- 1997: Volume 6, No. 6 Technical Tips of the Trade

Veteran traders will tell you that the market is full of idiosyncrasies which can be used to identify successful trading strategies. Bruce Kamich, Walter Burke, Roy Justice and Vincent Jay of MCM MoneyWatch, a real-time global market analysis firm, have collected 15 "tips" which they have formulated from their observations of the bond market.

1. The first tip is "a close above the high of the low day" as a short-term signal to cover shorts or to go long. The reverse also works at tops-"a close below the low of the high day" as an effective signal to sell out longs or to go short. The most recent buy signal was August 28th when Dec. Bonds closed at 113-09, above the 112-12 high of August 26th-the low day. The last sell signal was August 4th when Dec Bonds closed at 114-09, below the 114-12 low of August 1st which was the high day.
2. Outside days and higher closes are pretty good buy signals at bottoms. The big outside day and higher close on August 13th ended the decline from the beginning of the month. Outside days and lower closes tend not be as good a tool to identify bond market tops.
3. Bond futures often make major "runs" of 10-points (though the long-term charts going back to 1977 also show some 12- and 14-pt. runs), often followed by corrections or reversals. For instance Dec. Bonds rallied from 105-25 the week of April 14th to 116-20 the week of July 28th.
4. We have a "rule of thumb" for double tops and bottoms. Because chart patterns are rarely "textbook", we have learned to look for the second top to occur anywhere from one to two percent below the first high to one to two percent above the first high (for a double bottom reverse the method). We believe this occurs from traders not waiting to sell the last tick or from early sellers getting stopped out if the market makes a slight new high before starting its decline.
5. Though there are some 50 or so candlestick patterns that can provide timely and accurate signals, we wanted to focus on the patterns that seem to occur most frequently and across different time frames in bond futures. We have found these to be hammers and hanging men, bullish and bearish engulfing patterns, dark cloud covers and piercing patterns, and the ever faithful doji. A hammer can be seen on the daily chart for 8/26 marking a low. A doji on 8/19 is the end of the August bounce. 8/12-13 is a piercing pattern while 8/28-29 and 9/2-3 are dark cloud covers. A weekly candle chart of Dec. Bonds shows a bullish engulfing pattern for 4/21-28 which would have gotten you long very close to April lows. Intra-day candle charts starting with 60 minutes and working down to five minutes can pinpoint turning points. Textbook examples of these patterns can be seen on our Web Site www.mcmwatch.com under MoneyWatch Market Reports and Charts.
6. Triangles used as a measuring and timing tool. Often a market will spend days, weeks or even longer trading within a triangle pattern. While frustrating to trade, they do carry an important message of a price and time objectives for the move out of the triangle. The price objective is simply the widest part of the triangle applied to the ensuing breakout. An example can be seen on the Dec. Bond chart. The widest part of the triangle is about 2-1/2 points, if this is applied to the breakout point of 112-12 an objective of 110 is possible. The approximate time that move from the breakout from the triangle will occur is found by determining the point in space where the apex would intersect. The apex for the current triangle is in late September Therefore we look for 110 to be achieved by that time. One related rule or guideline that we like to see triangle patterns adhere to is that the breakout occur somewhere between two-thirds to three-fourths of the way through the pattern. We have found at this point that the triangle is coiled enough to have the energy to achieve the measurement objective, and thus many triangles do not reach the apex (coils and wedges more often reach the apex). Intra-day triangles work, too.
7. Put/Call Ratio-a measurement of market sentiment. The P/C ratio is calculated on a daily basis by taking the total volume of put options divided by the total volume of call options. The P/C ratio is used as a contrary indicator, i.e. a high P/C ratio signals excessive pessimism among traders and investors. The best signals generated by the P/C ratio is not by day-to-day fluctuations, but when the ratio registers extreme readings. To smooth out the peaks and troughs, we like to use a 10-day moving average of the daily P/C ratio. Reading above 1.2 on the 10-dma indicates a strong bearish market sentiment, whereas a ratio below .8 indicates a strong bullish sentiment. Some believe the P/C ratio is a better measure of sentiment because it measures what investors are actually doing with their money, unlike the numerous sentiment surveys performed by a variety of institutions who survey market analysts. For example, the P/C ratio for T-bond options produced readings of over 2.0 on April 1 and 2 (10 dma - 1.3), just ahead of the April bond market low. On July 30, the P/C ratio reached a low of .35 (10 dma - .66), a two-day lead before the August 1 peak.
8. Reactions in strong markets run one to three days against the trend; buy the middle of the second day. We can't remember where this trading tip came from but you can often see it on daily bar charts. In July of this year there were two good examples. On July 17, 18 and 21, the bond market corrected lower before resuming the rally on July 22. Then on July 23, we hit another high and pulled back the next two days before moving up again. Our rule of thumb is to watch the market closely in the middle of the second day (up or down). More often than not the market may be oversold or overbought on an intra-day basis, suggesting the correction may have run its course. Intra-day candle reversals may be seen. This also works in strong downtrends. Similar to this is (11) moving average envelopes. Gerald Appel probably did the most work on this tool, but this tip comes from the financial futures department of EF Hutton. Traders can take profits on bond futures when they trade at the upper or lower 5-day moving average envelope, see the bottom of page (7935), and then rebuy or reshort when the market gets to within one percent of the 5-day average. Normally the upper and lower bands are two to 2.5% around the five-day moving average.
9. Time cycles. Martin Pring defines time cycles as "an observable price pattern or movement that occurs with some degree of regularity in a specific time interval. A market or indicator that is observed to have a relatively consistent price low at six-week intervals is said to have a six-week cycle." Many technicians hark on the importance of price movements and patterns, but fail to take the dimension of time into their analysis. Time cycles can help technicians confirm price patterns and aid in predicting the direction of the subsequent breakout. We have found some accurate and reliable long- and short-term time cycles for bond futures, consisting of the 40-week, 20-week, 7-week, and the 11-day cycles. (The cycle periods we like to use on bonds come from research by Jim E. Tillman, CMT of Cycletrend, Inc.) Our long term 40-week cycle last bottomed on April 7 (the week of the bond market low), and is due to bottom again on January 12, 1998. The 20-week cycle, which also bottomed on April 7, successfully marked the recent low the week of August 25. In addition, we have found an accurate 9-day cycle for short-term Eurodollar traders. For daily commentary on the 9-day cycle see page (7928).
10. Flags. According to the late technical analysts Robert Edwards and John Magee, flag patterns normally form within three to four days, rarely more. Last year we decided to refine the time parameters specifically for bond futures, by analyzing daily bond charts dating back several years. We found that the average bond futures flag pattern develops within four to five days after the "pole" or sharp move up or down. Knowing that bull and bear flags in bonds take four to five days on average can be useful to a great many traders. It can give you an early warning signal on when to anticipate another breakout to the upside on bull flags or the downside on bear flags. Let's look at the price action this week. Tuesday we rallied sharply to form what could be a pole to a potential bull flag. Assuming we trade mostly sideways through the rest of the week, we should be looking for a breakout Monday or Tuesday. A breakout sooner would change the pattern to something else and tell us this advance was stronger or weaker than our first analysis.
11. The NOB Spread. We have often used the NOB spread (notes over bonds) as a coincident or leading indicator for the bond market. If one assumes that spreaders can take a view on the markets earlier than out-right positions because they are not exposed short or long, then it makes sense that if bonds rally faster than note futures this is telling us something bullish about bonds and if bonds fall faster it would imply traders are becoming bearish or defensive on bonds. The NOB spread was showing a bullish divergence on July 31 and a warning sign for bonds even though the outright bond chart did not show a divergence. Looking back in history to the 1993-94 bond market slide, we remember when in the summer of 1994 a number of visible technicians proclaimed that bonds had broken their 10-year uptrend line-if you followed the NOB spread you did not see confirmation of this and if you kept up a long-term point and figure chart the 1994 bear market only tested the major multi-year uptrend line on the p&f chart.
12. More timing ideas. We have noted in the past that when bonds rally, prices tend to move in waves or runs of 10-plus points. In that analysis, we neglected the value of time. Our latest research shows that a price move of 10-plus points tends to last an average of 17-1/2 weeks. The research was done on all sharp rallies of over 10 points since the late 1980s. The average run during this period was 12-1/2 points, with 16-4/32 (March to May 1995) being the largest, and 10-pts (November 1994 to February 1995) the smallest. In respect to time, the longest advance was 28 weeks (June 1991 through January 1992), and the shortest advance was 11 weeks (March to May 1995, and September to December 1990). This knowledge of potential time constraints can help traders in anticipating market peaks. The latest advance from the April low to the August peak was a typical advance in respect to both price and time. Within 17 weeks, prices tacked on 11 points, both "normal" when compared to the averages in our study.
13. Bullish consensus numbers. There are three popular series of market sentiment numbers that traders follow. These are compiled by Market Vane (Bullish Consensus), Consensus Inc. and MBH Commodities (DSI). While all series are reliable measures of market sentiment, we prefer the Market Vane mainly because we possess historical data on this series. We use the weekly Market Vane number which is available only to subscribers on Tuesday evenings and to the public in Barron's on Saturday. We then take this weekly series and smooth it by calculating a four week moving average. We have found that historically bonds are oversold and at a low risk point for longs when the 4-week ma is in the mid to low-thirties. Conversely, bonds are overbought and at a high risk point for longs when the 4-week ma is in the upper fifties or higher. Recent history shows that in early April the 4-week ma achieved an oversold reading of 31 which was nearly coincident with the bond market bottom. The 4-week ma then increased to 66 in late July, just as bonds were establishing their peak. Currently, the 4-wk ma has eased to 47 which is considered neutral.
14. Neural Network Forecasted Trend Changes. Available exclusively to MoneyWatch clients via Parallax Financial Research, our neural network program uses chaos theory to accurately forecast when a market could change direction. By definition from the parallax website (www.pfr.com), a turn marks a change in price behavior or direction. In other words, these are used as timing points to mark price highs, lows, or points where prices could turn flat after a period of rising or falling. Therefore, turns should not be used in isolation, but in conjunction with other technical indicators to determine if a market is overbought, oversold or due for a consolidation. The turn works in a variety of time frames, with weekly, daily, and hourly the most significant. Statistically, turn dates have an 80 percent probability of occurring within plus or minus one bar of the forecasted date (week, day, hour). Short-term traders can see forecasted hourly and daily turns starting on page (7095), whereas long-term traders might be interested in visiting www.pfr.com/ta/ta-w97.html for large scale weekly turns.
15. The best bar chart patterns. We have been looking at charts of bond futures since approximately 1980, commodities since 1973 and equities going back into the late 60s. Through the years and the various markets, we have noticed that each market will have its own personality. Bonds don't trade like pork bellies and bellies don't trade like IBM. For bond futures we have mostly seen double bottoms and tops. 1982-83 was a great double bottom as was 1984. The charts show a great double top in late 1989 that started the 1990 bear which preceded the 1990-93 bull run. This rally started with a double bottom and ended with a double top. Spike tops or v-bottoms can be seen frequently in the commodity pits, but the 1986 bond top and the 1987 bottom are good examples in bonds. The bottoming process in 1996 could be a triple bottom but we have found this to be as rare as diamond tops. Rising wedges have been seen more than falling wedges. Many of the 10-point runs are often composed of measured moves-two legs separated by a correction. Triangles are often spotted on the charts and are good for price objectives and timing.

Bruce Kamich, CMT, Walter Burke, CMT, Roy Justice and Vincent Jay are technical analysts with MCM MoneyWatch, a real-time global market analysis firm headquartered in New York. For further information, contact mcm@kisnet.com.