Continuation patterns and gaps
The answer
A gap is a real, physical event: a range of prices at which no trade occurred, because the market was closed or because a rule stopped it. A continuation pattern is a picture of one thing only — the range of daily prices getting narrower — and narrowing range predicts the size of the next move, never its direction.
Why this costs you money
Two different mistakes live in this article, and they cost money in opposite ways.
The first is trading a triangle for direction. You see an ascending triangle: a flat resistance level with rising lows underneath it. Every book says this breaks upward. You buy in anticipation. It breaks downward, which happens often, and you are now holding a position taken on a rule with no evidence behind it. The real content of that shape was that the daily range was contracting, which told you a larger move was coming. It never told you which way.
The second is believing that gaps always fill. You see a stock gap down 12% on a result. You buy, because gaps fill. Sometimes that works within days. Sometimes the gap is still open 4 years later, and you have spent 4 years holding a position based on a sentence with no time limit attached to it.
"Gaps always fill" is not a claim. A claim has a deadline. Without one it can never be shown to be false, which is why it survives.
There is a third cost that is purely Indian and purely avoidable. A stock goes ex-bonus or splits, and the chart shows a 50% overnight fall. That is not a gap. Nothing happened. If your platform has not adjusted the history, you will find "support levels" and "gaps" that are arithmetic from a corporate action. People trade these every week.
How it works
Part 1: continuation patterns, which are all the same pattern
Here are the shapes, stated plainly.
| Shape | What it looks like |
|---|---|
| Flag | A sharp move, then a small tilted range against the move |
| Pennant | A sharp move, then a small triangle |
| Symmetrical triangle | Lower highs and higher lows, converging |
| Ascending triangle | Flat highs, rising lows |
| Descending triangle | Falling highs, flat lows |
| Rectangle | Flat highs and flat lows — this is simply a range |
| Wedge | Both boundaries sloping the same way, converging |
Now notice what every one of them shares. The distance between the high and the low of each day is getting smaller. That is the only property common to the whole list, and it is the only property with evidence behind it.
Volatility clustering, which is real
Markets have a well-documented statistical property: quiet periods follow quiet periods, and violent periods follow violent periods. Robert Engle described this formally in 1982 and received a Nobel prize for the work in 2003. It is one of the most robust facts in financial economics, found in every liquid market anybody has examined.
The practical consequence is exactly what the pattern books claim, and for once the claim is supported: a period of contracting range is more likely than average to be followed by an expansion. A squeeze does precede a move.
Now read the limit carefully, because this is where the entire subject goes wrong. Volatility clustering is a statement about size. It contains no information about direction. None. The mathematics describes the magnitude of returns, not their sign.
So the honest translation of every continuation pattern is one sentence: the daily range has been narrowing, so a larger move is more likely than usual, and nothing here tells me which way it goes.
What is decoration
The direction rule. "An ascending triangle breaks upward." There is no mechanism for this and the studies that have looked have not supported it as a reliable rule. the specific literature on triangle breakout direction.
The measured move. For a flag, you are told to project the length of the "flagpole" from the breakout point. This is the same convention as the reversal targets in the previous article. No order sits at the target.
The names. Whether a shape is a pennant or a symmetrical triangle depends on how many days it took and how much it sloped. Nothing in the market knows the difference.
The trend context claim. "Continuation patterns continue the prior trend." This is true in the weak sense that trends do have some tendency to persist, which is already covered in the first article of this cluster. It is not an extra fact supplied by the shape.
There is a real and useful part left over, so keep it. The boundaries of a contracting range are often genuine levels for the reasons in the second article of this cluster: a flat top with 3 touches on real volume is resistance, and the orders there are real. Trade the level. The triangle is just how the level was drawn.
Part 2: gaps, which are genuinely real
A gap is a range of prices at which no trade took place. If a stock closes at ₹500 and opens at ₹540, there is no record of anybody buying or selling between ₹500 and ₹540. That is not an interpretation. It is a fact about the trade record.
Gaps exist for exactly 3 reasons.
1. The market was closed. Information arrives at all hours. Prices cannot move while the exchange is shut, so the adjustment happens in one step at the opening auction.
2. A rule stopped trading. A price band, a circuit limit, or a trading halt prevents the price from passing through a range even while the market is open.
3. A corporate action changed the number. A split, a bonus issue, a rights issue or a large dividend changes the price mechanically. This is not a gap. It must be removed before you look at anything.
The 4 traditional gap types
| Type | Where it appears | What it usually means |
|---|---|---|
| Common | Inside a range, on low volume | Nothing much. Fills often and quickly |
| Breakaway | Out of a base, on very high volume | New information. Often stays open a long time |
| Runaway | In the middle of an established trend | The trend accelerating |
| Exhaustion | After a long trend, on very high volume | The last buyers. Often fills fast |
Be honest about this table. The distinction between a runaway gap and an exhaustion gap can only be made after you know whether the trend continued. In real time they look identical. So 2 of these 4 categories are labels applied in hindsight, and you should treat them as descriptions of history rather than as tools.
The common and breakaway distinction is more usable, because it rests on 2 things you can check on the day: whether the gap occurred inside a well-established range or out of one, and whether volume was ordinary or several times normal.
The fill question, answered properly
A gap is "filled" when price trades back through the range that was skipped.
The right way to ask the question is: what fraction of gaps fill within 5 trading days, within 30, and within a year? Once you attach a window, the sentence becomes testable, and the answers are useful. Small gaps inside ranges fill often and fast. Large gaps out of a base on 5 times normal volume frequently do not fill for years.
The reason is not mysterious. A small gap in a quiet range is a temporary absence of sellers. A large gap on high volume is a change of opinion, and there is no reason for opinion to travel back through a price everybody has abandoned.
What it tells you, and what it does not
A contracting range tells you that a larger move is more likely than usual. It tells you nothing about direction. Anyone who tells you otherwise is repeating a convention.
A gap tells you that a repricing happened without trading. It tells you that there is a price region where nobody holds a position from, which is genuinely useful: there are no trapped holders inside a gap, so the usual resistance mechanism from the second article is absent there. Price often moves quickly through a gap region for exactly this reason, in either direction.
A gap does not tell you that a move will continue. It does not tell you that a move will reverse. And the size of a gap is partly a fact about the market's rules — its opening hours, its price bands — rather than about the news.
The decision rule
For a contracting range: prepare for size, not for direction.
If the daily range has been narrowing for several sessions, then reduce your position size, widen your stop or stand aside, because the next move is likely to be larger than recent ones. Take a direction only from a break of a level that has real volume behind it.
For a gap: classify it by volume and by context on the day it happens.
If the gap is small, inside a range, on ordinary volume, then it is likely to fill soon, and it is not information.
If the gap is large, out of a base, on several times normal volume, and tied to identifiable news, then treat it as a new price and stop expecting the old one. This is the gap people lose the most money waiting on.
Unless the gap was caused by a corporate action or by a price band, in which case it is not a gap at all and no interpretation applies.
Try this now
This is a census, it takes about 5 minutes, and almost nobody has ever done it on their own holdings.
- Open a daily chart, 1 year, of a stock or index you actually follow.
- Go through it and mark every day where the open was outside the previous day's high-to-low range. That is a gap. Most people find between 5 and 25 in a year on an Indian stock, and fewer on a US one.
- Remove the false ones first. Open the company's corporate actions page on the NSE, BSE or the company's investor relations site and check for splits, bonus issues, rights issues and large dividends. Any "gap" on those dates is arithmetic. Delete it from your list.
- For each remaining gap, write 3 things: the size as a percentage, whether the volume that day was ordinary or several times normal, and how many trading days passed until price traded back through the gap.
- Count. What fraction filled within 5 days? Within 30? How many are still open?
What you should see. A clear split, and it is the same split for nearly everybody.
The small gaps on ordinary volume mostly fill, and mostly within a few days. The large gaps on very high volume mostly do not fill quickly, and some of them will still be open at the end of your year.
If you did this on an Indian stock, you will also notice how many of the gaps have no news attached to them at all. They are the result of a closed market plus ordinary overnight drift, and they are the reason Indian charts look choppier than US charts of the same underlying volatility.
A second step, 60 seconds. Take the largest unfilled gap on your list and write down the date it happened and what the news was. If you cannot find the news in 60 seconds, that gap was probably common, not breakaway, and it may fill after all.
Three real cases
1. The NIFTY 50, 3 and 4 June 2024 (India) — 2 enormous gaps in 2 sessions Exit polls for the general election were published over the weekend of 1 and 2 June 2024. The index gapped up sharply at the open on Monday 3 June and closed up about 3.3%. The actual results arrived on 4 June and the index gapped down and fell about 5.9%, one of its largest single-day falls in years. Both gaps were filled within days as the index recovered. the exact percentages and the date on which each gap was filled. The case is instructive because the size of both gaps was a direct product of market structure: the information arrived while the exchange was closed, and there was no continuous pre-market session in which opinion could adjust gradually.
2. Meta Platforms, 2 and 3 February 2022 (United States) — the gap was an artefact of which session you look at Meta reported after the close on 2 February 2022 and the shares fell heavily in after-hours trading within minutes. By the time regular trading opened on 3 February, the price had already been trading near its new level for hours. The stock closed down roughly a quarter that day. On a daily chart the move appears as a very large gap, and it is genuinely a gap in regular-hours data. But it was not an unfilled auction in the way an Indian gap is. Thousands of trades occurred across that price range between 4:00 pm and 9:30 am. the exact closing decline and the after-hours price path. The lesson generalises: in the United States, the daily chart hides a session where the repricing actually happened.
3. The S&P 500, 23 and 24 March 2020 (United States) — a breakaway gap that did not fill The index reached its pandemic low on 23 March 2020. The Federal Reserve announced open-ended asset purchases that morning. On 24 March the index gapped up and rose about 9%. That gap was a breakaway gap by every criterion: it came out of a collapse, on very heavy volume, on identifiable news that changed what every future dollar of profit was worth. It has never been filled. the exact percentage gain on 24 March 2020. A trader who sold into that gap expecting a fill was waiting for a price that a large number of buyers had already decided was wrong.
The question that resolves it
A novice looks at a gap and asks: will it fill?
An expert asks: did anybody change their mind, or did the market just reopen?
The second question has an answer you can find on the day, from volume and from the news. A gap on ordinary volume with no news is a mechanical consequence of an exchange being closed. A gap on 5 times normal volume with a clear cause is a new opinion, and old prices have no claim on it.
What would make this wrong
If contracting ranges carried directional information, then a rule stated in advance — for example, buy every ascending triangle breakout across 500 stocks over 20 years — would beat a rule that took the same trades in both directions. whether a well-designed study of this exists. The claim in this article is that no such directional edge has been established, while the volatility clustering result is very well established.
If gaps filled reliably, then a rule of fading every gap would be profitable after costs, with a stated holding period. It is not, and the reason is visible in the census above: the gaps that fill are small and the gaps that do not are large, so the strategy wins often and loses badly.
Three honest limits.
First, the volatility clustering result is about the statistical distribution of returns over many observations. It does not promise that any particular squeeze resolves into a large move. Some squeezes simply continue being quiet.
Second, the classification of gaps in this article depends on volume, and volume is noisier than it looks — fragmented across venues in the United States, and distorted by expiry and index events in both markets.
Third, this article treats the boundaries of triangles as ordinary levels, which carries over all the limits from the article on support and resistance, including the requirement to fix a tolerance before you count anything.
In India
Indian charts contain far more gaps than US charts, and 3 features of the market explain it.
There is one session and no continuous pre-market. Continuous trading runs from 9:15 am to 3:30 pm. There is a pre-open call auction from 9:00 am, with order entry until about 9:08 am and matching until about 9:12 am, which produces a single equilibrium opening price. That auction is a price-discovery mechanism, not a trading session. So all overnight information is expressed in one step. the current pre-open session timings on the NSE website.
Price bands turn 1 event into several gaps. Most stocks outside the derivatives segment carry a daily price band, commonly 5%, 10% or 20%. When news is larger than the band, the stock locks at the band and stops trading. The next day it opens at the band again and locks again. A single piece of information then appears on the chart as 3, 4 or 5 consecutive gaps on consecutive days. Nothing new happened on days 2 through 5. The rule simply rationed the move.
This is exactly what happened to several Adani group companies from late January 2023, after a short seller published a report on 24 January 2023. Indian markets were closed on 26 January for Republic Day, and over the sessions that followed several of the group's stocks repeatedly locked at their lower price bands, with Adani Enterprises withdrawing a large follow-on public offer on 1 February 2023. the exact sequence of circuit locks and dates. Read that stretch of chart as 1 event, not as 6.
Corporate actions are frequent and large. Bonus issues and stock splits are much more common in India than in most markets. A 1-for-1 bonus halves the quoted price overnight. Charting platforms differ in whether and how they adjust history. Always check the corporate actions page before you interpret a large overnight move in an Indian stock.
There is also an index-level protection. Market-wide circuit breakers halt trading across the exchanges when the index moves 10%, 15% or 20%, with the length of the halt depending on the level and the time of day. the current thresholds and halt durations in the SEBI and exchange circulars.
In the United States
Extended-hours trading absorbs much of the news. Pre-market trading is available from 4:00 am Eastern time and after-hours trading until 8:00 pm on most retail platforms, against a regular session of 9:30 am to 4:00 pm. Most US companies report results outside regular hours precisely so that the market has time to digest them. The consequence for charts is direct: a great deal of what would be a gap in India is instead a continuous move in extended hours, and the regular-hours opening price is often close to where the stock was already trading.
Volume in extended hours is thin, and that matters. A price set at 5:00 am on a few thousand shares is not the same quality of information as a price set at 10:00 am on millions. So extended-hours trading reduces the size of gaps without necessarily making the resulting price more reliable.
There are no daily price bands on individual stocks. The US uses limit-up-limit-down, which pauses trading for 5 minutes when a stock moves outside a band around its recent average, and then allows trading to resume. A stock can therefore complete a 40% fall in 1 session. Market-wide circuit breakers halt all trading at 7%, 13% and 20% moves in the S&P 500, with different rules late in the session. the current level 3 rules and timings.
Opening auctions are large and concentrated. The New York Stock Exchange and Nasdaq opening and closing auctions handle a very large share of daily volume, and the closing auction in particular has grown enormously as index funds have grown. A price set in the closing auction is a real, high-volume price and is a better anchor for a level than an intraday print.
Where they differ, and what that tells you
This is the clearest structural difference in the whole cluster, and it changes how you should read a chart.
In India, a gap is usually a genuine unfilled auction. Nobody traded in that price range, anywhere, at any time. There is no record of a single transaction between the 2 prices. That makes the gap a real feature of the order book's history: no holder in the world bought inside it, so there are no trapped holders there, and price tends to travel through the region quickly when it returns.
In the United States, a gap on a daily chart is usually a gap in the regular-hours record only. The price range was traded, often heavily, between 4:00 pm and 9:30 am. Holders do exist inside it. The gap is a fact about which data your chart shows you, not about whether trading occurred.
What that tells you is a practical rule, and it runs in both directions.
If you are in India and you read US-written material about gap trading, remember that the American author's gaps were partly filled before the open. Their fill statistics do not transfer. Your gaps are larger, more frequent and more often caused by nothing more than the market being shut.
If you are trading US stocks, look at an extended-hours chart before treating a gap as untraded space. Many platforms let you switch extended hours on with 1 click. The "gap" often disappears entirely, and with it the reason you expected price to move through it quickly.
And in India specifically, before you interpret any large overnight move, run 2 checks that take 30 seconds each: the corporate actions page, and whether the stock was locked at a price band. Those 2 checks remove most false gaps from an Indian chart, and almost nobody performs them.
Carry this
- A narrowing range predicts the size of the next move, never the direction.
- A gap is an auction that never happened. Check volume and news before you interpret it.
- "Gaps always fill" with no deadline is not a claim. Add a window and count.