The second case was on February 17 and 18 when the U.S. maricet stayed locked at the limit for 2 dajs. A trader entering the martlet in London on the prior day at 241.91 (3187.50 sterling) could have gotten out on the next day at 247.07 (3255.07) for a loss of 516 points. The corresponding New York trader would have entered at 246.00 and exited at 254.50 for a loss of 810 points

Remarkably, the London coffee prices had moved to 2 52.04 by the time the NewYork martlet was trading again. London prices had gained 1013 points during the time that a locked-limit U.S. market gained only 850 points, in both cases, the volatility of the London market with no limits was much greater than the U.S. equivalent, and the results do not show any disadvantage to trading limits.

FIGURE22-11 Comparing limit moves in U.S. and London coffee martlets, •

MN FEB FtB FEB FbB MAR

vs (May 1977»

.-. London Cuffw ( 1977»

ConvMed IV \ *1 CI = SI 70

The third case of locked-limit moves began on Februarj 24. On the day prior to the first locked-limit day, the U.S market traded actively even after the London Exchange moved up an equivalent of 698 points. There should have been ample time that day or even the prior day, when the U.S. market traded at the limit but closed lower, for any trader to exit an undesirable position. Figure 22-11 shows the larger moves of the London market peaking and starting down to meet the plodding U.S. prices. The market with no limits has overreacted to the bullish news by at least 1,000 points Of course, for traders who didnt get out of NewYork coffee soon enough, the overreaction of the London market was not very gratifjing.

Changing the Rules

Due to the frequency of locked-limit moves during the exceptional periods, and the need to keep changing them as Inflation forced prices higher, most exchanges have changed the way daily trading limits are established. Al one time, an exchange committee was convened to review limits; now, limits automatically expand to meet conditions. The S&P, along with some other financial markets, have circuit breakers that cause the market to halt when specific limits are reached. After a short pause they reopen and can trade to new levels before another circuit breaker is

activated. At an extreme level, however, the S&P will also stop trading, putting it in the same situation as New York coffee versus London coffee. The years have changed but the problems remain the same

The normal policy for agricultural maikets is that, following a trading day in which prices closed at the limit limits expand to 11/2 their initial value. They may remain at this level for 1 or 2 dajs, provided prices continue to close at either limit (higher or lower). Limits will then expand to twice the initial range; if a locked-limit move persists, limits are entirely removed for 1 day. After this, or after any day that prices do not close at the limit, the attem begins again with the initial limit value. Occasionally, the initial value may be altered by the exchange.

GOING TO EXTREMES

In 1974, public awareness of inflation caused an overwhelming interest in all forms of hedging, with large numbers of naive investors purchasing silver and gold as currency protection. Many sophisticated investors turned to the futures markets, which offered leverage on their purchases; some learned how to use the maikets themselves and others relied on advice.

One investment sjstem that was sold at this time was more of a leveraged substitute for the purchase of bouillon than a trading strategj. At the time it was published, it had ahvajs worked-the sponsor of the sjstem stood behind it with his reputation. The rules of the sjstem were:

1. Trade silver futures because of their intrinsic value, historic performance, potential, and fundamental demand with short supply.

2. Use the futures contract between 3 and i months from delivery to combine the advantages of liquidity and duration.

3. Always buy, never sell, because it is ahvajs successful.

4. Buy whenever you like. Although any sophisticated method can be used, it wont matter in the long run-any guess will do just as well. Follow the same method for adding to positions or reentering after closing out a trade.

5. Close out the position when there is a profit, not before.

6. Meet all maigin calls-dont let the maiket beat you

7. Invest $5,000 per confract (5,000 froy ounces). This will allow a $I-per-ounce adverse move (silver was al $4.50 per ounce).

8. Whenever you need reinforcement, reread the reasoning behind this sjstem.

9. Do not let anyone or anything interfere with following this sjstem.

How did the investors do? It depended on when they started. Entering in the first half of 1975, shortly after the sjstem was released, showed individual contract retums ranging from profits of 5 10 to losses of 180 per ounce. A drop in silver prices then produced losses of up to $1.02 per ounce for the next 9 months (slightly over 100° o), followed by a 2-month rally, and losses again. This sjstem would have worked from 1979 to February 1980 and then lost for the nexl 15 years. An investor following these rules, entering at a chance place would have had more than a 10: I likelihood of losing. Any profit made would have been given bad; if the investor persisted in following the method.

In trying to understand how a sjstem such as this can have a chance of reaching the public, remember the time in which it was infroduced. With prices increasing drastically, food shortages, and publicity over the devaluing U.S, dollar, the rationale of the sjstem seemed justifiable. There was serious talk about leveraged inflation funds using futures. These sjstems fill an immediate need, without regard to long-term consequences. There IS no doubt that, if high inflation retums, sjstems just like this one will reappear. Given the bull maiket in stocks for the past 20 years, it is surprising that more of these sjstems are not in evidence. With only small changes in the nine rules for trading silver, we could create a scenario that would be very believable to many people looking to enter the stock market.

SIMILARITY OF SYSTEMS

A primarj concern of both the government regulatory agency (CFTC) and individual traders is the similarity of computer-based sjstems that are used to manage large positions. If the propagation of technical trading sjstems has similar foundation and testing methods, it is possible that many of them will produce buy and sell signals on nearly the same dsy. That is, these sjstems may appear different in their rules and parameters, but they could be essentially quite similar-too many people looking at the same markets and using the same tools.

First, consider the exfreme case in which two sjstems with the same frend-following philosophy have the same profit over the same period of time. Will the signals occur on the same daj? Are they on the same side of the markel most of the time? Figure 22-12 shows one combination that has a low correlation given this situation.

Sjstem A is a slow frend-following method, which avoids the choppy markets but nets only a small profit. Sjstem jumps in and out, capturing profits quickly but with frequent losses. They both rehim the same profits; Sjstem is on the same side of the market as A about 50°o of the time because the overall frend was up; therefore, il cannot he negatively correlated, but it could have a low positive correlation. If sjstem were not as fast as in this illustration, the positive correlation would be greater. As the time periods used for calculation become closer, the correlation increases.

if computer-based, frend-following sjstems with different calculation periods are not highly correlated, then il is possible to reduce portfolio risk by trading a variety of different techniques and time periods; however, the reality of the issue may be different from the theory. Lukac, Brorsen, and Irvin performed such a shidy. They compared 12 popular frading techniques (mostly frendfollowing) over 12 varied futures markets for the years 1978 to 1984. Each sjstem was optimized, using 3 years of data, and the best parameters used to frade the next year. The sjstems selected were

1. Cbannel sjstems

a. Closing price channel (CHL)

b. price channel ( )

c. L-S-0 price channel (LSO)

Louis P Lukac, G. Wade Brorsen, and Scott H. Irwin, Similatity of Computer Guided Tecimical Trading Sjstems (CSFM-124, Working Paper Series, Columbia Futures Center, Columbia University Business School, New York March 1986). Also by the same authors, "Do similar signals from trading sjstems move prices?" Futures (November 1987).

FIGURE 22-12 Two frend-following sjstems with low correlation.

2. Momentum/oscillators

a. Directional indicates (DRI) b. Directional movement (DRM) c. Range quotient (RNQ) d. Reference deviation (REF)

3. Moving averages