Multi-Timeframe Analysis on MT4: M15 Over M5 Was the Only EURUSD Pair That Helped, and on Gold the Filter Won 113 of 144

My confluence test showed that on EURUSD an honest higher-timeframe filter was close to a coin flip, and the comparison with a slow moving average showed that an H1 filter is really just a long EMA. This guide turns that into settings you can use in MetaTrader, and adds gold, where the result was very different.

How do you set up multi-timeframe analysis in MT4?

There are three ways. All work the same in MT5, which also has more timeframes to choose from, as I covered in MT4 vs MT5.

1. Two windows. Open the pair twice (File → New Chart), set one to M5 and one to M15 or H1, then Window → Tile Vertically. Add Moving Average (Insert → Indicators → Trend → Moving Averages), method Exponential, period 50, to the higher one. You trade the M5 and only look at whether the higher chart’s price is above or below its EMA.
2. One chart, longer average. Multiply the period by the timeframe ratio: M15 EMA 50 ≈ M5 EMA 150, H1 EMA 50 ≈ M5 EMA 600, H4 EMA 50 ≈ M5 EMA 2400. On EURUSD, M5 EMA 600 and H1 EMA 50 agreed on 93.3% of bars.
3. An MTF indicator that draws the H1 average on the M5 chart. Convenient, but check how it treats the unfinished bar (below).

Save the finished layout with right-click → Template → Save Template, as in my multi-indicator template guide, so every new chart opens the same way.

Which timeframe pair should you use?

I tested four entries on EURUSD M5 — EMA 9/21 cross, MACD 12/26/9 signal cross, RSI 14 leaving 30/70 and a Stochastic 14/3/3 cross in the 20/80 zone — with a filter on M15, M30, H1 or H4, EMA 20, 50 or 100, and exits after 1, 2 or 4 hours. 16,926 bars, 1 pip of cost per trade, the closed higher-timeframe bar.

The common rule of “one timeframe four to six times higher” points to M30 or H1 for an M5 trader. On this data, the nearer M15 was better and H4 was worst. For trades that last one to four hours, the trend that matters is the last half-day, not the last week.

Here is the EMA 50 filter per entry with the 2-hour exit:

Only the Stochastic cross with the M15 filter finished positive, over 234 trades, and it held in both halves (+0.57 and +0.38). On EURUSD a trend filter trimmed losses more often than it created a winner.

Does the higher-timeframe filter work better on gold?

Yes, clearly. I ran the identical grid on 13,736 gold (XAUUSD) M5 bars, with a cost of $0.30 per trade, the same assumption as in my gold session test. Results are in dollars per ounce per trade.

On gold the filter did what the textbooks promise, and it did it for the trend entries — the opposite of EURUSD, where it helped only the Stochastic fade. Gold’s EMA cross was already positive without any filter, which tells you gold moved in longer, cleaner swings in this period. A trend filter can only help when there is a trend to follow; I saw the same split between the two markets in gold vs EURUSD with one rule.

The honest caveat: this is ten weeks of one metal. If gold starts to range, the filter will behave like it did on EURUSD.

How do you check whether an MTF indicator repaints?

A multi-timeframe indicator has to decide what to do with the higher-timeframe bar that has not closed yet. If it redraws its past values when that bar closes, its history will look better than it ever traded. On EURUSD that effect alone raised the profitable setups from 4 to 32 of 144, and on gold from 73 to 112.

1. Attach the indicator to an M5 chart and take a screenshot at, say, 10:25.
2. Take another at 11:05, after the H1 bar has closed.
3. Compare the M5 bars between 10:00 and 10:25. If the colour or arrows changed, it repaints.

In an EA, read the higher timeframe with shift 1, for example iMA(NULL, PERIOD_H1, 50, 0, MODE_EMA, PRICE_CLOSE, 1), which uses the last closed H1 bar. Shift 0 is the live bar — honest in real time, but a backtest in the MT4 Strategy Tester models it correctly only with “Every tick”. The full procedure is in my repainting arrows test and how to backtest an EA on MT5.

What settings would I start with?

• EURUSD M5: M15 EMA 50 (or M5 EMA 150), and only on dip-buying entries like a Stochastic cross. Do not expect it to fix trend entries.
• Gold M5: H1 EMA 50 (or M5 EMA 600) on EMA-cross or MACD entries, closed bar.
• Skip H4 as a filter for M5 on EURUSD. It was the worst pair here.
• One filter only. Requiring H1 and H4 to agree was worse than H1 alone for three of four EURUSD entries.

Frequently asked questions

How do I show two timeframes on one chart in MT4?

Either open two windows and tile them, or put a longer average on the lower chart: EMA period × timeframe ratio. An H1 EMA 50 equals roughly an M5 EMA 600; they matched on 93.3% of EURUSD bars.

What is the best timeframe combination for scalping?

On EURUSD, M5 with an M15 filter: it improved 22 of 36 setups. On gold all four pairs helped, H4 the most often (31 of 36).

Does multi-timeframe analysis work on gold?

In this test, yes. The closed-bar filter beat no filter in 113 of 144 gold setups, and an H1 EMA 50 took the EMA 9/21 cross from +$0.38 to +$1.51 per trade after a $0.30 cost.

Should my EA use shift 0 or shift 1 for the higher timeframe?

Shift 1 (the last closed bar) cannot repaint and backtests the same in every tester mode. Shift 0 is fine live but needs tick-accurate testing to give honest history.