Markets usually do not trend straight up, nor do they fall vertically downward. They usually retrace some of the advance, or decline, before resuming the prior trend. Some of the more popular retracement levels are the 50% level and the Fibonacci figures of 38% and 62% (see Exhibits 12.1 and 12.2). Fibonacci was a 13th century mathematician who derived a special sequence of numbers. Without getting into too much detail, by comparing these numbers to one another one could derive what is called-not surprisingly-the Fibonacci ratios. These ratios include 61.8% (or its inverse of 1.618) and 38.2% (or its inverse of 2.618). This is why the 62% (61.8% rounded off) and the 38% (38.2% rounded off) corrections are so popular. The popular 50% correction is also a Fibonacci ratio. The 50% retracement is probably the most widely monitored level. This is because the 50% retracement is used by those who use Gann, Elliott Wave, or Dow Theory.

Exhibit 12.3 illustrates how well retracements can help predict resistance areas in a bear market. The 50% retracements in gold over the past few years have become significant resistance levels. Let us look at three instances on this chart where 50% retracements melded with candlestick techniques to provide important top reversal signals.

Retracement 1-The highs at A in late 1987 (\$502) were made by a bearish engulfing pattern. The selloff which began in late 1987 ended with a piercing pattern at B at \$425. Based on a 50% retracement of this selloff from A to B, resistance should occur at \$464 (this is figured by taking half the difference between the high at A and the low at B and adding this onto the low at B). Thus, at \$464 you look for resistance and confirmation of resistance with a bearish candlestick indicator. A bearish engulfing pattern formed at C. At C, the high was \$469 or within \$5 of the 50% correction. The market began its next leg lower.

Retracement 2-The selloff which began at C ended at the morning star pattern at D. Taking a 50% correction from C’s high at \$469 to D’s low at \$392 gives a resistance area of \$430. Thus, at that level, bearish candlestick confirmation should appear. Gold reached \$433 at area E. During this time (the weeks of November 28 and December 5 (at E)) gold came within \$.50 of making a bearish engulfing pattern. Another decline started from E.

Retracement 3-From the high at E to the low at F in 1989 (at \$357), prices fell \$76. (Interestingly, all three selloffs, A to B, C to D, and E to F fell about \$77.) There were no candlestick indicators that called the lows on June 5. The second test of these lows in September came via a hammerlike line.

The next resistance level, a 50% retracement of the decline from E to F, is \$395. Not too surprisingly gold surpassed this level. Why wasn’t this a surprise? Because, in late 1989, gold pierced a two-year resistance line. In addition, gold built a solid base in 1989 by forming a double bottom at the \$357 level. Thus, we have to look out farther to a 50% retracement of the larger move. This means a 50% retracement of the entire decline from the 1987 high (area A) to the 1989 low (area F). This furnishes a resistance level of \$430. Near this \$430 level, at \$425 on the week of November 20 at area G, the market gave two signs that the uptrend was in trouble. Those signs were a harami pattern and, as part of this pattern, a hanging man. A few weeks later, on the week of January 22, the highs for this move were touched at \$425. The following week’s price action created another hanging man. Gold declined from there.

Look at Exhibit 12.4. The combination tweezers and harami bottom at \$18.58 (A) was the start of a \$3.50 rally. This rally terminated at \$22.15 (B) with a bearish engulfing pattern. A 50% correction of the A-B thrust would mean support near \$20.36. At area C, a bullish piercing pattern formed at \$20.15. The market then had a minor rally from C. This rally ran into problems because of the dark-cloud cover at D. Interestingly, D’s high at \$21.25 was within 10 ticks of a 50% bounce from the prior downleg B-C.

Exhibit 12.5 reveals that a Fibonacci 62% retracement of rally A-B is \$5.97. This also coincides closely with the old resistance level from late January and February at \$5.95. That old resistance converted to support. On the pullback to this level, on April 2 and 3, this \$5.97 held as support. These sessions formed a harami pattern that signaled an end to the prior minor downmove. Then, just for good measure, there was an additional test of this support in mid-April, and away went the beans!

Exhibit 12.6a shows from crude oil’s July low at L, to its October high at H, there was a \$21.70 rally. A 50% retracement of this rally would be at \$29.05. Thus, based on the theory that a 50% retracement level from a rally should be support, we should look for a bullish candlestick indicator near that \$29.05 area on the brisk selloff from October’s high. This is what unfolded. On October 23, after prices touched a low of \$28.30, a hammer developed on the daily chart. The market rallied over \$5 from that hammer. On the intra-day chart of the price action on October 23 (see Exhibit 12.6b) we see the first hour’s action also formed a hammer. Thus, the daily candlestick chart on October 23 and the first hour on the intra-day candlestick chart on October 23 both had hammers. This is a rare, and as we see, significant occurrence. Note how, on the intra-day chart, the brisk rally that began with the hammer ran out of force with the emergence of the hanging man on October 26.