The Wyckoff method
The answer
The Wyckoff method reads price and volume together to judge whether large buyers are absorbing supply or large sellers are distributing stock. Its central premise is mechanically true: a very large order cannot be filled in one trade without moving the price against the person placing it, so it must be worked over time, and that leaves traces in volume. Where the method becomes uncertain is in naming those traces in advance.
Why this costs you money
You already use a version of this method, and you probably use it backwards.
The common retail habit is to buy a breakout on high volume. Price leaves a range, volume is large, and the move looks confirmed. Sometimes it is. Often the high volume at the breakout is the large seller finishing the job, and the breakout is the last convenient moment for them to sell into willing buyers. You supplied the willingness.
The reverse costs the same. Price breaks below a long sideways range on heavy volume. It looks like the floor gave way, so you sell. In Wyckoff's language that break may be a spring: a move below support that fails quickly and recovers, because the purpose of the move was to trigger stop orders and buy the shares those orders released. You supplied the shares.
Both mistakes come from reading one bar. Volume on a single bar is close to meaningless, because a large number is created equally by heavy buying, heavy selling, and the exchange of one from the other. What the volume means depends on where it sits in a sequence, and on whether the price movement it produced was large or small.
That last comparison is the whole method in one sentence, and it is the part worth taking. Big volume with a small price range means somebody is absorbing what is being sold. Big volume with a large price range means nobody is.
How it works
Richard Demille Wyckoff was born in 1873 and died in 1934. He began working on Wall Street as a teenager, ran a brokerage, and founded a publication that became The Magazine of Wall Street. He wrote Studies in Tape Reading in 1910 under the name Rollo Tape, How I Trade and Invest in Stocks and Bonds in 1924, and later organised his work into a correspondence course.
He interviewed the large operators of his era and tried to describe what they did, rather than describing what a chart looked like. That is why his framework is built on a mechanism.
The 3 laws
1. Supply and demand. Price rises when buying pressure exceeds selling pressure and falls when the reverse is true. On its own this is a definition rather than a discovery. Its use is that it forces you to ask which side is being satisfied at a given price, not merely which way the price moved.
2. Cause and effect. A period of sideways trading builds a "cause", and the move that follows is the "effect". The longer and wider the sideways period, the larger the move it can support. Wyckoff measured cause using the horizontal count on a point and figure chart, which counts the number of columns across a range and multiplies by the box size and the reversal amount.
Be careful with this one. The horizontal count depends on the box size you choose and on where you decide the range began and ended. Two analysts using different box sizes get different targets from the same range. The qualitative version — a long, wide range can support a long move, and a 3-day pause cannot — is defensible. The numerical target is not a measurement.
3. Effort versus result. Volume is the effort. The price range covered is the result. When effort is large and result is small, something is absorbing the effort, and that is the signal. When effort is small and result is large, there is nothing in the way, and the move can continue easily.
The third law is the one that carries the method. It is also the only one of the 3 that gives you a comparison rather than a description.
The Composite Operator
Wyckoff asked his students to imagine a single large operator behind all market action, and to ask what that operator would want at each point.
This is a thinking device, not a claim about the world. There is no single operator. What exists is many large participants whose orders are individually too big to execute at once. The Composite Operator is a way of remembering that the other side of your trade is often a professional working a plan over weeks, not a person reacting to the same news you just read.
Used that way it is useful. Used literally it becomes a conspiracy story, and the story version is what most modern "smart money" teaching sells.
The 4 phases
Wyckoff described a cycle with 4 parts.
- Accumulation. A wide sideways range after a decline. Large buyers acquire stock without lifting the price.
- Markup. The advance. Supply has been absorbed, so less buying is needed to move price.
- Distribution. A wide sideways range after an advance. Large holders sell into strength without breaking the price.
- Markdown. The decline.
The accumulation events
Wyckoff's schematics name the events inside a range. The abbreviations are standard and you will see them everywhere.
| Event | Name | What it is |
|---|---|---|
| PS | Preliminary support | First large buying inside a decline. Price steadies briefly. |
| SC | Selling climax | Very heavy volume, wide range down, and a close well off the low. |
| AR | Automatic rally | A sharp bounce because selling has temporarily exhausted. Sets the top of the range. |
| ST | Secondary test | Price returns toward the low on lower volume than the climax. |
| Spring | Spring or shakeout | A break below the range that fails and recovers quickly. |
| Test | Test of the spring | A second, lower-volume visit to the same area that holds. |
| SOS | Sign of strength | A wide-range advance on rising volume, usually to the top of the range. |
| LPS | Last point of support | A pullback that holds above the previous resistance. |
| BU | Back-up | The final pullback before markup. |
Distribution has mirror events: PSY for preliminary supply, BC for a buying climax, UT for an upthrust, UTAD for an upthrust after distribution, SOW for a sign of weakness, and LPSY for a last point of supply.
The range is also divided into phases A to E. Phase A stops the previous trend. Phase B is the long building period. Phase C contains the spring or the upthrust. Phase D is the move toward the edge of the range. Phase E is the trend outside it.
Why the premise is actually true
This is the part that separates Wyckoff from the other 2 methods in this cluster. The claim that a large order moves the price, and therefore must be split up over time, is not folklore. It is one of the better-documented facts in market microstructure.
Albert Kyle published a model in Econometrica in 1985 in which an informed trader spreads their order over time precisely to avoid revealing information through price impact. The measure of how much price moves per unit of order flow is still called Kyle's lambda.
Robert Almgren and Neil Chriss published a framework in the Journal of Risk in 2000 for splitting a large order across time to balance market impact against the risk of the price moving while you wait. Every institutional execution algorithm in use today — volume-weighted average price, time-weighted average price, implementation shortfall — exists because of this problem.
Louis Chan and Josef Lakonishok examined institutional trade packages in the Journal of Finance in 1995 and found that large institutional trades are commonly worked over multiple days, with measurable price impact accumulating across the package.
And Jonathan Karpoff surveyed the price-volume relationship in the Journal of Financial and Quantitative Analysis in 1987, documenting a robust positive relation between volume and the size of price changes across many markets.
Put together, these say: large orders are worked over time, working them leaves a footprint, and volume carries information about that footprint. Wyckoff described this in 1910 from watching the tape. He was right about the mechanism before anybody had the mathematics for it.
Where identification becomes subjective
Now the honest half, and it is large.
Every event is named after it resolves. A break below a range is a spring if price recovers and a breakdown if it does not. There is no rule available at the moment of the break that separates the 2. The same bar carries both names depending on what happens next.
The range boundaries are chosen. Two analysts drawing the top and bottom of the same sideways area will disagree by a few percent, and that few percent is exactly the region where springs and upthrusts occur.
"Volume" is ambiguous. A high-volume bar with a narrow range is read as absorption. It is equally consistent with a large index fund rebalancing, an expiry-related transfer, a block trade crossed at an agreed price, or a single institution moving stock between its own accounts. None of those involve any opinion about the price.
The schematics are drawn from completed examples. Every published Wyckoff diagram is a finished chart. Finding the same shape on a finished chart is easy and proves nothing, which is the same trap as every other visual method.
Selection is invisible. Practitioners show the ranges where the method worked. Nobody publishes the count of ranges that produced no clear structure or that resolved the wrong way. That count is the only number that would tell you whether the method has an edge, and it is the number the exercise below asks you to produce for yourself.
What it tells you, and what it does not
Wyckoff tells you where a range's supply is being absorbed and where it is not, and that is a real question with a real answer at least some of the time. Comparing effort with result is a genuinely better habit than looking at price alone, and it costs nothing to adopt.
It tells you to slow down. A method organised around multi-week ranges will not let you trade 6 times a day, and for most readers that constraint alone is worth more than the analysis.
It does not tell you who is buying. It tells you that a lot of shares changed hands with little price movement. The inference to "an institution is accumulating" is an interpretation, and other interpretations fit the same data.
It does not give you an entry the moment the range ends. The spring, which is the highest-conviction Wyckoff entry, is only identifiable after it has already recovered, which means the good price is gone by the time the name is available.
It does not have a body of tests behind it. There is very little published academic testing of the Wyckoff schematics as such. What has support is the underlying microstructure, not the pattern-naming layer on top of it. That is still more support than Elliott or Gann can claim, and it is much less than the courses imply.
The decision rule
Judge effort against result, over a range, not over a bar.
If price has been sideways for several weeks, and volume on the down days is falling while volume on the up days is rising, and attempts to break below the range keep failing within a few sessions, then supply is being absorbed and the range is more likely to resolve upward.
Unless the sideways period follows a long advance rather than a long decline. The identical volume behaviour inside a range at the top of a trend is usually distribution, and the same absorption reading points the other way.
Where you are in the larger trend decides which reading applies. Get that wrong and every event inside the range is named backwards.
The practical form: never interpret a range without first stating what came before it. That single discipline removes most Wyckoff errors, because the most common error is reading a distribution range as an accumulation range while holding stock.
Try this now
Fifteen minutes, on your own charts, and it produces a number nobody selling this method will give you.
- Open a daily chart of a liquid share or index you follow, with a volume panel below the price. Set the range to about 5 years.
- Scroll through and find every period where price moved sideways for at least 4 weeks inside a band of roughly 10% or less. Mark each one. On a 5-year chart you will usually find between 6 and 12.
- For each range, look at the volume inside it and write down 1 of 2 answers: rising volume or falling volume across the range.
- Also write down what came before the range: a decline or an advance.
- Now write your prediction using the method, before you look right. A range on rising volume after a decline predicts an upward resolution. A range on rising volume after an advance predicts a downward resolution.
- Look at what actually happened over the 40 trading sessions after price left each range. Mark each one right or wrong.
- Count. Write 2 numbers: how many you got right, and how many of the ranges resolved upward regardless of what you predicted.
What you should see. Most people find between 6 and 12 usable ranges and get roughly half of them right. Then look at your second number. In a market that rose over the 5 years, most ranges resolved upward anyway, so a person who wrote "up" every time would have scored something similar.
That comparison is the point. The benchmark is not 50%. The benchmark is whatever "always up" scored on your sample. If your Wyckoff reading did not beat it, the method added nothing on this sample, and you now know that from your own charts rather than from an argument.
Then do 1 more pass. For the ranges you got wrong, write down whether you could have known at the time. If your reason for the error was "the spring turned out to be a real breakdown", note that clearly. That is not a mistake you made. It is the method's central ambiguity, and how often it appears in your sample is the most useful thing this exercise produces.
Three real cases
1. Kyle 1985 and Almgren and Chriss 2000 — the mechanism, confirmed and then engineered Wyckoff's premise was that a large operator must work an order over time because buying all at once would move the price against them. Albert Kyle formalised exactly this in Econometrica in 1985, with a model in which an informed trader deliberately spreads their trading to hide inside the noise. Robert Almgren and Neil Chriss turned the problem into an engineering one in the Journal of Risk in 2000, deriving how to split an order across time.
The result is that essentially every large order in modern markets is executed by an algorithm that slices it into small pieces across hours or days. Wyckoff's Composite Operator is not a fiction. It is now a piece of software, and it is running in both markets every day.
That is the strongest thing that can be said for any method in this cluster: the mechanism it assumed turned out to be real and is now standard practice.
2. Berkshire Hathaway's purchase of Apple shares, 2016 to 2018 — a large buyer working an order over quarters, visible only afterwards Berkshire Hathaway first disclosed a position in Apple in its Form 13F filing for the quarter ended 31 March 2016, and increased it substantially across the following 2 years.
Two things in this case matter. First, the accumulation took many quarters, which is precisely the behaviour Wyckoff described. Second, nobody outside the firm could see it in real time. A Form 13F is filed within 45 days after the end of a quarter, so the information arrives up to 4 and a half months late. Anybody claiming they identified this accumulation on a chart while it was happening is making a claim that cannot be checked, and the disclosure record shows how much was genuinely invisible.
3. The March 2020 decline in both markets — what a selling climax looks like, and what naming it in advance would have required Between late February and 23 March 2020 equity markets in India and the United States fell sharply on very high volume. The S&P 500 recorded its low on 23 March 2020 and the Nifty 50 recorded its low on 24 March 2020. Volume across those sessions was among the highest on record in both markets.
In Wyckoff terms this has the shape of a selling climax followed by an automatic rally. After the fact, the labelling is clean and almost everybody agrees on it.
Now look at what using it would have required. On 23 March 2020, high volume with a wide downward range was also entirely consistent with a market that was going to fall another 20%. The events that separate a selling climax from an ordinary heavy decline — the automatic rally and the secondary test on lower volume — occur afterwards. The label is reliable and it is late. Both statements are true at the same time, and holding both is what using this method honestly looks like.
The question that resolves it
A novice asks: was that volume high?
An expert asks: how much price movement did that volume produce, compared with the movement the same volume produced a month ago in this instrument?
The first question has an answer with no content. Volume is high on the day of an index rebalance, on expiry, on a block trade and on a panic. The second question is a ratio, it is comparable across time, and it is what Wyckoff meant by effort against result. A person who asks it will never again treat a high-volume bar as a signal on its own.
What would make this wrong
The claim in this article is that the mechanism is real and the pattern-naming is subjective. Here is how each half fails.
The mechanism half would be wrong if large orders did not move prices and did not need to be worked over time. That would require the microstructure literature from Kyle onward to be mistaken, and it would require the entire execution algorithm industry to be solving a problem that does not exist. This is about as close to settled as anything in this field gets.
The pattern half would be wrong if somebody defined the Wyckoff events precisely enough to detect them with a program, applied that program to many markets and many periods without adjusting the definitions afterwards, recorded every detected range including the failures, and found that the resolution rate beat the base rate after costs. Nothing prevents this study. It has not been done convincingly as far as I am aware. Until it is, the honest description is that the method has a sound premise and an unmeasured application.
The strongest objection to the method, stated fairly. If the events can only be named after they resolve, the method is a description of history. A defender will answer that experienced practitioners assign probabilities in real time and manage risk accordingly, which is a reasonable answer. But it moves the claim from "this identifies accumulation" to "this organises my judgement", and those are very different products. The second one is worth having. It is not what is usually sold.
And the honest limit of the criticism. Effort against result is testable and mostly holds. High volume with a narrow range is followed by different behaviour than high volume with a wide range, in most instruments, most of the time. A reader who takes only that from this article has taken something real.
In India
India gives you something the theory could not have expected: partial disclosure of the large orders themselves.
Bulk and block deals are published. Exchanges publish bulk deals, where a single client buys or sells more than 0.5% of a company's listed shares in a day, after the close. Block deals executed in the separate block deal windows are also published, with the buyer, the seller, the quantity and the price. This is free on the NSE and BSE websites.
That means an Indian reader can do something Wyckoff could not. When you identify a range and conclude that absorption is occurring, you can check whether any named institution actually traded in size during that period. Sometimes there is a bulk deal on exactly the days you marked. Sometimes there is nothing at all, and the absorption you saw was your interpretation of an ordinary volume pattern.
Shareholding patterns are quarterly. Listed companies file a shareholding pattern within 21 days of the end of each quarter. It shows promoter, foreign portfolio investor, domestic institutional and public holdings. Changes across quarters give you a slow but real check on whether institutions were net buyers.
Daily flow data exists. The exchanges and SEBI publish daily net buying and selling by foreign portfolio investors and by domestic institutional investors. This is aggregate, not per share, but it tells you what the largest 2 groups were doing on the days you marked.
Three Indian complications are worth knowing.
Price bands break the schematic. A share at its upper or lower band cannot trade freely. A spring that would have occurred is prevented, and a wide-range climax bar is truncated. Wyckoff reading is more reliable on indices and on the largest shares, which have wider or no bands, than on small companies.
Expiry distorts volume badly. Indian index derivatives volume concentrates enormously into expiry sessions. Volume on those days is shaped by settlement mechanics. Reading effort against result on an expiry bar is reading an artefact.
Surveillance measures change behaviour administratively. A share moved to the Additional Surveillance Measure or Graded Surveillance Measure framework can be put into 100% margin or trade-for-trade settlement. Volume collapses for a regulatory reason, not an economic one. Check the exchange surveillance list before interpreting a volume change in a small company.
In the United States
The United States has better disclosure of ownership and worse disclosure of volume than most readers assume.
Ownership disclosure is detailed. Institutional managers above a size threshold file Form 13F quarterly, within 45 days of quarter end. Anybody crossing 5% of a company's voting shares files a Schedule 13D or 13G, with 13D carrying a short deadline for activist intent. Company insiders file Form 4 within 2 business days of a trade. All of it is public and free on EDGAR.
Volume is fragmented, and this matters more than anything else here. American shares trade across many exchanges and many alternative trading systems. A large share of volume executes away from the lit exchanges and is reported through a trade reporting facility. FINRA publishes weekly and monthly volume data by alternative trading system.
The consequence for a Wyckoff reader is direct. The volume bar on your American chart may not include everything, and the composition of what it includes changes over time. A rising volume trend can partly reflect a change in where trades are printed rather than a change in participation. This is a real measurement problem and almost no Wyckoff teaching mentions it.
Retail order flow adds another layer. A large portion of American retail orders is routed to wholesale market makers rather than to an exchange. Those trades are reported, but the order flow itself is internalised. What reaches the public tape is a filtered version of activity.
The academic base is here. The microstructure literature that supports Wyckoff's premise is overwhelmingly American, using American data. If you want to read the evidence for the mechanism rather than the method, it is all in one place.
Where they differ, and what that tells you
The method is the same. The verification available to you is not.
In India you can sometimes check your interpretation against a named trade. Bulk and block deal data tells you which clients bought size and on which day. That is an unusual gift. When you mark a range as accumulation, go and look. If no large trade is recorded across the entire range, the honest conclusion is that you inferred an operator from a volume pattern, and the operator may not exist.
In the United States you can check ownership but not timing. Form 13F tells you an institution's position at the end of a quarter, up to 4 and a half months after the fact. Schedule 13D tells you somebody crossed 5%, which is a large position and a rare event. You learn who owns, not when they bought.
The volume number itself means different things. Indian equity volume is consolidated on 2 exchanges with a common clearing system, so the printed volume is close to the whole picture for a given share. American volume is fragmented across dozens of venues, with a substantial portion executed off-exchange. An effort-versus-result comparison across several years of American data is comparing numbers whose composition changed underneath you.
What that tells you is practical and slightly uncomfortable. The market with the weaker institutional research coverage gives you the better real-time trace of large orders, and the market with the deeper literature gives you a noisier volume series. If you want to test whether your absorption readings correspond to anything real, Indian bulk deal data is the easier place to do it, and it takes about 10 minutes per range.
Carry this
- Effort is volume. Result is the price range it produced. Compare the 2, never volume alone.
- Big volume with a small range means absorption. Big volume with a large range means nothing is in the way.
- The same range means opposite things depending on whether a decline or an advance came before it. State that first, every time.
- A spring and a breakdown are the same bar. Only the recovery separates them, and the recovery comes later.
- In India, check bulk and block deal data before believing an operator was there. Often there was nobody.
Knowledge check
Related
- Elliott Wave Theory
- Gann analysis: angles, squaring and time cycles
- Volume analysis and market breadth
- Point and figure charts
- Market profile and volume profile
- ← Classical methods: Elliott, Wyckoff, Gann