Volume Based Support and Resistance

What support and resistance already means

Most of this will be familiar, and it is worth restating only because everything that follows is a departure from it.

Support is a price where demand has been strong enough to stop a decline — a floor, found by looking at previous lows. Resistance is the mirror: a price where selling has been strong enough to stop an advance, found from previous highs. Dynamic versions of both move with price rather than sitting still, and are usually a moving average or a trend line.

The weakness is not that any of this is wrong. It is that a level found this way is found from price alone — from where the market turned, not from what it cost to turn it. Two levels that look identical on the chart can have completely different amounts of business behind them, and nothing in the classical definition distinguishes them.

What follows adds that missing dimension.

A horizontal support level under price and a resistance level above it, with a sloping dynamic level cutting through both — The classical picture. Every level here was found by looking at where price turned, and at nothing else. Click to enlarge

Two kinds of level, and only one of them is new

Under the model in this lesson there are exactly two sorts.

Structural levels are the ones you already have. They are the P2 and P3 points from lesson one, read on the daily and hourly charts. Nothing new is needed to find them.

Statistical levels are computed rather than observed. They come from collecting price, time and volume together and asking, in effect, where the market has agreed to do business — which is a question about probability rather than about shape.

He attaches one restriction to the structural side and it is worth repeating because it contradicts what most people do with these points:

In a downtrend, P2 is not tradable. P3 is, using the setups from lesson two.
In an uptrend, the same: P2 is not tradable, P3 is.
The structural levels are used mainly for partial exits, not entries. They are the "no trade" zone from lesson one, and the reason is unchanged — the end of a movement is not something you can price.

Structural levels taken from the trend points beside statistical levels computed from price, time and volume — One kind you already know how to find. The other has to be computed, and most of this lesson is about the computing. Click to enlarge

A downtrend and an uptrend with the tradable and non-tradable points marked, headed structural support and resistance — Which structural points may be traded and which may not, marked on both directions. P2 is never one of them.

Why bother with a statistical level at all

Because of what sits at one. A statistical level is a price where a measurable amount of business has already happened, and that has consequences the chart shape does not.

It marks where the market previously met real buying or selling pressure, which is a statement about the psychology of the people in it rather than about the drawing. It gives you somewhere defensible to put a stop or a target. And it concentrates activity: these are the prices where volume, volatility and the chance of a turn are all higher than average.

There is a harder reason too, and he is direct about it. A great deal of what trades at these prices is algorithmic. Automated systems are built on exactly this kind of statistical point — they are back-tested on it and they manage risk with it. So the level is not only where people once traded. It is where a lot of machines are currently instructed to act.

Four different calculations, all producing the same kind of thing: a price with a reason attached to it. Click to enlarge

A chart carrying the daily, weekly and monthly weighted averages together with the daily pivot — Three lengths of weighted average and the daily pivot on one screen — the levels themselves before anything is read into them.

The toolkit

Four sources, in the order he introduces them.

Pivot points are computed from the previous period's high, low and close. The pivot itself — PP — is the one that matters most.

VWAP, at three lengths. Daily, weekly and monthly, plus the Midas anchoring from lesson two. Short-term traders read the intraday one to judge whether price is cheap or dear today; longer-term ones use the weekly or monthly to see the average over a stretch that matters. He notes three ways to use it: comparing price against it, watching the crossing, and putting deviation bands around it — the last of which gets its own section below.

The volume journal, which files volume by price rather than by time. It gets its own section too.

Volume profile and TPO, which are the rest of the lesson.

One property of VWAP is worth pulling out because it separates it from almost every other tool on a chart: it is instantaneous. Unlike a moving average it carries no lag, so it is usable in a live market rather than as a description of one that has already happened.

The same statistical level taken once where it agrees with the trend and rejected where it runs against it — Two identical levels. What separates them is not the calculation — it is which way the structure was already going. Click to enlarge

A chart where the pivot and the weekly weighted average meet inside a correction, with a second cluster below marked as unusable — The combination he wants: pivot and weekly VWAP landing together inside a correction — and the cluster below it he refuses because it faces the wrong way.

The rule that outranks everything else here

This is the most important paragraph in the lesson and it is his, almost word for word.

Every statistical element must line up with the structure of the market. If the statistics contradict the structure, they are not used. Structure and the direction of the movement take precedence over everything.

So the purpose of all the calculating is narrower than it first appears. It is not to generate signals. It is to find the prices where several of these measurements land together — because that is where the most algorithms are triggered, and the more of them coincide at one price, the more the price is worth. But the whole cluster is worth nothing if it faces the wrong way.

His own example is the clearest statement of it. A pivot and the weekly VWAP meeting inside a correction in an uptrend is a level to work with. Lower down the same chart, a monthly VWAP and a pivot support meet just as neatly — and he refuses it, because it is against the direction and comes after the trend has broken. It may still produce a reaction. Taking it is still high risk.

Horizontal lines drawn at prices where unusually large positive or negative delta was recorded on earlier days — Levels that are older than the window they are drawn on — which is the point of keeping the journal separate from the chart. Click to enlarge

A twenty day chart carrying horizontal lines drawn from volume recorded far earlier, with entry points marked where price meets them — A twenty-day window carrying levels from four hundred days of volume. The tool is ICF Market's own, written for NinjaTrader.

The volume journal

Volume filed by price, not by time

This one is worth explaining carefully because it does something the others do not.

A volume journal files the volume that has been spent over time at each specific price. It can be grouped by day, week, month, contract or any custom period, and the volume can be separated by delta, by time, by size and by price.

The property that matters is that it works independently of the chart. If your chart is showing the last twenty days, the journal can still draw levels built from the last four hundred. Those lines tell you where traders were interested in a price, and that information is simply not present on an ordinary chart at all.

What to look for is the lines carrying a very large positive or negative delta. Those are the ones that act as support and resistance when price returns to them, and combining them with the other factors above is where the clustering he is after comes from.

The tool is ICF Market's own — his slide says it was first developed by the ICF Market programming team for the NinjaTrader platform.

The parts, named. Everything later in this lesson is read off these four things. Click to enlarge

Both profiles on one chart, with the value area at 68 per cent and the point of control marked on each.

The seven parts named: profile, value area, point of control, high and low volume nodes, and the two value area edges.

Volume profile and TPO, part by part

Two tools that answer neighbouring questions.

A volume profile shows how much traded at each price over a chosen period. An ordinary volume histogram shows volume against time; this shows it against price, which is a different picture of the same data and a far more useful one for finding levels.

TPO — time price opportunity, from Market Profile — shows how much time the market spent at each price instead. The longer it stayed, the more significant the price. Volume says how much was done; TPO says how long they stayed to do it.

The components are the same vocabulary either way:

Point of control (POC) — the price with the most volume. Treated as where the market found equilibrium, or fair value.
Value area — the band containing roughly 70 per cent of the activity. On his charts it is drawn at 68.
Value area high and low (VAH, VAL) — the two edges of that band, and the levels most often referred to later.
High and low volume nodes — prices with unusually heavy or unusually thin trade. A thin one is a price the market passed through without wanting to stay.

A volume profile is a distribution lying on its side. Once you see that, the shapes stop being patterns and start being statistics. Click to enlarge

Where the shape comes from

He takes a detour into statistics here, and it pays off in the next section.

The bell curve — the normal distribution — is the shape data takes when most values cluster near the mean and fewer occur far from it. It is symmetrical: the two halves mirror each other around the middle.

A volume profile is that curve turned on its side. The point of control is the mean, the value area is the fat middle, and the thin ends are the prices few people wanted. Nothing more mysterious than that is going on.

Which makes skewness the interesting measurement. A perfectly symmetrical distribution has zero skew. Positive skew means the data is concentrated to one side with a long tail running the other way; negative skew is the mirror. On a profile, skew is simply which end the volume piled up at — and that is what the next section is about.

Three distributions, three names. The letters are just the silhouette the volume leaves behind. Click to enlarge

The three shapes on real charts — P, b and D — with the profile that produced each one beside it.

P, D and b

Three shapes, and he attaches a number to each so the reading is not a matter of opinion. The measure is SPP, and the thresholds are his:

P — volume concentrated at the top. Positive skew, SPP above +0.4. A buyer-dominated period.
D — volume balanced through the middle. SPP between −0.4 and +0.4. Supply and demand in agreement.
b — volume concentrated at the bottom. Negative skew, SPP below −0.4. A seller-dominated period.

This is the part people skip. The shape does not carry a direction on its own — the trend it appears in supplies that. Click to enlarge

Several small diagrams of price moving between value area highs and lows across successive sessions — How the value area of one session becomes the reference for the next, drawn across several days.

A diagram of the hourly secondary trend with the powerful patterns marked, under the condition that the trend has broken — The powerful patterns under a trend-break condition, with the percentages he attaches to each side.

A wall of small diagrams sorting continuation and reversal patterns by whether the close was inside or outside — Every combination of pattern and trend he sets out, sorted by whether the close landed inside or outside.

The same shape means opposite things

Here is why the previous section stopped at the definitions.

A P in an uptrend is continuation: buyers hold the upper prices and the move is likely to carry on. The same P in a downtrend is a warning of reversal — significant buying has happened at higher prices despite the selling, which is power moving from sellers to buyers.

A b reverses the logic. In a downtrend it is continuation, sellers pressing at the lows — though it can also mark the point where they are exhausted. In an uptrend it is strength gathering at the low prices, which usually means sellers are building and a turn may follow.

A D is agreement, and agreement is ambiguous. In an uptrend it can be consolidation before the next leg, or the end of the advance. In a downtrend it can be the market finding a floor. His instruction for a D is not to read it as a signal at all but to go looking for further evidence.

Which is the same instruction as the rule further up this page, arriving from a different direction: the shape is one input, and the structure decides what it means.

The same value area three times with the closing price inside it, above it and below it — Three closes, three readings. This is the cheapest useful thing in the lesson — it needs one profile and one closing price. Click to enlarge

Where the close sits relative to the value area

A short section with a good ratio of usefulness to effort.

Inside the value area. Supply and demand are roughly balanced, price is where most of the business was done, and the market has not settled on a direction yet. Expect stability in the short term rather than a move.

Above it. Buyers were able to push price out of the area where most trade happened and keep it there to the close. That is a real statement about pressure, and it argues for continuation upward.

Below it. The mirror image, and it argues for continuation downward.

None of the three is a trade by itself. All three are read together with the profile shape and the trend, which is the combination the next section sets out.

Three candle closes against the point of control: a reversal, a shock and a continuation — His own grouping. Each one also carries a delta condition, which is what stops the shape being read on its own. Click to enlarge

A grid of candle closes sorted into reversal, shock and continuation patterns, each with its delta condition and buy or sell label — His full grid: reversal, shock and continuation, each with the delta condition attached and the side it argues for.

TPO, single prints and excess

TPO adds two ideas the volume profile does not have, and both are about prices the market refused to spend time at.

A single print is a price touched only once in a time bracket — one lonely block with nothing beside it. It says the market went through that price quickly and did not linger, which usually means a strong move or an event. Those thin prices tend to matter later: when price comes back, there is little business there to absorb it, so it either hesitates or turns.

A single print excess is the same thing at the end of a move rather than in the middle of one. Price has gone further than the market was willing to support, and the prints mark the endpoint. He treats these as a warning that a trend is finishing and another may be starting.

He also names AVO peak levels — a proprietary calculation in his own software, sensitive to fast algorithmic moves, which marks those points on the chart. Reversals at them, he says, tend to react well.

Combining the two profiles is where this gets useful. Where the volume profile and the TPO overlap, price and volume agree, and the overlap is a strong consensus area. Where their highs and lows sit far apart, the amount traded and the time spent disagree — which can mean supply and demand are shifting and a new move is starting.

A time price opportunity chart with its blocks stacked by price, one price touched only once, and an excess print at the extreme — The prints the market left behind at prices it did not want. They are worth more later than they look at the time. Click to enlarge

A TPO chart with the point of control, the value area, single prints and an excess print at the extreme all marked — The TPO components on a live chart: the value area at 68 per cent, the single prints, and the excess at the turn.

An overlap strong enough to stop the market — his own example of the two profiles agreeing.

The deviation bands

Put standard deviation bands around VWAP and you get a measure of how stretched price is against the average it has actually traded at.

Reaching the upper band means price is a long way above the day's weighted average — overbought, in the ordinary sense. The lower band is the mirror. These are the levels that attract traders looking for a turn, because they look unsustainable.

His warning is the one that matters: price can stay stretched for longer than you expect, so the bands confirm rather than trigger. What he actually watches at them is the algorithmic activity inside and outside the range.

The settings are his and they differ by instrument, because the two behave differently:

Currency pairs — deviation of 2 to 2.5. Rate differentials, data releases and geopolitics all move them, so the normal range is wider.
Indices — deviation of 1.5 to 2. Sentiment and the constituent companies drive them, and the typical excursion is smaller.

A volume weighted average with bands at two standard deviations either side and price tagging the upper one — How far price has gone from where the business was done. Stretched is not the same as finished. Click to enlarge

The euro future with weighted average deviation bands drawn at plus and minus two and a half, and the algorithmic peaks marked — The bands on 6E, with his note to watch what the algorithms do inside and outside them.

Merging profiles

Merging joins several days' profiles into one. The purpose is to see a range area properly — where the balance between supply and demand actually sits when a single day is too small a sample to show it.

He is unusually careful to say what this is and is not. It is descriptive rather than a strategy, it needs a lot of practice, and he says plainly that if you are going to use it you should have tested it thoroughly first. His conditions:

The point of control matters more after a merge than before. A daily POC is not normally traded; a merged one is a different proposition.
Do not merge more than a week. The weekly profile is what the analysis is built on.
Only merge days that overlap enough. Never merge a movement together with a correction.
The aim is a D. If the merge does not produce one, work out whether it produced a P or a b and read it accordingly.
It must not destroy the other analysis. If merging removes the information you had, it was the wrong merge.
It has to agree with TPO. If the two profiles come out very different, or inverted, do not use the merge.
The merged area should be covered by higher than average volume.
Areas that form before a significant news release are among the most useful to merge — but remember that behaviour during the release itself can be temporary, so watch how price treats the area afterwards.

Same data, two readings. The merged view is what makes a range area legible; it is also the easiest thing here to do badly. Click to enlarge

The same stretch of chart shown twice, once with separate daily profiles and once with them merged — Before and after, side by side. The separate days say one thing; merged they say another.

The worked example: a b profile at 1.75, an extended point of control, and the rejection that came off it.

The opening range, and the initial balance

Two related ideas, and he uses one of them far more than the other.

The initial balance is the high and low set during the first hour of the regular session — two hours for European equity indices, from 8am to 10am CET, which covers the old futures open through the first hour of equity trading.

The opening range is the same idea over a shorter window, and the window depends on the market: half an hour for US equities from 9:30, half an hour to an hour for the DAX, a session open such as London's for forex, half an hour after the New York open for commodities — and the first two minutes for futures, because of how violent that market is at the open.

His own practice is specific. Because he trades around the US futures open, he does not use the initial balance there — the volatility is too high — and uses the opening range instead. The same applies in European futures such as the DAX. The initial balance is used only at the very start, to get an overall view and a first set of levels.

Why the levels work at all is the same answer as everywhere else in this lesson: liquidity is sitting there. And the same restriction applies — the placement and delta of the opening range matter, and it is not to be used against the structure of the market.

The high and low of the first minutes of a session drawn forward as levels the rest of the day reacts to — Two minutes of a futures session, carried across the rest of it. The reaction is liquidity, not the drawing. Click to enlarge

About a month of the S&P future with each session's opening range marked and price reacting to them — A month of ES with every opening range marked. The reaction is liquidity sitting at those prices.

What to take from this one

Three things, in order of how much they matter:

A level is worth what agrees with it. The point of the pivots, the averages, the journal and the profile is not any one of them — it is finding the price where several land together, because that is where the algorithms are pointed.
Structure outranks statistics. A cluster that faces against the trend is not used, however neat it looks and however well price reacts to it. That is his rule, not a caution added here.
The same profile shape means opposite things in an uptrend and a downtrend. P, D and b are descriptions of where volume piled up, not directions — and reading them without the trend is the main way this material gets misused.

The three in the order they matter. The second one is what stops the first from becoming a signal generator. Click to enlarge

Next: the order book, the tape and market speed

This lesson found levels from what has already traded. Lesson five is what is waiting rather than what is done — resting bids and offers, the running tape, iceberg orders, and what a change in speed tells you that price does not.

Back to the twelve lessons