Choose the Frame Before the Trade

Why the same move looks different at different distances

A price series has a fractal quality: the shape repeats at every scale, but the amount of randomness inside a given percentage move does not. At the five-minute scale, a 1% move can be one institutional order, a news headline being digested, or the mechanical flow from an expiring option — three unrelated causes that look like one trend on the chart. At the weekly scale, a 1% move is a rounding error, and the move worth studying is 10%. The practical version is a signal-to-noise ratio that improves as the frame lengthens: a daily chart compresses six and a half hours of argument into one decision, and a weekly chart compresses five of those into one, so the pattern a learner sees on the weekly is made of far more evidence than the pattern they see on the five-minute. This is why the frame is a risk decision rather than a preference. A trader who analyses on five-minute candles has chosen an instrument where most of the movement is uninformative, so their stops are hit by flow and their win rate is dominated by noise; the same trader on the daily is trading slower signals but cleaner ones. The question is not which chart is better in the abstract — it is which frame matches the horizon over which the position will be held. If the plan is to hold for weeks, the daily is the frame where the position’s ordinary movement lives, and a five-minute wiggle is not information about it; if the plan is to hold for hours, the five-minute is the operating frame and the daily is context. Then the two-frame habit, which is the discipline this lesson installs. Every trade is analysed on a context frame — where the trend and the levels are — and timed on an execution frame, where the entry is taken. Day trader: daily and 60-minute context, 5-minute execution. Swing trader: weekly context, daily execution. Position trader: monthly context, weekly execution. The pairing is deliberate: the execution frame is never used to decide anything about direction, and the context frame is never used to pick an entry price. Without that split, a learner drifts into whichever chart agrees with them, which is the next failure mode. Frames by horizon — Minutes to hours: Context: daily and 60-minute · Execution: 1–5 minute · Days to weeks: Context: weekly · Execution: daily ← · Months to years: Context: monthly · Execution: weekly · Deciding direction: The higher frame only — never the lower one ← Aggregation is not free information. A weekly candle contains the daily candles inside it, so the weekly never says something the daily contradicted — it says which of the daily moves mattered. Reading both is reading the same data at two resolutions, which is exactly why the two-frame rule works.

Frame conflict, and the honesty of standing aside

The two frames disagree routinely: an instrument in a weekly downtrend will contain dozens of daily rallies, and each one looks like an opportunity on its own frame. The rule that resolves this is priority, and it is a rule rather than a judgement call because the alternative — deciding case by case — is where discretion becomes rationalisation. The higher frame owns direction; the lower frame owns timing; if the lower frame conflicts with the higher and there is no specific reason (a level, a completed reversal on the higher frame, a fundamental change), the correct action is no trade. Standing aside is not a failure of nerve. It is the arithmetic of the previous paragraph: the counter-trend move is real, its expectancy is worse than the noise it is made of, and declining it is a decision with a positive expected value. There is a second failure mode that looks like diligence and is not: frame-shopping. The habit is to pick a chart, see that it disagrees with the idea, and switch to another frame until one agrees — five-minute for the entry, then daily to justify holding when the entry goes wrong, then weekly to justify holding longer, then back to the five-minute to argue that the bounce is coming. Every frame contains both directions somewhere, so this process can always find confirmation, and the confirmation is worth exactly nothing because the frame was selected by its answer rather than before the analysis. The defence is mechanical and it is the sentence to remember: the frames are chosen before the trade or not at all. • Context frame sets direction and levels; execution frame sets the entry and the stop. • Frame conflict with no specific reason means no trade — that is a decision, not cowardice. • Choose frames from the holding period, in writing, before looking at a chart. • Never switch frames to justify a position already held: every frame will oblige you. • A lower-frame move never overrides the higher frame; it can only time an entry in its direction. A subtle version of frame-shopping: keeping the same chart but changing the timeframe drop-down until the indicator or pattern appears. The frame is the same object regardless of its label, so a learner who “finds” the setup at a different interval has usually found the interval, not the setup — and the correct response to a setup that only appears at one arbitrary interval is to treat it as noise.

The arithmetic that links two frames

Charts at different intervals are not different opinions about a market; they are the same price series aggregated over different windows, and the relationship between them is arithmetic rather than interpretive. A four-hour bar contains four one-hour bars, and its open is the first of them while its close is the last. That containment is what makes a higher frame authoritative about direction: a weekly close cannot be changed by anything that happens inside the week, so a conflict between a weekly downtrend and a daily rally is resolved when the week closes, not when the intraday looks convincing. This lesson’s rule that the higher frame holds the direction is a statement about aggregation, not about which chart is more powerful. The second relationship is statistical and it explains why the same percentage move means different things at different distances. Price movement scales roughly with the **square root of time**, so the typical range of a four-hour bar is about twice the typical range of a one-hour bar, not four times. A consequence worth carrying: a two-percent move inside an hour is extraordinary and a two-percent move across a week is unremarkable, and a stop sized from the daily range is a wider stop than the same percentage-sized stop on the weekly. Anyone who has wondered why a stop that looked generous on a five-minute chart was hit immediately has met this arithmetic without naming it. The third is a practical choice about how much history to keep on the screen. A moving average needs enough bars to be meaningful, so a two-hundred-period average on a daily chart is nearly a year of data while the same setting on a five-minute chart is about two days — which is why indicator settings do not transfer between frames and why a crossover signal means something completely different on each. When two frames are chosen, each should be read with settings appropriate to it, and the higher frame should carry enough history to contain a full cycle of the pattern you are relying on. Charts that show three weeks of hourly data cannot speak about a trend, no matter how suggestive the shape looks. One market, two frames, three consequences — A 4-hour bar contains four 1-hour bars: The higher close is fixed by the lower ones — it holds the direction · Range scales with the square root of time: Four times the duration is about twice the typical range · Indicator settings do not transfer: A 200-period average is a year on daily and two days on five-minute ←

The frame is a data decision, not only a clock

Two traders can both say they trade “the daily chart” and be looking at series that disagree, because a bar is a convention before it is a period. For a stock, the choice is the regular session of 09:30 to 16:00 against a twenty-four-hour bar that includes pre-market and after-hours prints; for a future, it is the exchange’s floor session against a nearly continuous overnight one. Add half-days, exchange holidays that differ between countries, and the treatment of the opening auction, and “daily” resolves to a dozen different objects. The differences are not cosmetic. A twenty-four-hour bar contains the overnight range, so it prints a wider average range than the session bar on the same instrument, which means a stop sized from one convention is a different stop on the other. And a gap that is real on a session chart — the two percent jump between yesterday’s close and today’s open — is simply filled inside the overnight session on a twenty-four-hour chart, which is why two traders can argue about whether a gap held and both be describing their own data accurately. The second data decision is adjustment. Equity series are usually back-adjusted for splits and dividends, which preserves return continuity at the cost of making past prices differ from the prices that actually printed, so a level drawn on adjusted data is not the level a participant saw. Futures series are stitched from rolling contracts and back-adjusted to remove the roll’s price jump, which is why a five-year-old chart can show a price the contract never traded. The unadjusted alternative avoids the fiction and introduces a different one: a raw series breaks at every ex-date and every roll, so a moving average or a support level computed across the break measures the adjustment rather than the market. Neither convention is wrong; a trader who does not know which one their platform is using is drawing levels on data they have not inspected. The operational consequence is the sentence worth keeping: choose the frame’s data convention with the frame itself, write both down, and measure every trade on the frame the plan was written for. The temptation is to review with whichever chart is open, and the result is a statistic that describes the wrong series. A two-week swing trade graded on five-minute bars shows a maximum adverse excursion several times its intended stop and a win rate that looks like failure, because the trade is being judged on the frame where its ordinary movement is noise — the exact point the lesson opened with, now applied to the journal instead of the entry. R multiples that were earned on the daily and reported from the five-minute chart are not conservative, they are mis-specified, and a trader who abandons a working system on the strength of that number has been defeated by a chart setting. Two conventions, two series, one instrument — Session bars 09:30–16:00: Gaps are visible; the average range is narrower · Twenty-four-hour bars: Overnight range included; the gap is filled off-screen ← · Split- and dividend-adjusted prices: Return continuity, but historical levels are restated · Unadjusted prices: Real prints, but every ex-date breaks the series ← The check is mechanical and takes a minute: put the session chart and the twenty-four-hour chart side by side and compare the last month’s range. If the average candle differs by a fifth, the stop distance and the position size computed from it differ by a fifth too, and the plan is only valid on the series it was built on.

Sources

• Multiple-timeframe analysisElder, “Trading for a Living”
• Noise and aggregation across horizonsMandelbrot & Hudson, “The (Mis)Behavior of Markets”
• Fractal structure of price seriesStandard quantitative finance treatments of scaling