What are pivot points in trading?

Every pivot family starts from three numbers and only three: the High, Low and Close of the prior session. Not the open, not the volume, not the shape of the day, not what happened the week before. Three published prices. The central pivot is their arithmetic mean, which is sometimes called the prior day's typical price, and it serves as the current day's notional line of balance. From that single anchor the standard method, inherited from the open-outcry floors where a trader needed levels they could compute in their head before the session, projects the prior day's range outward in both directions.

The figure below is built to make one point that most explanations skip: the ladder is an output. The session on the left is drawn first. Its High, Low and Close are then read off that drawn path, and every level on the right is computed from those three readings on the same price axis. Nothing is typed in twice and nothing is asserted. If you moved the path, every line on the right would move with it.

Read the distance column and something odd surfaces. R1 sits 166.67 above the pivot but S1 sits 133.33 below it: the ladder is not symmetric about P, even though R2 and S2 are. That asymmetry is not a quirk of this example. It is the close leaning, and it is the seed of a set of exact identities that the next section unpacks. Notice too what the table cannot tell you: whether any of these prices has ever mattered to anyone. Each one is a true statement about yesterday's arithmetic and an entirely unsupported claim about today.

Because the whole construction is arithmetic, you can do algebra on it, and the algebra says more about pivots than any amount of chart-gazing. Three identities fall straight out, none of which is a heuristic or a tendency. They are exact, they hold on every instrument and every session, and none of them appears in the usual explanation.

Put the first and the third together and the ladder stops looking like a projection of the future and starts looking like what it is: a rubber stamp. The corridor between R1 and S1 is the prior range, translated. Its midpoint is TC, 22,383.33, while the prior range's own midpoint is BC, 22,350.00. So the corridor is the previous day's range shifted by 33.33 points, in whichever direction the close leaned. That shift, and not the levels, is the only new information the close contributes. Everything else is the High and the Low wearing different names.

This matters for a practical reason. A trader who believes R1 is a place the market has some tendency to stall is believing something about the market. A trader who knows R1 is the prior High plus a shift is believing something about arithmetic, which is safer, because it is true. The corridor being one prior range wide also quietly explains the folk observation that price spends most of its time between R1 and S1: a session stays inside the R1 to S1 corridor about as often as it stays inside the previous session's range, which is often, and which is a fact about how ranges behave rather than anything the pivot did.

The pivot ladder does not project the prior session. It re-issues it.

If the arithmetic knows nothing, why do the levels behave like levels at all? The honest answer is the one that sounds too simple to be real, and it is worth stating carefully because it is the single most misunderstood idea in this topic. Pivots matter because a very large number of people can all identify the same price without talking to each other. That is not mysticism and it is not a self-help slogan. It is a well described phenomenon: Thomas Schelling's work on focal points showed that people who cannot communicate still converge remarkably often on whichever option is conspicuously obvious to everyone, precisely because it is obvious to everyone. A price that every terminal in the country prints identically at 09:00 is about as conspicuous as a number can get.

Compare it with what happens when levels are drawn by hand. Ask five competent traders to mark yesterday's resistance and you get five prices, because there are five defensible things to anchor to: the session high, the failed retest that made a lower high, the point where the rally first stalled, the round number just above, the middle of the rejection zone. None of them is wrong. That is exactly the problem. Discretion is what produces the dispersion, so the honest output of an eyeball reading is a zone, not a price. Now ask any five pieces of software for R1 and the dispersion collapses to nothing, because there was no discretion in the input.

Hold both halves of that at once, because the whole subject lives in the gap between them. The coordination is genuine, and it is the only honest argument for watching pivots at all. But coordination is a claim about traders, not about the instrument. It says orders may cluster at 22,533.33; it says nothing about whether sellers will be strong enough there to turn anything. And it degrades exactly where you would predict: the effect needs a crowd, so it is strongest on the index futures and the heavily traded names where thousands of screens are showing the same three lines, and it thins toward nothing on an illiquid counter where the level is real arithmetic that nobody is looking at. A focal point with no crowd is just a number.

A pivot level is not more accurate than a level drawn by eye. It is more agreed. Those are different properties, and only one of them is being sold.

Most charting packages offer the standard set alongside at least two rivals, Fibonacci and Camarilla, and the menu is usually presented as a matter of taste. It is more interesting than that. All three read the same prior High, Low and Close. All three are mechanical and drawn in advance. They differ in exactly one respect: the multiplier applied to the prior range. Which means you can settle the question of whether any of them is special by running all three on one session and looking at what they say.

They disagree, and not by a rounding error. Each family nominates a price it calls the first resistance, the level a trader is most likely to actually watch. On the illustrative session the standard method says 22,533.33, the Fibonacci method says 22,481.27, and Camarilla says 22,427.50. Those three sit 105.83 points apart, which is 35 percent of the prior day's entire range, derived from byte-identical inputs.

The existence of the disagreement is the finding. If one of these multipliers were tracking something real about where a market runs out of buyers, the other two would have been competed out of existence decades ago, because the evidence would have been overwhelming and free to collect. Instead all three ship, in every package, forever, which is what you would expect if the multiplier is arbitrary and the levels derive their weight from being watched rather than from being right. There is a sharper corollary that cuts against the variants specifically: since the mechanism is the shared attention, splintering the audience across three formulas weakens the very thing that made any of them work. The strongest argument for the standard set is not that its arithmetic is better. It is that it is the default, so it is where the crowd is.

The variant that dominates Indian intraday screens is the central pivot range. Instead of a single pivot line it draws a three line band: the Pivot P, the Bottom Central (BC), the average of the prior High and Low, and the Top Central (TC), which is P minus BC added back onto P. Because TC is defined as P plus the gap between P and BC, the band is symmetric about the pivot by construction. The lore attached to it is specific and everywhere: a narrow CPR is said to precede a trending day, a wide CPR a rangebound one, on the reasoning that the width inherits the prior session's compression.

That reasoning is wrong, and the formula says so in one line. Substitute the definitions and the width simplifies to TC − BC = 2 × (Close − mid-range) / 3. The prior range is not in that expression. The width does not measure how big yesterday was; it measures how far yesterday's close finished from the middle of its own day, and nothing else. A sprawling session that closed near its midpoint produces a hairline band. A dead quiet session that closed on its high produces a wider one.

There is a real grain of truth in the folklore, and it deserves to be stated precisely rather than discarded. Because the close can be at most half a range from the mid-range, the width can never exceed a third of the prior range. So a genuinely compressed day does cap the band: a 70 point session cannot produce a band wider than 23.33 points, no matter what. The implication runs one way only. A narrow range guarantees a narrow band; a narrow band tells you nothing about the range. The folklore takes a true one-way implication and runs it backwards, which is how a great deal of chart lore is manufactured.

Here is the test that decides what pivots are. Compute one ladder from one prior session, exactly as a trader would at 09:00, and then run three different days against it without changing a single line. The levels cannot adapt, because they were fixed before any of the three began. Whatever the ladder is worth has to be worth it in all three cases, since at the moment of drawing there was no way to know which one was coming.

Three failure modes are visible there, and they have different causes. The gap is the cleanest: pivots are computed from yesterday and drawn before the open, so when the market reprices overnight the ladder is a description of a session the market has already left behind. If you want the mechanics of why that repricing happens where it does, the guide on a gap up and gap down covers it. The trend is subtler: a level only holds if participants choose to defend it, and in a directional move they are not defending, they are chasing. Watching price slice R1, R2 and R3 is not the levels failing, it is the levels being exactly as informative as they always were, which is to say silent about what kind of day this is. The third is the thin counter, where the arithmetic is perfect and irrelevant, because the coordination mechanism from the last section needs a crowd and there is not one.

The standard advice is to use pivots “with confluence”, meaning to prefer levels where a pivot coincides with something else: a prior swing high, the edge of an unfilled gap, a multi-day congestion boundary, the session's VWAP. The advice is sound, and the reason it is sound is worth being blunt about. A pivot is a claim made by arithmetic. A swing high is a claim made by the market, a price where buyers demonstrably ran out. Those are different kinds of evidence, and a level carrying both has two independent reasons to matter instead of one. This is where pivots meet the material in supply and demand and in ordinary support and resistance: structure is earned and local, pivots are assumed and global.

What nobody says out loud is how seldom the coincidence happens. Audit the ladder against real structure on any given day and the arithmetic is honest about it. Seven levels are spread across roughly 900 points. The market has, at any moment, a handful of genuinely earned levels in that span. The chance that they land on each other within a few points is small, and that is not a defect: it is what you should expect when one set of prices is derived from a formula that has never looked at the chart.

Two out of seven is the honest number, and the near miss is the most instructive row in the table. P at 22,366.67 sits 27 points above a real congestion top at 22,340. Twenty-seven points is a meaningful distance intraday, comfortably enough to take a stop out. But it is close enough to be rounded into a confluence by a trader who has already decided to be long, and that rounding is precisely how confirmation bias enters a method that was supposed to be objective. The discipline is to fix what counts as a coincidence before you look, not after. Pivots do not save a trader from that judgement; the mechanical part ends the moment the lines are drawn, and everything expensive happens afterwards.

Strip out the claims that did not survive and something genuinely useful is left, but it is not what the levels are usually sold as. Pivots are a pre-marking device. Their value is almost entirely in the fact that they exist at 09:00, before the market has had a chance to make anyone feel anything. A trader who marks their levels in advance has moved a decision, where will I pay attention, out of the moment when they are least equipped to make it and into a moment when they are calm. That is worth real money, and note carefully that it has nothing to do with whether the level holds.

This is also why the objectivity matters more than it first appears. A level drawn by eye during a session is contaminated by the session: fear draws support lower, hope draws resistance higher, and the chart obligingly seems to agree. A level that was computed from published numbers before the bell cannot be moved by how the morning is going. It is not a better level. It is an uncorrupted one, and on a bad day that is the more valuable property.

1. Compute before the open, not during. The inputs are fixed and public the moment the prior session closes. Anything computed after the bell is contaminated by what you have already watched happen.
2. Mark the whole ladder, then decide which lines you will actually respect. Most of them will land on nothing. Deciding that in advance is the work; deciding it at 11:40 while a position is open is not deciding, it is rationalising.
3. Write down what counts as confluence before you look for it. A number, in points, fixed in advance. Otherwise every near miss becomes a hit in the direction you are already leaning.
4. Treat a level as a place to look, never as a place to act. The reaction is the evidence. A clean breakout through R1 that holds and a rejection at R1 are opposite messages from the same line, and the line cannot tell you which one you are about to get.
5. Know when the map is void. A gap past your levels or a day that has already cut two of them means the ladder is describing a market that no longer exists. Retiring a map is a decision that has to be made in advance too.

None of those five steps is about the formula. The formula took one line and a division. Every one of them is about the judgement wrapped around it: which level deserves respect, what would prove the idea wrong, when to stop believing the map. That upstream judgement, and not the arithmetic, is what the method we teach is built around, for the same reason it recurs in every topic on this site. The computable half of trading is the half that was already solved.

Pivots occupy an unusual position in technical analysis, and it is worth naming precisely. Almost every other tool on a chart requires a choice: which lookback for the moving average, which threshold on the oscillator, which swing counts as a swing. Each choice fragments the audience, and each is a place for a trader to fool themselves. Pivots have no choices in them at all. That makes them the cleanest object on the chart and, at the same time, the emptiest. There is nothing to tune because there is nothing to tune with: three numbers in, seven numbers out, no room for information to enter.

So the fair summary is narrow and, once you accept it, quite useful. Pivots give an objective, shared, discretion-free map of the session, drawn before it starts, that a large number of participants are looking at simultaneously. Used for orientation, for pre-marking, and as context alongside VWAP, real structure and actual price behaviour, they organise a chart and they organise a trader. Used as prediction they will disappoint, and the disappointment is diagnostic: it means the tool was being asked for the one thing it was never able to supply. A formula that has never seen a chart cannot tell you what a chart is about to do.

The deeper lesson generalises past pivots. The reason this ladder is so seductive is that it is certain, and certainty is scarce and comforting on a chart. R1 is {fmt(R1)}, exactly, and no one can argue. But the certainty is about the arithmetic, not about the market, and the two get conflated constantly, because a precise number feels like knowledge. It is the same error that makes a backtest with two decimal places feel more trustworthy than one with none. Precision is cheap. Being right is not, and nothing in this page's formula is even attempting the second thing.

Pivots are the honest half of a chart: fully mechanical, fully shared, fully known in advance, and fully ignorant. Everything that decides the outcome is in the half they cannot touch.

Two kinds of claim appear on this page and they deserve different treatment. The formula specifications below are conventions, documented consistently across charting platforms and trading education, and they are cited as such. The identities and the CPR width result are not citations at all: they are algebra, they are derived in full on this page, and any reader can check them in a spreadsheet in about a minute. That is the appropriate standard of proof for an arithmetic claim, and it is a higher one than a citation.

• Focal points and coordination. Thomas C. Schelling, The Strategy of Conflict (Harvard University Press, 1960), which introduced the focal point: the observation that participants who cannot communicate converge with surprising frequency on whichever option is prominent for the simple reason that it is prominent to everyone. This is the mechanism behind the shared-attention argument for pivots, and it is why that argument is real rather than mystical.
• Standard floor-trader pivot specification. P = (High + Low + Close)/3, R1 = 2P − Low, S1 = 2P − High, R2 = P + range, S2 = P − range, with R3 and S3 projected a further range out. This is the widely documented specification used across charting platforms and is the default pivot set on most of them.
• Central pivot range specification. P = (H+L+C)/3, BC = (H+L)/2, TC = (P − BC) + P, with the TC to BC distance read as the CPR width. The specification is documented consistently across Indian intraday-trading education. The widely repeated interpretation of the width as a measure of the prior session's compression is addressed and corrected on this page from the formula itself.
• Camarilla and Fibonacci specifications. Camarilla multiplies the prior range by 1.1/12, 1.1/6, 1.1/4 and 1.1/2 and adds and subtracts those from the prior Close. Fibonacci adds 38.2, 61.8 and 100 percent of the prior range to and from the central pivot. Both are standard alternative families offered alongside the floor-trader set.
• The arithmetic on this page. Every level, identity and spread quoted here is computed from the single illustrative prior session drawn in the first figure (High 22,500, Low 22,200, Close 22,400), and the figures derive their levels from that same drawn path rather than from typed-in values. The identities R1 − S1 = High − Low, TC = midpoint(R1, S1) and width = 2(Close − BC)/3 hold for all inputs and were verified exhaustively against randomly generated sessions.