Confirming Divergences

That's true. But a trader can "get out" any time. The trick is to get out before losing significant amounts of capital — while at the same time staying with the trade long enough to give prices time to do their work. Another way of putting this is to say that traders in general — and top 'n bottom traders in specific — need a method of knowing as soon as possible whether or not the trade is one that is likely to work.

Swing traders, who rely on a short 3-5 day window in which to enter and exit trades, tend to be the most sensitive to this imperative. I can't help but think of trader Linda Bradford Raschke's reply to Jack Schwager in his book, New Market Wizards, when he suggested, "It sounds like you would be profitable in every month." Her response? "Every month! My philosophy is to try to be profitable every day!"

Raschke wasn't boasting. She was simply pointing out that her methodology and mentality as a trader required a higher winning percentage than other traders. Elsewhere, in her book, Street Smarts (coauthored with Larry Connors), Raschke encourages short-term traders not to "carry losing positions overnight." I've long been personally enamored of trend trading. But the fact of the matter is that my constitution simply requires more numerous, shorter-term trades. As Paul Tudor Jones put it more eloquently in his interview with Jack Schwager in Market Wizards, "I just can't stand the kind of drawdowns you find in trend trading."

So if you are going to play tops and bottoms, then not only do you need to look for and/or create excellent risk/reward scenarios, but you will also need some sort of confirmation that a top or bottom has actually occurred. Many commonsensical ways of accomplishing this — such as stopping out with a new high if trying to short or with a new low if trying to buy bargains — unfortunately expose traders to outsized losses too frequently. This look at trading tops and bottoms uses stochastic divergences as a method of spotting tops, and uses a very straightforward confirmation strategy to let traders know when a reversal from a top or bottom has a good chance of producing the sort of follow-through that can yield profitable trades.

Calling the top versus confirming the top

For our purposes here, I'll focus on one technique for calling tops and bottoms: the stochastic divergence. However, most of what I'll be writing about is relevant to other oscillator-based ways of spotting both positive and negative divergences. One of the problems with using divergences to spot reversals is that, in strongly trending markets, divergences will pop up from time to time as markets move back and forth between being overbought and oversold in the course of an overall trend. Think of it like a price moving back and forth between the upper and lower boundaries of a trend channel. Yes, when prices move to the lower boundary of an upward trend channel, they stand a chance of breaking through that boundary. But if the trend is strong, then that lower boundary will serve as a trampoline, with prices bouncing back from this support toward the upper boundary.

But the problem is that when stochastic divergences work, they often work spectacularly, providing traders with a low risk/high reward entry that would make even the most ardent trend trader green with envy. Consider, for example, the positive stochastic divergence in the S&P 500 near the end of October that I wrote about for Traders.com Advantage (Divergences And The Bush Bounce," Traders.com Advantage, November 9, 2004). A trader or investor observant of that divergence could have easily entered the market, say, in the days leading up to the US presidential election, and would be sitting on a near 70-point gain in the S&P and more than 700 points in the Dow. So, to my way of thinking, divergences provide too much potential reward for them to be ignored. The trick is to minimize the risk for those instances — for example, during strong trends — when those divergences crap out.

How does the stochastic divergence work? Consider it from the long side. A stochastic divergence occurs when the stochastic oscillator is out of sync with price action. Typically, this takes place when prices make a new low or high and the stochastic fails to confirm with a like new low or new high. In a downtrend, what is called a positive divergence occurs when sequentially higher stochastic troughs match sequentially lower price troughs. In an uptrend, what is called a negative divergence occurs when sequentially lower stochastic peaks match sequentially higher price peaks. In both instances, the divergence signals the potential for a reversal in trend.

Nevertheless, there is an important difference between calling or signaling a top or bottom and actually confirming that a top or bottom has occurred. Ideally, a trader or investor wants to be as early as possible in both instances. But while the divergence itself provides an alert to an opportunity, the divergence itself cannot give us a particularly reliable "signal" to buy or sell. For that, I am suggesting that traders and investors turn away from the indicators and back to price action itself.

What can price action tell us — and tell us as soon as possible — about the likelihood of a reversal in the wake of a stochastic divergence? In short, and as the examples will show, traders need to see price action follow through in the direction of the divergence before they should feel confident about being aggressive in pursuing the reversal. For example, Larry Connors (author of How Markets Really Work as well as coauthor of Street Smarts) describes one reversal setup he calls Turtle Soup. This setup essentially involves fading breakouts and breakdowns. After a few instructions on the "buy" version of this setup, Connors writes, "After the market falls below the prior 20-day low, place an entry buy stop 5-10 ticks above the previous 20-day low." If this "one day only" buy stop is hit, then Connors has traders enter a stop-loss just under the more recent 20-day low. This approach to playing reversals has both of the key virtues I've been talking about: The potential for early entry and the follow-through in the direction of the expected reversal.

Having faith in follow-through

Now let's add the influence of stochastic divergences to this set of virtues. Requiring a stochastic divergence to trade setups like Connor's Turtle Soup does limit the number of opportunities, but at the same time, the best divergences often occur with markets that are particularly primed for a reversal. Also, rather than make the entry "5-10 ticks" away from the previous low or high, we will look at simply waiting for prices to reverse and rebound through the high (if playing a positive divergence) or the low (if playing a negative divergence) of the previous divergence low or high, respectively.

That probably sounds more confusing than it really is. So let's go to the charts. Figure 1 is a 30-minute chart of the S&P 500 since early December.

Figure 1. A positive stochastic divergence announces the opportunity of a bottom in this 30-minute chart of the S&P 500. Confirmation comes swiftly as the market rallies beyond the high of the initial low.

Here we have the 30-minute S&P 500 making a low late on December 7, then following up with a lower low late on December 9. At the same time, the 7,10 stochastic made a low on December 7, and a higher low on December 9. This was the positive stochastic divergence that signaled the potential for a bottom and a subsequent reversal. But when would traders know that the bottom was in and the reversal confirmed? I believe that as soon as prices rose above the bar or candlestick that accompanied the initial low back on December 7, the reversal was established and could be pursued aggressively. More specifically, as soon as prices rose above 1179.65, the S&P 500 was a buy. This number was hit by the wide-range bar on December 8 around noon. The market that day closed at 1189.24 — nearly 10 points above the entry level after the first day.

Let's take a look at an instance — this time involving a negative stochastic divergence — in which the divergence signal takes place, yet the anticipated short trade does not materialize. Figure 2 will show how the requirement of follow-through can also help keep traders and investors out of losing bets.

Figure 2. A negative divergence early in an uptrend should be looked at with suspicion. But suspicion or no, the lack of follow-through to the downside doomed chances of a reversal here.

Shortly after the positive divergence just discussed, the market ran into a negative stochastic divergence. While the uptrend was still relatively young — and not the type of mature uptrend that is far more likely to succumb to a negative divergence — it is worthwhile to see whether following the guidelines presented here would result in a losing trade. The negative stochastic divergence in Figure 2 is a running divergence, one that features multiple matches of consecutively lower stochastic peaks along with consecutively higher price peaks. Running divergences can be tricky to read, as traders try to determine which peaks are the more relevant ones. But, as you will see, any set of divergences will often suffice for purposes of determining the likelihood of a reversal.

Let's go through the methodology. We have a pair of consecutively higher price peaks on Thursday, December 9, and Friday, December 10. These peaks are matched by consecutively lower stochastic peaks. So there is the negative divergence. Confirmation? The reversal would be confirmed if prices moved — and preferably closed — below the low of the bar/candlestick that matches up with the initial stochastic peak. That 30-minute bar/candlestick was formed at the end of trading on Thursday, with a low of 1188.10. Now, for a trader to go short the S&P 500 — or for an investor to go to cash — based on this method, the S&P 500 would need to make a 30-minute low close below 1188.10. As it turned out, the final half-hour of trading on Friday saw a 30-minute close at 1188.00 — 10 cents below the target number. With a stop at the high of the bar/candlestick corresponding with the second stochastic peak (at about 1191.45), traders would have learned first thing on Monday morning that the trade was a loser and best exited with a relatively small loss.

Odds and ends

Interestingly, those looking at a somewhat larger negative divergence would have been spared even that small loss. Using the December 13 price peak as the second peak (instead of the less impressive, late-in-the-day December 10 peak), traders would have still had the same "confirmation number" of 1188.10. But that number never would have been hit, as prices retreated to an intraday low of 1190.25 before rallying steadily over the balance of the day.

As the second example shows, even a confirmed reversal can result in a loser. The point, however, is not merely to guess right, but to be protected when the guess goes wrong. By providing for as early an entry into a potential reversal trade as possible, this approach to trading stochastic divergences helps minimize the damage in the event that the reversal does not follow through as significantly as expected. One additional feature I add to this method is to use half the range of the peak/trough (bar/candlestick) to provide some space from the previous low or high that will serve as a confirmation price. For example, with Figure 2, rather than enter the trade short at 1188.10, I would have taken the range of that confirmation bar/candlestick (approximately 2.41 divided by 2 for 1.21) and subtracted that from 1188.10 for a confirmation price of 1186.89. Rather than use a set number of ticks or cents, using the range of the confirmation bar helps take into account the volatility of that specific environment. That is, with this approach, a wide-ranging confirmation bar (suggesting higher volatility in prices) would inevitably produce a confirmation price that was farther away, minimizing the likelihood of getting whipped into a trade.

I should add that this conversation about trading tops and bottoms is not just for those who actually do enjoy trading tops and bottoms. Even dedicated trend traders can benefit from an alert that a trend may be topping. In addition, long-only investors looking to buy when stocks are cheapest can put these strategies to good use — even if these methods may appear geared solely to those bungee-jumpers and high-wire artists who thrive on top- and bottom-picking.

David Penn may be reached at DPenn@Traders.com.

Suggested reading

Charts courtesy Prophet Financial Systems, Inc.

Current and past articles from Working Money, The Investors' Magazine, can be found at Working-Money.com.