Multi-Timeframe Analysis: Direction First, Then Location

Different trends on different timeframes are normal

It is common for the same coin to be rising on the daily chart and falling on the 1-hour chart. Just as one 4-hour candle bundles four 1-hour candles, a higher-timeframe candle bundles several lower-timeframe candles and smooths away the ripples (see the timeframes article). A single pullback on the higher timeframe can look like a downtrend lasting days on the lower one. Part of the difference comes from the participants. People who trade in minutes or hours and people who hold for weeks or months trade the same price for different reasons, and lower timeframes carry more traces of the former, higher ones more of the latter. So asking which timeframe shows the "real trend" has no answer. Multi-timeframe analysis does not try to remove this difference; it gives each timeframe a different question to answer. A common mistake is to assume that when the lower timeframe's trend turns, the higher one will soon follow; most lower-timeframe declines end without leaving a trace on the higher timeframe.

Top down: direction, price zones, location

The usual order is to work down from higher timeframes to lower ones. First, on the highest timeframe, look at the direction in which highs and lows are stacking up and at the slope of the moving averages. Next, on the middle timeframe, mark meaningful price zones such as the latest swing highs and lows or support and resistance. Finally, on the lowest timeframe, see where price currently sits relative to those zones and where the structure is that a stop could be placed against. The higher timeframe answers "which way is the market flowing?" and the lower one answers "where in that flow are we now?" If you reverse the order, the shape you saw on the lower timeframe settles in your mind first, and the higher timeframe tends to become a place to look for evidence that supports it. If the higher timeframe shows no clear flow, that is information in itself. Lower-timeframe signals cannot set a direction then either, so many people simply mark the upper and lower edges of the range.

• Higher (e.g., daily): direction of the flow, slope of the moving averages
• Middle (e.g., 4-hour): key price zones, the latest swings
• Lower (e.g., 1-hour): where price is now, structure for a stop

The ratio between timeframes

A common convention is to space the three timeframes roughly 4 to 6 times apart. Daily to 4-hour is 6x, and 4-hour to 1-hour is 4x. If the ratio is too small (1-hour and 2-hour), the two charts show almost the same thing; if it is too large (weekly and 5-minute), the moves in between drop out and the two no longer connect. Because crypto trades 24 hours a day, one weekly candle is seven daily candles and one daily candle is 24 hourly candles. Which combination fits depends on how long you intend to hold a position, and there is no set answer. A common way to anchor it is to put your intended holding period on the middle timeframe. Someone looking a few days ahead might put the 4-hour in the middle, reading the flow on the daily and location on the 1-hour. Some people change indicator periods along with the timeframe, but it matters to know that the same EMA20 summarizes about a month on daily bars and a little under a day on 1-hour bars.

Higher timeframe: see the direction

The figure shows 60 hypothetical daily candles. The close is above the EMA20 (the 20-bar exponential moving average), the EMA20 is above the EMA50, and both lines are rising. Highs and lows have also stepped higher one after another. On the daily chart this can be summarized as being inside an uptrend. The reason to read both the order and the slope of the moving averages on the higher timeframe is that either one alone struggles to tell a range from a trend. Price can sit above a flat moving average in a market that is just going back and forth, and if price often dips below a rising average the flow can be called weak. The last two candles, however, are bearish. These two days, which look like a small pullback within a long advance on the daily chart, look quite different on the 1-hour chart in the next figure. One or two bearish candles on the higher timeframe do not change the flow by themselves; a close below the latest swing low or the reference moving average is the more widely used test.

Lower timeframe: see where price is

Spread the same two days over 48 one-hour candles and you get a move in which highs and lows step lower one after another, with price below the 1-hour EMA20. Looking at the 1-hour chart alone, it is a shape you could call a downtrend. If you then carry the daily EMA20 over as a horizontal line, you gain information about location: price is currently coming down toward the higher timeframe's reference line. A common way to use this method is to draw the price zones set on the higher timeframe onto the lower one and watch how the lower timeframe's structure changes near them. Keep in mind that the daily EMA20 carried over is its value as of the last closed daily candle. While today's daily candle is forming, the actual daily EMA20 drifts a little, so it is more accurate to treat it as a zone with some width than as a single line. Changes people often watch for on the lower timeframe are the first time a falling high clears the previous high, or long lower wicks forming near the reference line. Neither, by itself, confirms the pullback is over.

Two readings when the timeframes disagree

When the higher timeframe is rising and the lower one is falling, there are usually two readings. One is a pullback (a dip) within an uptrend: the decline on the lower timeframe stops near a higher-timeframe price zone, and the lower timeframe's highs and lows start rising again. The other is the start of a trend change: the decline on the lower timeframe continues until the higher timeframe also closes below its most recent low. The problem is that in the moment you cannot tell the two apart. Even a large decline starts out looking like a pullback on the lower timeframe. As a pullback stretches on, earlier buyers hit their loss limits one by one, and that selling pushes price down further, which is how a pullback can turn into a reversal. That is why the method is sometimes used less to pick an interpretation than to decide in advance where you will admit you were wrong (for example, a daily close below the most recent swing low on the daily chart). The opposite case, higher timeframe falling and lower timeframe rising, is read the same way, watching whether the lower-timeframe bounce stalls near higher-timeframe resistance.

The scores in the Multi-Timeframe Trend Matrix

This site's Multi-Timeframe Trend Matrix checks three conditions on closed candles for each of the 15-minute, 1-hour, 4-hour, daily and weekly timeframes: whether the close is above the EMA20, whether the EMA20 is above the EMA50, and whether the EMA20 is higher than it was 5 bars ago. Each scores +1 or −1, so the total is one of +3, +1, −1 or −3. When the daily and weekly are both up but the 15-minute or 1-hour is down, it is flagged separately as "Pullback in uptrend". The score summarizes where the moving averages sit, so it only shows whether there is a trend right now; it does not mean the trend will continue. When reading it, look at the pattern across timeframes more than the size of the score. If all five cells agree, the flow is evenly one-directional; if higher and lower timeframes split, it is the conflict described in the previous section. +1 and −1 mean the conditions are pulling against each other, which often happens in ranges where the moving averages are tangled. Because only closed candles are used, the weekly cell changes only when a week ends, and sharp moves in between show up first in the lower-timeframe cells.

A common trap: more timeframes make the answer easier

With five or so timeframes open, at any given moment some are rising and some are falling. That makes it easy to pick whichever timeframe agrees with the direction you had already settled on and use it as your evidence. This is confirmation bias at work, hopping from one timeframe to another (see the article on the limits of chart analysis). One way to reduce it is to decide in advance which timeframe plays which role and not change that midway. Also remember that the higher the timeframe, the later its signals. For a change in the weekly trend to be confirmed, several weekly candles have to close, and by then price has already moved a good deal. Another trap is switching timeframes between entry and exit: closing a position taken on the daily flow because of a wobble on the 5-minute chart, or, the other way round, holding on to a losing position taken on 5-minute evidence by pointing to the daily. A common way to reduce this is to set the exit rule on the same timeframe that gave the reason to enter.

What looks different in crypto and stocks

The ratios between timeframes assume a 24-hour market, so they do not carry over to stocks unchanged. Regular sessions in both Korea and the US last six and a half hours, so one daily stock candle is about seven 1-hour candles and one weekly candle is five trading days. Compared with crypto, where a daily candle is 24 hourly candles and a weekly candle is seven daily ones, the same "daily and 1-hour" pair is much closer together. Stocks also gap, because news from while the market was closed piles into the first bar of the next day, so the first 1-hour bar is often unusually large and the bars after it relatively calm. Two days that look like a clean pullback on the daily chart of Samsung Electronics or SK Hynix may, on 1-hour bars, consist of one gap and some sideways drift. In crypto, each exchange has a time at which it cuts the daily candle (many use 00:00 UTC, 9 a.m. in Korea), and the daily candle closes at that time. For US indices and large tech stocks, the shape of lower-timeframe bars depends on whether extended-hours trading is included, so it is worth checking which kind of chart you are looking at.

Checklist in practice

The most confusing part of stacking timeframes on a live chart is that the higher-timeframe candle is still forming. Today's daily candle changes all day, and this week's weekly candle changes until the weekend. While a few lower-timeframe bars close, the higher timeframe's alignment can flip on screen and flip back, so it is more consistent to judge the higher timeframe on its last closed candle and treat the one in progress as reference only. That is also why this site's trend matrix scores only closed candles.

• Decide the three timeframes and the role of each in advance
• Note the flow on the highest timeframe as of its last closed candle
• Mark the latest swings and key price zones on the middle timeframe
• Carry those zones onto the lower timeframe and see where price is now
• When they conflict, do not pick pullback or reversal; set the price at which you would call it wrong
• Set the exit rule on the same timeframe that gave the reason to enter