Higher Time Frame Mean Reversion
One of the main problems in technical analysis is distinguishing exhaustion from strength. Exhaustion is indicative of overextended moves that will soon reverse; real strength (or weakness) occurs with moves that should have continuation in the same direction. If it were always possible to tell one from the other, we would always know whether to fade moves or to enter pullbacks for continuation, but, of course, it is not possible to make these distinctions every time. Though there are characteristics associated with each, there are also many similarities and even our best tools work within the laws of probability.
There is no certainty in trading; the best we can hope for is to find something that will give a slight, reliable bias or tilt. Some type of well-calibrated band or channel—for instance, the Keltner channels I use and discuss in this work—can provide a good visual and quantitative reference for overextension.
On a single time frame, you would not want to make a standard practice of buying strength above the upper band or shorting weakness below the lower band. Figure 7.21 shows a pullback on daily Wheat futures that might have been buyable, taken out of context: the market consolidated after a strong upthrust, put in a series of smaller-range bars, and eventually showed a return of upside momentum at the spot marked on the chart.
However, the weekly chart shows a long tail and a clear exhaustion above the channel, greatly reducing the attractiveness of the potential long entry on the daily chart. Many traders think that a good analytical system will get them into more winning trades, but equally valuable is a system that will reduce the number of losing trades. Being aware of higher time frame overextension and potential exhaustion is one tool that can help traders avoid trades that immediately fail and collapse under the force of higher time frame mean reversion. (Note that in all of these charts the left pane expands on the highlighted section of the right pane.
The lower time frame is in the left pane, the higher time frame in the right.) Figure 7.22 shows another example, this time in the broad U.S. equities market. At the end of the daily chart, an aggressive trader could have justified taking a shot at a short. True, this was not the best possible trade setup, but the market had just completed a sloppy complex pullback after making new lows, and had been locked in a strong downtrend for many quarters. However, the weekly chart again puts this trade in context.
After three pushes down, the last of which was on a glaring momentum divergence (i.e., price made a new low and the MACD fast line was unable to do so), it would have been unrealistic to expect a clean short. There are two valuable lessons here: First, the weekly chart does not clearly support a long trade, but it did offer enough upside
evidence to negate a potential short setup on the lower time frame. In this case, a set of conditions that might not have fully supported a long entry were enough to justify not taking a potential short setup. Second, most of the trades that are contradicted by higher time frames are also not excellent setups on one time frame. To some extent, focusing on the geometry of a single time frame will also wrap in some of the multiple time frame factors, though there is still valuable information to be gleaned from analysis of neighboring time frames.
Higher time frame considerations are not limited to providing filters to skip trades; sometimes they also can be motivation enough to justify trade entries. Figure 7.23 shows LULU, a market-leading stock in February 2011, extended above the upper band on the weekly chart, again with momentum divergence on the MACD.
The tight consolidation on the daily chart might potentially be a place for aggressive longs to consider adding to positions or entering new positions. Looking only at the daily chart, we would never consider shorting under such a consolidation, because a high and tight flag is usually indicative of real conviction from buyers; furthermore, shorting an extremely strong market leader in any context is usually a recipe for pain. It is far better to spend your time and mental energy figuring out how to buy those market leaders and how to short laggards, in most cases. However, in this case, the weekly chart provided another layer of information, and the clear overextension made it possible to justify a quick short under the little range on the daily chart. There is also another lesson here: after a longer
pullback, this stock spent some time consolidating and eventually headed much higher. Swing traders cannot be greedy.
Your job is to take what opportunity the market offers and to be quick to realize when the trade is over. These examples have focused on longer time frames, but the same principles apply to much shorter time frames as well. Figure 7.24 shows 5- and 30-minute charts of the USDJPY in the early evening (New York time) on March 16, 2011.
The 5-minute chart shows a structure that could potentially be shortable for at least a retest of recent lows, though the presence of a selling climax, indicated by the overextension past the lower band, should give shorts pause for concern. In this case, the 30-minute chart provides context that is not as clearly visible on the 5-minute chart. The market is drastically over-extended beyond the lower band on the higher time frame; this usually happens in response to a significant exogenous shock—in this case the aftermath of the 2011 Tohoku ¯ earthquake in Japan. In a situation like this, you may still choose to take the short trade, but you have to realize that the character of the trade, especially if you are wrong, will be different.
There is a much higher probability of a dramatic failure and reversal when the market is overextended on the higher time frame, so a trade taken under these conditions has a completely different risk profile. Skip the trade altogether, take it and manage the risk with a tighter stop, or choose to get out at the first sign you might be wrong—these are all acceptable alternatives. What is not acceptable is ignoring the clear message of the higher time frame extension and taking an unnecessary large loss as you continue to short into the mean-reverting higher time frame.
Higher Time Frame Pullbacks To understand the effects of higher time frame pullbacks on the trading time frame, first review some of the characteristics that define high-probability pullbacks on a single time frame:
- Presence of a good impulse setting up further continuation.
- Market is not overextended.
- Market is not on the third (or later) trend leg.
- Absence of momentum divergence.
- Absence of buying or selling climax.
- Pullbacks show generally lower activity and volume than the trend legs.
These conditions and patterns, when they appear on the higher time frame, can provide motivation for and add confidence to trades on lower time frames.
Figure 7.25 shows a pattern on the daily chart of Crude Oil futures that might not be compelling by itself. It is obviously some form of a pullback, but the move up off the lows was maybe a little too far too fast to justify shorting; it could also be read as a potential setup for a long Anti trade, but it is not an example of the best possible trade
on this time frame. However, the weekly chart shows a clean bounce, which was the first reaction after the collapse from the overextended highs and the first pullback after a strong impulse move pushed prices below the lower channel. The weekly pattern added confidence to the daily consolidation, and justified a short attempt on a breakdown out of this pattern. Figure 7.26 shows another example on weekly and daily charts of Goldman Sachs from October 2010. The pattern on the daily chart was anything but convincing: many large gaps and sudden reversals. Though the market was in an uptrend from the September lows, it might have been difficult to see this at the time. However, the weekly chart shows a pattern that we know well: several marginal new lows on slackening momentum followed by a sharp reversal off those lows. The upward reversal (not visible on the daily chart) made new momentum highs on the weekly chart and set up an Anti trade. This weekly chart pattern put a clearly bullish context on an otherwise difficult daily chart, and traders would have been justified aggressively pursuing long setups on the daily chart. In this case, there were many small bullish setups on this time frame: failure tests of support, breakouts of small daily flags, and, finally, an overextended climax just beyond the weekly measured move objective for the weekly Anti. At this point, realizing that the weekly chart had basically fulfilled its expectations for a measured move would have removed much of the bullish context from the daily patterns, justifying a reduction of risk or perhaps a complete exit from the trade.
Higher time frame pullbacks can also be a filter to skip trades that set up against those pullbacks. Figure 7.27 is an important example that shows what might have been seen as a good setup for a breakout on the daily chart: the consolidation up against resistance normally suggests the possibility of a strong break above that level, and is the opposite of the price rejection that normally accompanies a level holding. However, the weekly chart provided context, showing that the breakout was actually coming near the top of what was more likely to be bearish complex consolidation on the weekly chart. In this case, the higher time frame provides a bearish bias, which clearly contradicts the smaller pattern on the daily chart. The consolidation on the weekly could certainly fail, and it would do so through the success of the bullish breakout on the daily, but the probabilities favor downside continuation. There are a number of things you can do with this information, ranging from skipping the trade on the daily to taking it on smaller risk (which is not usually a good idea), or perhaps even watching for a breakout failure and entering a short trade on the daily. This last possibility combines a small pattern on one time frame with a bigger-picture bias from a higher time frame, which is an excellent use of multiple time frame information. At the very least, beware of patterns on one time frame that are clearly contradicted by strong patterns on higher time frames, especially when the patterns occur at potential inflection points.
