Overview
“Gaps” (as they are called in the West) and “windows” (their Japanese counterparts) have always attracted the attention of technicians – most probably because they are nearly impossible to be missed on a price chart. After all, a trading session lying completely outside of the prior day’s range, which is what gaps and windows are by definition, must carry some kind of predictive power. However, a critical question remains – are gaps and windows indicative of the beginning of a new trend (as it was the case for AAPL in Figure 1), or are they simply an overreaction and are subsequently quickly filled (as it was the case for MMM in Figure 2)? Notice how in the former case the gap stayed opened (and it still is) for more than a year, whereas in the latter case the gap was filled/closed within two months (i.e. subsequent price action in September completely overlapped the range of the gap).
Figure 1. AAPL Daily
Figure 2. MMM Daily
Note, from here on, only the term “gap” is used, even though there is an important difference between the two – “gaps” look at intraday prices when determining if they are “filled”, whereas “windows” only look at “closing” prices. At the bottom of the article, you can find references to several works on the price phenomenon, in which the difference between the two variations is discussed in-depth.
Given that gaps occur quite frequently (see referenced materials for details), it is easy to understand how any “gap” strategy can be depicted to have predictive power. Most traditional books on technical analysis include a list of gap trading strategies followed by a few stellar charts that are supposed to prove the strategies’ validity (charts similar to Figures 1 & 2). Furthermore, given gaps’ conspicuous nature, most traditional trading strategies have their “entry” on the day the gap occurs (or Gap Day). For example, proponents of gaps being a continuation pattern would suggest that taking a position in the direction of the gap on Gap Day is profitable. On the other hand, technical analysts who believe that “all gaps get filled” suggest taking a position on Gap Day in the opposite direction of the gap. In both cases, action is taken on Gap Day.
After trading and analyzing gaps for many years, I was certain that traditional theories do not work the way they are described to. Probably the most illogical stipulation made by various technical authors was that “the gap itself should serve as support or resistance” and “once filled, gaps become insignificant”. On the contrary, I had found that the gap itself is rarely a strong support or resistance and that very often the most significant gaps are those that have already been filled. This is when, in 2016, I developed a new theory on gaps (“K-Divergence), which significantly “diverges” from traditional theories.
Before discussing what the K-Divergence theory entails, an explanation of the most popular traditional theories is presented.
Traditional Gap Theories
1. A gap is a continuation pattern.
Strategy – taking a position, on Gap Day, in the direction of the gap.
This theory is based on the idea that if, on any given day, prices jump/fall significantly enough to never touch the prior session’s price range, something significant must have occurred and changed the market’s sentiment on the company. In this case, on Gap Day, prices are assumed to reflect the changing opinion of the stock only partially, and thus, further movement in the direction of the gap is expected. In the case of AAPL’s gap (Figure 1), on February 1, 2017, the company reported better-than-expected 1Q17 earnings on the heels of record breaking iPhone sales. Subsequently, the price continued moving higher in a swift fashion, leaving the up-gap behind it. Often, proponents of this theory use support and resistance levels, or technical indicators, as a confirmation that the gap has occurred at an important juncture and that it can be trusted. For example, zooming out and looking at the stock’s price action since 2015 (Figure 3), traders who utilize gaps as continuation patterns can claim that “the breakout occurred above the interim high of the multiple bottom formation, and therefore, carried high predictive power”.
Figure 3. AAPL 2-Day Chart
2. A gap is an overreaction.
Strategy – taking a position, on Gap Day, in the opposite direction of the gap
Advocates of this theory are convinced that gaps are a result of market participants overreacting to news (or “noise”) and that once participation subsides, the gap is expected to get filled. The famous adage “all gaps get filled” is often used in an attempt to support this supposition. Similar to the previous strategy, support/resistance levels and technical indicators are expected to provide further confirmation if the particular gap is to be filled. For example, looking once again at the MMM chart (2-day chart – Figure 4), one may say that the down-gap took prices close to a well-established uptrend (green trendline) and to a key moving average (100 SMA – yellow line). Also, to further support the thesis that prices will reverse, one may point to the positive reversal in RSI (not to be confused with a positive divergence), which indicated that the correction has taken the stock to oversold levels during the uptrend (i.e. RSI making a lower low, while prices making a higher low).
Figure 4. MMM 2-Day Chart
The above two strategies are a perfect example of technical analysis being more of an “art” than “science”, where it is up to the technician’s discretion to decide what action to take after observing a gap. While, after testing both strategies (see K-Divergence section), I found that neither the simple continuation nor reversal strategies are profitable on a systematic basis, there are definitely specific situations where the probability of a gap reversing is higher and vice versa. Unfortunately, the next strategy, the one found in almost any TA book, is one of the reasons why technical analysis has a bad name among most non-technicians.
3. A gap could be either a continuation pattern or an overreaction based on its “classification”.
Strategy – an ambiguous one based on hindsight
As it had become evident that gaps cannot be all “continuation patterns” or “overreactions”, the popular gap classification system was born – where gaps are categorized as either “breakaway”, “continuation/runaway”, “exhaustion” and “common”. This classification is based on two criteria – 1) the location of the gap relative to preceding price action and 2) whether the gap gets filled or not. However, as one can imagine, there is no way to know on Gap Day whether a gap will be filled in the future. That is, the classification system is based on hindsight. Let’s prove this point by looking at an example. Figure 5 shows an up-gap after a prolonged uptrend. Based on the widely-used classification system this gap can be “runaway” (if the gap does not get filled and prices continue higher), “exhaustion” (if prices quickly reverse, fill the gap and continue lower) or even “common” (if prices fill the gap but do not reverse or consolidate). Given the colour of the candle and the long upper wick, it seems like it is an “exhaustion” gap, right?
Figure 5. Daily Chart (real chart, ticker hidden)
Clearly, it is only in hindsight that this gap can be classified. In this case, the gap turned out to be of the “runaway” type as it did not get filled and the stock (MSFT) continued propelling higher (Figure 6). The point is, the classification system is futile for making decisions on Gap Day.
Figure 6. MSFT Daily Chart
K-Divergence (K-Div) Theory
4. Most gaps occur after prices have moved away from a significant support or resistance levels.
Strategy – taking a position in the direction of the gap, only after prices have returned to pre-gap levels
More specifically, the theory suggests that in most cases an up-gap transpires after prices have already jumped from a key support level and a down-gap – after prices have already fallen from a key resistance level. The theory is based on the premise that before a gap occurs prices have already reached a key level and have bounced from it. It is only later on, after most market participants agree on the direction of the next move and take positions in the same direction that gaps occur. This means that it is not the gap itself that should serve as a support or resistance, but rather the range of prices preceding it (pre-gap range). The most important implications of the theory are – 1) the gap itself should not serve as support or resistance and 2) a filled gap is not “insignificant”.
So why does the K-Divergence make sense from a technical point of view? After all, if prices gapped due to “news” that nobody was aware of, this would mean that gaps are nothing more than prices adjusting to the new information. Any such conclusion should render fundamental and technical analysis useless, for it would imply that no analyst is able to purchase a security before news gets disseminated. On the contrary, the K-Divergence assumes that the most astute market participants (i.e. the best fundamental and technical analysts, quants and even “insiders”) are able to trade in advance of the gap occurring. Therefore, true support and resistance levels lie prior to the gap transpiring and subsequent filling of the gap does not render it “insignificant”. It is best to illustrate this with an example. I will use one of my most recent predictions based on the theory, which was sent to one of my clients. First, I will describe the rational in detail with an updated chart (Figure 7), which will be followed by screenshots from the day the signal was given.
After the close on February 1, 2018, Google reported its 4Q18. The next day, the stock opened sharply lower and continued falling into the close (Feb 2 – Down Gap in Figure 7). The stock continued falling along with the market until the Feb 9 low was set. Subsequently, while NASDAQ was making new highs in early March, GOOG reached the pre-gap range (Bearish K-Divergence Range – violet horizontal trendlines) and started stalling. Due to the strong bounce by the broader markets, the stock recovered and filled the gap. However, when the stock started trading at the pre-gap range, market participants were given a second chance to sell the stock for the same price it was trading at before the 4Q18 earnings were released. Price action confirmed the bearishness of the set-up (GOOG March 13 & 16 – Figures 8 & 9).
Figure 7. GOOG Daily Chart
Figure 8. GOOG March 13
Figure 9. GOOG March 16
In order to validate the theory, I developed two trading strategies based on it (one with the gap and one with the window variation) and backtested them along with 5 variations of the traditional gap strategies discussed above. Figure 10 shows the 1-, 2-, 5-, 10-, 20-, 30- and 44-day period returns of the 7 strategies (#6 & 7 being the two based on the K-Divergence theory) and Figure 11 shows the annualized returns for the those same periods. The backtest took into account a total of 14,219 gaps over nearly a 2-year period.
Figure 10. 1-, 2-, 5-, 10-, 20-, 30- and 44-day period returns
Figure 11. Annualized 1-, 2-, 5-, 10-, 20-, 30- and 44-day period returns
The two K-Div strategies were profitable throughout all periods. The only other consistently profitable strategy was “Fading the Gap” strategy which entailed taking a position in the opposite direction of the gap on Gap Day, but closing it immediately after the gap was filled. This once again goes against traditional theories which suggest that once a gap is filled, prices should continue going against the gap’s direction as, supposedly, an important support/resistance was breached.
It is noteworthy that the K-Divergence theory does not suggest that all gaps have occurred after important support/resistance levels or that they can all be traded profitably in a similar fashion as the GOOG example. Rather, it provides a framework for analyzing the gap phenomenon, on that all active investors/traders should believe in, which assumes that some market participants are able to act ahead of major moves (i.e. prior to the appearance of gaps). Furthermore, it eliminates the use of the “hindsight” gap classification system.
For more on gaps, I recommend reading Julie R. Dahlquist and Richard J. Bauer’s “Technical Analysis of Gaps” book, where they conduct, one of the first on the topic, objective investigations of the phenomenon. For a much more in-depth coverage of the K-Divergence and my research on gaps, you can view my thesis for the Master of Financial Technical Analysis (MFTA) Program, published in the 2018 IFTA Annual Journal.
Conclusion
In the future, regardless of whether you look for opportunities to trade gaps on Gap Day (strategies 1 & 2) or decide to use the K-Divergence as part of your trading arsenal, I hope this article would make you think more critically the next time you hear terms such as the “runaway” gap. And even more importantly, will push you to analyze gaps even after they have been filled, and according to traditional theory, have become insignificant.
Happy gap trading.
Featured image courtesy of Shutterstock.
