Analysis · Timeframe
Intraday Trading and Its Arithmetic
Shorten the holding period and the move available shrinks with the square root of time, while the cost of a round trip stays where it was. That single relationship decides more about short-horizon trading than any indicator does.
The sum that comes before everything else
A price series’ variation grows roughly with the square root of elapsed time. If an instrument moves about 1.6 per cent in a typical session, then over a quarter of a session it moves about half that, and over a hundredth of a session about a tenth of it. That is a property of how prices accumulate, and it holds well enough for this purpose on any liquid instrument.
The cost of getting in and out does not shrink in the same way. The spread you cross and the fees you pay are the same whether the position is held for forty seconds or four days. So the shorter the horizon, the larger the share of the available move that the round trip consumes.
Nothing about that figure depends on a view of the market. It is two numbers and an exponent, and it is worth doing with your own cost and your own instrument before any indicator is chosen. Two things about how to read it. The bars are generous, because they credit the full expected move to every trade, and capturing all of a move is not something any method does, halve the captured fraction and the shortest horizons go negative. And the shape is what matters rather than the levels: the cost is a constant and the move is a square root, so the two lines cross somewhere for any pair of inputs, and where they cross is the only question worth asking first.
It also explains why the activity is viable for some participants and not others. A firm that receives a rebate instead of paying a fee, and that is quoting rather than crossing the spread, has a different cost term, sometimes a negative one. That is a structural advantage rather than a better method, and it is the honest answer to why professional short-horizon trading exists while the retail version of it mostly does not.
Which measures survive a change of timeframe
| Measure | Transfers? | What has to change |
|---|---|---|
| Average true range | Cleanly | Nothing structural, it is expressed in the instrument’s own units and computed on whatever bars you give it. |
| Relative volume | Only with a new baseline | The denominator must be the same time of day across recent sessions. Against a flat daily average it measures the clock. |
| Oscillators with fixed levels | No | Every threshold was calibrated on daily data. Recompute as a percentile of the intraday series, or expect dozens of crossings a session. |
| Anything using the close | Weakly | A daily close is where a session settled; an intraday bar’s close is an arbitrary instant. Measures built on closing direction lose most of their meaning. |
| Breadth measures | Not really | They are counts of issues on a session’s close. An intraday version exists on some platforms and is a different measurement with no published history behind it. |
The pattern in that table is worth naming: what transfers is anything expressed relative to the instrument itself, and what fails is anything carrying a number borrowed from someone else’s sample. That is the same distinction the method page draws between a description and a threshold, and shortening the timeframe is the quickest way to expose it.
The shape of the session, and why it fools volume measures
Intraday volume is not evenly distributed and never has been. The opening and closing periods carry a disproportionate share of the day’s trading, with a pronounced trough in the middle, and the pattern is stable enough to be almost a constant of the market.
Any volume comparison that ignores it produces the same wrong answer every day: the open looks like an extraordinary event, midday looks like a market that has stopped, and the close looks like a surge. None of that is information; it is the timetable. The fix is the one the average volume page sets out for intraday work: compare each interval against the same interval over recent sessions.
Two further distortions belong in the same paragraph, because both are scheduled rather than behavioural. The opening and closing auctions are separate mechanisms whose volume arrives in a single print, so it should not be read as continuous trading at all. And on index rebalance dates the closing auction is enormous for reasons that have nothing to do with that day’s market, a fact worth knowing before an intraday volume reading is called unprecedented.
Where VWAP fits, and why it exists
Volume-weighted average price is the one genuinely intraday measure with a solid reason to exist, and its reason is execution rather than prediction. An institution filling a large order over a day is measured against the day’s volume-weighted price, because that is the benchmark for whether the order was worked well.
That makes it a real reference level rather than a signal, and it is why price often behaves near it: a lot of participants are being measured against that number and are trading in relation to it. Treating it as a forecast reverses the causation, and it is the clearest example on this site of a level that matters because of market structure rather than because of anything technical.
What this page is for
It does not recommend a holding period, and the reason is not squeamishness: a recommendation about holding period would be a claim about expected return, and nothing in this reference supports one.
What is here instead is the part that gets left out. The cost arithmetic above is elementary, rarely written down, and decides the question before any indicator is chosen. The transfer table is a list of which measures keep their meaning when the bars get shorter. And the session-shape warning is the specific error an intraday volume reading makes by default. All three are checkable against your own data, which is the only standard this site holds itself to.
Frequently asked questions
What is scalping?
Trading with a holding period measured in seconds to minutes, aiming to capture a small part of a move many times over rather than a large part of one move. It is a real activity carried on by market makers and specialist firms, and the reason it is difficult for anyone else is arithmetic rather than skill: the cost of a round trip is roughly fixed while the size of the move available shrinks as the holding period gets shorter.
How much does the cost matter?
It decides the whole question, which the figure on this page computes. A move scales with the square root of time, so a one-minute horizon offers a small fraction of a day’s move, while the spread and fees on a round trip are the same as they would be for a five-day position. On the figures used here a round trip consumes about three quarters of a typical one-minute move and around a tenth of a full session’s, and that comparison assumes the entire move is captured, which no method does.
Can indicators be used on one-minute data?
The arithmetic works on any series; the interpretation frequently does not. Every threshold in the published literature was calibrated on daily data, and an oscillator on one-minute bars crosses its levels dozens of times a session, so a rule with a fixed level generates a volume of signals that has nothing to do with what the level meant. Recompute any threshold as a percentile of the series you are actually using, or the number is borrowed from a different measurement.
Which measures do transfer to an intraday timeframe?
The ones that are relative rather than absolute. Range measured in the instrument’s own units, such as average true range on the bars you are using, transfers cleanly. Relative volume transfers if — and only if — the comparison is against the same time of day rather than against a flat daily average. Anything with a published fixed level does not transfer, and neither does anything requiring a settled close, since an intraday bar’s close is an arbitrary moment.
Why does intraday volume need a time-of-day baseline?
Because trading is concentrated at the open and the close. Measured against a flat average of the whole day, every opening half-hour looks extraordinary and every midday period looks dead, on every instrument, every day, so the reading describes the clock rather than the market. The correct comparison is this half-hour against the same half-hour over recent sessions, which removes the shape of the session and leaves what is actually unusual.
What about the opening and closing auctions?
They are different mechanisms and their volume should not be read as continuous trading. An auction concentrates orders into a single price at a single moment, so its volume arrives in one print and reflects a queue that built up beforehand. Including auction volume in an intraday average distorts the first and last buckets of every session, and index rebalance dates make the closing auction enormous for reasons that have nothing to do with that day’s trading.
Is there any edge available at that timeframe?
Yes, for participants with structural advantages: a fee rebate rather than a fee, colocation, and an inventory-management reason to be in the market continuously. Those advantages change the cost term in the arithmetic, which is precisely why the activity is viable for them. Without them the same strategy faces the full round-trip cost on every trade, and the sum on this page is the one to do before anything else.
What does this site recommend?
Nothing: this is a reference rather than an advisory, and a page that recommended a holding period would be making a claim it cannot support. What is here is the arithmetic of the costs, the list of which measures survive a change of timeframe, and the reasons the intraday volume patterns mislead. Those are checkable, and they remain true whatever anyone decides to do with them.