well be discarded-it is more often that crowd behavior moves the market. IVIedia Indicators

In a delightfiil article, the author Grant Noble argues that the news recognizes events when they are cresting, and most often provides a countertrend trading opportunity. First, the American media should be considered to provide signals in three major time frames:

43r.iutlbble,T]iebeEttr.\diiieindic.*T-fiiemedia," Tectjiical Analysis .-& Cjumodities (-ttober 1 !: "

• Long-term-as given by the large circulation news magazines such as Trme and Newsweek. With the timeliness of a brontosaurus, they profile moves over many years

• Medium-term-represented by Barrons, Forbes, and Business Week, cover a period of about 3 months.

• Sbort-term-held captive by the Wall Street Joumal and the New York Times, which also provide intermediate prediction.

By reading these periodicals you find that the Wall Street Joumal has run headlines on "killing drought," "dusl bowl," and the New York Times on "Drought... imperils crucial wheat crop" just as the wheat price makes its highs for the current move. In another case, Barrons cover article adced "Is the bull leaving you behind?" in August 1987. Its not surprising that the media would highlight events only after they have become a popular concern. In that sense, we might say that there is a high public consensus, a topic addressed in the section "Contrary Opinion." It may be perfectly valid to conshiict a consensus indicates based on the number of square inches of news coverage given to an event in a combination of publications.

E\T:NT TRADING

The largest price moves and the greatest volatility are the result of reactions to unexpected news. These marfcel events pose the greatest risfcs to all traders because they are unpredictable and of such great magnitude that they are out of proportion with normal trading risfc. It may be possible to trade for a number of years without experiencing an adverse price shock., therefore, many traders do not plan properly for these situations. Those readers interested in price shodes and their effects should also refer to Chapter 21 ("Testing").

Not all price shocfcs are of such magnitude that they present an unmanageable problem, it may be that price moves are either the reaction to news or the anticipation of news. This news is most often a release of an economic report by a govemment agency or monetarj authority, but can also be a natural disaster associated with weather or earthquakes, or a political event, such as an outbreak of war, a military coup, or an assassination. The U.S. govemment releases economic data on an announced schedule, many of them at 7:30 A.M. Chicago time, that create regular disruptions to price movements in world economic marfcets.

The frequency and size of these price moves, triggered by unexpeaed news, makes these events a natural candidate for a trading sjstem. The profit opportunity, however, does not lie in taking a position shead of the anticipated market reaction, but in studjing the sjstematic pattems that come after the initial price reaction to the news. Because the news is unexpected, you cannot predict the results nor the extent of the reaction; therefore, taking a maifcel position in advance of a govemment report would have a 50°o chance of success and often very high ride. Studies might show that there is a bias in the direction of the price shock due to the way the monetarj authority plans economic growth and controls inflation; however, the risfcs would remain high. This section win only loofc at positions entered on the close of the event day, the day on which a sharp marfcet reaction occurred.

Maifcet Reactions to Reports

To determine whether there is ample opportunity to profit from the price move that follows an event, it is necessarj to shidy the direction of the maifcet move and the subsequent price changes that occur over the next few dajs. These lagged reactions are the results of marfcet inefficiencies; it is unlifcely that prices could immediately junp to the exact price

1 . 1 1*. entTra win. )

find consistent results; without those values, the best approach is to work backward fran the reaction to infer the accuracy of expectations.

Measuring an Event

There are many economic statistics released each year, as well as special reports in the wall Street Joumal and

that economic principles require, despite the Efficient Maricet Hypothesis. With large price shoclcs there is often an over- or underreaction that is corrected during the next few dajs Sometimes prices junp one direction and immediately begin to go the other way until they have completely discounted the price shock.

Figure 14-1 illusfrates the types of price movements that might follow an upward price shock. Where there is an underreaction the prices move liigher over the next few dajs; when there is an overreaction prices move lower. Pattems are not as orderly as shown in Figure 14-1 because volatility is high and many traders react impulsively, or by financial necessity, to the move. To decide whether a market is a candidate for event trading, you must study the pattem of moves following the reaction to news and decide whether:

1. The size of the average move is enough to generate a profit.

2. The retums are worth the risk of loss associated with these volatile events.

When studjing these events, there may be a direct relationship between the size of the reaction and the tjpe of pattem that follows. For example, a small reaction to news may be followed by a steadj continuation of the direction of the price shock. If unemployment junps by IWb in a month, ttiere should be a reaction by the govemment to stimulate job growth. The same initial reaction would occur if unemployment junped by 1 full percent, but the number would be so unexpectedly large that it may be considered an error in which case it is not clear to what extent the government would respond. The market may overreact to a large shock but underreact to a small one. The only way to discover this is by testing these events.

Fundamental to understanding price shodcs is that the shock is based on the difference between the expectations and the actual reported data, not just the reported dataThe market alwajs discounts what it believes is its best guess at what the report will say., therefore, if bond prices rise in advance of an important unemployment report, we can say that the market expects unemployment to increase. If bond prices have moved up by l/2>o (equivalent yield) in expectation of a very bad report, and the report comes out neuttal, then prices will drop sharply to offset the incorrect anticipation. When studjing market reactions from historic records of economic data, you musl have market expectations to

FIGURE 14-1 Price reaction to unexpeaed news, including delayed response.The upward price junp on the event day (0) may be followed by (a) a continuation move up or (b) a reverse move down over the next 3 dajs.

other influential periodicals. To record the dates of each news item would be impractical for most traders, a simpler approach that might prove effective is to identify the importance of the event by the size of its reaction. This can be done by comparing the volatility of the current day with the average volatility of the recent past using the true range calculation:

, ...... true range,od,y

volatility ratio = --------

average true range „.. .

When the volatility ratio is below 1.0, current volatility is less than the average volatility over the past n days. When the ratio is greater ttian 3.0 we are likely to have identified a day in which a news event caused a price shock The laiger the ratio, the greater the surprise to the market. Note that the true range is used because the shock can occur ovemight, causing an opening gap, or during the trading session, when the United States tjpically releases reports.

Key Govemment Reports

Even when there is public concern over the state of the economy, the marfcet does not react in a similar way to all economic data. Some reports seem to be more important than others, and the marfcet appears to focus on one reporl at a time. The most significant data seem to be unemployment and the Consumer Price Index (CPl). During a period of sustained low unemployment, as seen in 1997, even amodest increase will not cause much concern. Ajunp in the CPL however, will always wam the maifcet to expect a preemptive shifce by the Fed, cutting off potential inflation by nudging rates up slightly. The Trade Balance report was particularly popular in the late 1980s when the deficit with japan seemed to be at the root of the U.S. deficit. Sagging U.S. exports and overconsumption of foreign-made products by Americans loofced as though the U.S. work force could not compete in the world marfcet.

Other reports can also attract the attention of traders, but may be more difficult to interpret. Durable goods, retail sales, budget, and tax legislation all directly affect the economy and prices; however, it is not ahvajs clear how to relate the changes in durable goods orders with price change, or how the latest news on tax law will contribute to the eco. nomic well-being. More important, it is difficult to assess how the govemment will manipulate interest rates in reaction to these data.

Trading the Event Lag

Once you have shidied the way prices move following an event day, or price shocfc, you can establish trading rules. For example, you find that a change in the Consumer Price Index of between 0 and ----.24 has no particular reaction, provided marfcet expectations were similar to the data released. A change between +.2 and +A or -2 and -A should give enough surprise for a noticeable maifcet reaction, provided expectations were low, and data greater than a change of ±.4 should cause a price shock regardless of the anticipation. In addition, we might find that changes over ±4 tended to react in the opposite direction to the normal govemment action because of overanticipation

Based on the results of shidjing these pattems, Warwick established the following trading rules:

1. Buy on the close if the maifcet closes in the upper 20°o of the trading range on an event day (a price shock), if prices have shown a continuation pattern.

2. Hold for a predetermined number of dajs, based on the lag pattem of the marfcet.

3. Use a stop-loss to limit ride.

rn an older study by Arnold Larson applied to the com marfcet, it was found that 8Po of the price changes occurred on the event day, there was a typical price reversal of 8°o over the next 4 days, and a net change in price of 27°o in the original direction over 45 dajs. Although we should not expect those same numbers now, the pattem of a large move, a reaction, and a longer directional move may still be valid.

Results of Event Studies