. The open trade also is included.
The method produces a total gross point gain of 766.90 ($766,900), before slippage and commissions. It is 100% accurate over a 17-year period. As may be seen, the swings are increasing over time. Volatility is growing rapidly. Our data for crude oil begins in January 1984. The last signal for the PMI prior thereto is December 1983, a short trade. The results are nowhere near as good, especially when the gigantic price increase of 2008 is remembered as a drawdown on the 2004-08 short trade (see “Slick profits,” below). Research may reveal a time lag that improves these results.
Similarly, a test of cotton prices beginning in 1973 produces eight wins and eight losses for a net gain of 50.02¢, including the open trade profit of 23.35¢. However, four of the gains took place from November 1999 forward. Again, research could be done to determine whether a time shift would be beneficial to the result.
From this analysis, we can conclude that the ISM PMI is a useful tool for predicting the direction of the S&P 500 and lumber. It is less useful for oil and cotton. Other research should be done for any other commodities likely to be closely in tune with the business cycle -- for example, copper.
On average, these trades last three years each. In May 2010, the index rose above 60, approaching the historic turning point Position traders could enter for the long haul. Short-term traders should add this to their arsenal of directional tools and consider options strategies that favor current shorter-term market fluctuations.
Arthur M. Field, Ph.D., is a former commodities broker and co-manager of Fidelity’s Pacific Fund. He wrote “The Magic 8: The Only 8 Indicators You Ever Need to Make Millions.” E-mail him at TheMagicEight@hotmail.com.