Multiple Time Frames

Every chart you open shows one time frame, but no trade ever happens in just one. A 15-minute scalper still glances at the hourly trend before pressing the button; a swing trader checks the weekly before acting on the daily. That "glance up" is what professionals call multiple time frame analysis, and AmiBroker lets you bake it directly into a formula - so your system can see the bigger picture without you flipping charts.

The idea is powerful and a little dangerous in equal measure. Done right, a higher time frame becomes a calm trend filter that keeps you on the correct side of the market. Done carelessly, it quietly leaks future information into your signals and produces a backtest that can never be repeated. This chapter teaches both the tools and the discipline.

Why look at a higher time frame

The same EMA crossover that looks like a great buy on the 15-minute chart can be a doomed counter-trend bet when the hourly is rolling over. A higher time frame answers a question the lower one cannot: which way is the larger tide flowing? The classic professional rule is simple - take lower-time-frame entries only in the direction of the higher-time-frame trend. We will build exactly that.

AmiBroker keeps your chart on its base interval (say, 15-minute) and lets you temporarily "switch up" to compute something on a coarser interval, then bring the answer back down. Three functions do all the work.

TimeFrameSet and TimeFrameRestore

TimeFrameSet(interval) compresses the price arrays to a higher interval. Everything you calculate after it - moving averages, RSI, anything - is computed on those bigger bars. TimeFrameRestore() switches you back to the chart's native interval. Always use them as a matched pair, like opening and closing a bracket.

// Switch up to the hourly bars
TimeFrameSet(inHourly);
    hourlyEMA = EMA(Close, 20);   // 20-period EMA of HOURLY closes
TimeFrameRestore();               // back to the chart's own interval

AmiBroker gives you ready-made interval constants: inDaily, inWeekly, inMonthly, inHourly, and in5Minute, in15Minute and friends. You can also build any custom interval in seconds with inDaily * n style arithmetic, but the named constants cover almost everything.

TimeFrameExpand: bringing the value back down

There is a catch. After TimeFrameRestore(), the hourlyEMA array has one value per hourly bar, but your chart has many 15-minute bars per hour. To plot or compare it against 15-minute price, you must stretch it back out so every 15-minute bar carries the right hourly value. That is what TimeFrameExpand does.

TimeFrameSet(inHourly);
    hourlyEMA = EMA(Close, 20);
TimeFrameRestore();

// stretch the hourly series back onto the 15-minute chart
hourlyEMA_exp = TimeFrameExpand(hourlyEMA, inHourly);

Plot(Close, "Price", colorDefault, styleCandle);
Plot(hourlyEMA_exp, "Hourly EMA(20)", colorYellow, styleThick);

The third argument controls how a higher-time-frame value fills the smaller bars beneath it, and the choice matters:

• expandLast (the default) - each finished higher-TF bar's value appears on the lower bars after it closes. This is the honest, non-repainting choice for trading.
• expandFirst - the value appears from the start of the higher bar. Convenient for display, but it shows the hourly close on 15-minute bars that formed before that hour finished - look-ahead.
• expandPoint - the value is placed on a single bar only, leaving the rest empty.

TimeFrameGetPrice: a prior higher-TF value

Sometimes you just want a single higher-time-frame price - yesterday's daily close, last week's high - without setting and restoring. TimeFrameGetPrice is the shortcut:

// the PRIOR completed daily close, available on every intraday bar
prevDayClose = TimeFrameGetPrice("C", inDaily, -1);

// today's developing daily high (use with care - it is still forming)
todayHigh    = TimeFrameGetPrice("H", inDaily,  0);

The first argument is the field ("O", "H", "L", "C" or "V"), the second is the interval, and the third is the offset - 0 is the current higher bar, -1 is the one before it. The -1 is the safe one: the prior day has finished, so its close is fixed. Reaching for 0 on the current day means you are reading a value that is still changing, which is fine for a live readout but rarely safe inside a signal.

A higher-time-frame trend filter

Now we put it together into the professional rule: trade the 15-minute chart only in the direction of the hourly trend. The lower time frame gives the entry; the higher time frame gives permission.

// --- Higher time frame: hourly trend ---
TimeFrameSet(inHourly);
    hrEMA = EMA(Close, 20);
TimeFrameRestore();
hrEMA_exp = TimeFrameExpand(hrEMA, inHourly);   // expandLast by default

hourlyUp   = Close > hrEMA_exp;                 // hourly trend is up
hourlyDown = Close < hrEMA_exp;                 // hourly trend is down

// --- Lower time frame: 15-minute entry trigger ---
fast = EMA(Close, 9);
slow = EMA(Close, 21);

// Only take the 15-min cross in the hourly trend's direction
Buy   = Cross(fast, slow) AND hourlyUp;
Short = Cross(slow, fast) AND hourlyDown;

Sell  = Cross(slow, fast);     // exit long on the opposite 15-min cross
Cover = Cross(fast, slow);     // cover short on the opposite cross

Buy   = ExRem(Buy, Sell);
Sell  = ExRem(Sell, Buy);
Short = ExRem(Short, Cover);
Cover = ExRem(Cover, Short);

The hourly EMA filter throws away every 15-minute crossover that fights the bigger trend. You will take fewer trades, but the ones you take lean with the larger flow instead of against it - the multi-timeframe version of the trend filter we used for RSI pullbacks last chapter.

Pitfalls and alignment

Multi-timeframe code has a few sharp edges worth memorising:

• Always pair Set and Restore. Unmatched calls silently corrupt the rest of the formula.
• Expand before you compare or plot. A raw higher-TF array does not line up with your chart bars until you TimeFrameExpand it.
• Prefer expandLast and -1 offsets for anything a signal depends on, so you never act on a higher-TF bar that has not closed.
• Mind the data. To compute an hourly or daily value you must actually have enough intraday history loaded; thin data makes the higher series ragged or empty.

Try it yourself

• Plot an hourly EMA(20) on a 15-minute BANKNIFTY chart and watch how it steps once per hour.
• Add prevDayClose = TimeFrameGetPrice("C", inDaily, -1) and plot it as a dashed line - your intraday pivot reference.
• Run the multi-timeframe crossover as a backtest with and without the hourlyUp / hourlyDown filter and compare trade counts.
• Switch the expand mode to expandFirst and look closely - can you see the line bending inside an hour? That is the look-ahead you are avoiding.

Recap

• Multiple time frame analysis lets a lower-time-frame system see the higher-time-frame trend.
• TimeFrameSet / TimeFrameRestore switch the calculation interval up and back - always as a pair.
• TimeFrameExpand stretches a higher-TF array back onto your chart; use expandLast for trading.
• TimeFrameGetPrice("C", inDaily, -1) grabs a single prior higher-TF value safely.
• The professional rule - enter on the lower time frame only in the higher time frame's direction - filters out trades that fight the bigger tide.

Next we make a system speak up - firing alerts to a Telegram channel and even out loud the moment a signal triggers.