Configuring the Ichimoku EA for USDJPY: Every Indicator Value

An Ichimoku EA is an Expert Advisor, meaning a program that runs inside MetaTrader and places orders automatically according to conditions built from the Ichimoku indicator. The USDJPY build documented here executes on the M5 chart and reads three inputs: Ichimoku with values of 9, 34 and 72, Williams Percent Range at a period of 180, and MACD on two higher charts as confirmation. Every value below comes from the original setup by Petko Aleksandrov, founder of EA Academy. This page covers the configuration in full. It does not cover the entry and exit conditions, which were documented separately, and it says so plainly rather than implying otherwise.

Indicator	Chart	Values used here	Platform default
Ichimoku	M5	Tenkan-sen 9, Kijun-sen 34, Senkou Span B 72	9, 26, 52
Williams Percent Range	M5	Period 180	Period 14
MACD	M30	12, 42, 9	12, 26, 9
MACD	H1	12, 14, 9	12, 26, 9
Instrument	USDJPY	Executes on M5	Not applicable

What This Automation Actually Is

Three terms describe the same thing, and the overlap confuses newcomers. EA is short for the label MetaTrader itself uses for an automated program attached to a chart. Traders shorten that to EA. Marketing material calls the same object a trading bot or a robot. So an automated forex robot built on those lines and the program discussed here are one concept wearing several labels.

What the automation does is narrow and worth being precise about. It watches the conditions you defined, and when they are all satisfied it sends an order without asking. It does not think, adapt or improve, and it will keep placing losing orders indefinitely if the conditions it encodes stop working. Execution is mechanical; the judgement was yours, made earlier, when you specified those conditions.

That distinction matters for anyone comparing automation approaches. A grid system, a breakout system and a momentum system are all EAs, and they differ entirely in what they encode. Our overview of the main types of EA logic sets out how those families behave differently, and what an EA is in forex covers the mechanics of automation itself in more depth than there is space for here.

Where the Indicator Came From

Ichimoku Kinko Hyo was the work of Goichi Hosoda, a Japanese financial journalist who worked on it from the late 1930s and published the method in 1969 after roughly three decades of refinement (Wikipedia, “Ichimoku Kinkō Hyō”, accessed September 2026; Swissquote, “What is the Ichimoku trading method?”). The name translates loosely as a one-glance equilibrium chart, which describes the design goal: read direction, support and resistance from a single view.

Five lines make up the display. Tenkan-sen and Kijun-sen are the fast pair, drawn in red and blue respectively on a default MetaTrader chart. Senkou Span A and Senkou Span B form the shaded band between them, which most material calls the Ichimoku cloud. Chikou Span plots the close shifted backwards.

The best-known reading treats that shaded band as a support and resistance zone: price trading above it means the band tends to act as support, and price below it means the band tends to act as resistance. That is the conventional interpretation rather than a rule this particular build relies on.

The Values This Build Uses

None of the three indicators here runs on its factory configuration, and the gaps are large enough to be deliberate rather than incidental.

The Trend Lines at 9, 34 and 72

MetaTrader ships this indicator with 9, 26 and 52, the periods Hosoda published. This build keeps the first at 9 and lengthens both of the others: Kijun-sen from 26 to 34, and Senkou Span B from 52 to 72. Longer periods produce slower, smoother lines that react less to short bursts of movement, which is a reasonable direction to move when the execution chart is as fast as M5. The Kijun-sen is the line the system reads.

Williams Percent Range at 180

Here the change is dramatic. Williams %R defaults to 14, and this configuration uses 180, nearly thirteen times longer. An oscillator stretched that far stops behaving like a short-term overbought and oversold gauge and starts behaving like a slow positioning measure across a much wider window of price history.

The original setup keeps the -20 and -80 level lines on the chart for visual reference while noting that the system does not read those levels. That is a useful detail, because it tells you the oscillator is contributing something other than a threshold cross.

MACD on Two Charts

M30 runs MACD at 12, 42 and 9. H1 runs it at 12, 14 and 9. Both differ from the 12, 26, 9 default, and they differ from each other, so the two charts are not duplicating one measurement at different speeds. The slow line at 42 on M30 produces a considerably more sluggish reading than the 14 on H1, which is an unusual arrangement and one of the more distinctive things about this build.

Why Two Higher Charts at All

Execution happens on M5. The M30 and H1 charts contribute nothing to timing and exist to filter, which is the standard reason for multi-timeframe construction in forex trading: a fast chart generates far more candidate setups than are worth taking, and a slower reading of the same instrument discards some of them.

Signal frequency is the quantity being traded away here. Each additional condition removes candidates, and on an M5 chart that reduction can be severe, turning a system that would have traded several times a day into one that trades a few times a week. Fewer entries mean less exposure to spread and slippage, which is the benefit. They also mean a thinner record from which to judge anything, which is the cost, and it is a cost that only becomes visible once you try to evaluate the results.

Not every build in the same course uses this arrangement. According to the original documentation, the confirmation charts apply to the GBPUSD and USDJPY systems only, while the remaining systems run on a single chart. That is a meaningful detail for anyone assuming multi-timeframe filtering is always an upgrade. It is a design choice with costs: more conditions mean fewer trades, and fewer trades mean a smaller sample from which to judge whether anything works.

If you want the broader argument for how multiple timeframes fit together on this instrument, our USDJPY multi-timeframe framework covers it properly.

What This Page Does Not Tell You

Straight answer: the entry trigger, the exit condition, and exactly what the Kijun-sen and the Williams %R readings must do before an order is placed.

Those conditions were covered in a separate lesson and are not reproduced here. Publishing a guess would be worse than publishing nothing, because a plausible-sounding reconstruction of someone else’s rules is indistinguishable from the real thing until money is on the line.

What you can take from this page is the configuration, which is genuinely useful on its own: it tells you which indicators are involved, which chart executes, how far the values sit from the defaults, and what kind of behaviour those choices imply. If you are building your own Ichimoku automation, that is a starting configuration you can test against your own conditions.

Test It Before Trading It

Any automation built from an optimised configuration deserves scepticism, and values this far from the defaults are exactly the kind that reward checking. Work through the following before a live account is involved:

1. Evaluate on data the configuration never saw. Develop on one portion of history, hold the rest back, and compare.
2. Move the values slightly. If 34 works and 33 and 35 collapse, the number is describing your data file rather than the market.
3. Shift the start date. Results that depend on where the measurement began are not results.
4. Degrade execution assumptions. Widen the spread beyond your broker’s average and add slippage, which matters disproportionately on an M5 system.
5. Compare live behaviour against the simulation on a demo account before funding anything.

Sample size deserves its own line here. A configuration filtered by two higher charts will produce relatively few entries, and a record of thirty or forty trades tells you very little about whether an edge exists, however good the equity curve looks. Judge the length of your evaluation by the number of trades it produced rather than by the number of months it covered, and be sceptical of any result built on a handful of them.

Our guide to walk forward analysis covers the first point in depth, and why simulated results differ from live ones covers the fourth. Neither process guarantees anything. Both reduce the chance that an impressive equity curve is an artefact of the search that produced it.

Loading It Onto the Platform

Installation works the same way for any EA. Copy the .ex4 or .mq4 file into the MQL4/Experts folder of your MetaTrader data directory, restart the terminal, and the program appears in the Navigator panel. Drag it onto a USDJPY M5 chart, enable AutoTrading, and confirm the smiling face appears in the top right corner of the chart. Without that confirmation, the program is loaded but inert.

Latency matters more for a system reading three charts than for one reading a single chart, since each filter has to be current at the moment the fast chart produces a candidate. Traders running automation continuously tend to move it onto a hosted server rather than leaving a home machine on overnight, which removes the risk of a dropped connection mid-position. That is a practical consideration rather than a performance upgrade.

Two setup notes specific to this build. The higher-timeframe data must be available locally, so open the M30 and H1 charts once and let them download history before you rely on any filter reading them. And if you do not have the terminal yet, our MetaTrader 4 download and install guide covers the process on both Windows and Mac.

Honest Limits

This configuration was documented in 2020, on one instrument, by one person, with no published performance record. Treat it as a worked example of how a multi-indicator automation gets assembled, not as a recommendation. Market conditions on USDJPY have changed considerably since then, and no indicator values survive changing conditions automatically. The yen has been through several distinct regimes in that time, including intervention episodes and a sustained shift in rate differentials, none of which the values here were calibrated against.

That is not an argument against the approach. It is an argument for treating any published configuration as a starting point for your own testing rather than as a finished product someone has handed you.

FAQs

What does EA stand for?

It stands for Expert Advisor, the label MetaTrader applies to any program that runs on a chart and places orders automatically for you. The one described here draws its conditions from a single trend indicator plus two higher-timeframe filters. You will see the same object called a robot or a bot in marketing material, and all three words describe identical software. The name itself says nothing at all about quality, since anyone can compile a set of conditions into a file and distribute it.

Can these lines be automated at all?

Yes. Every line in the display is a defined calculation on past highs, lows and closes, so any condition you can describe in words about them can be written into code. The difficulty is never the automation itself. It lies instead in specifying conditions precise enough to be programmed, since much conventional reading of this indicator depends on judgement about market context that no program can exercise on your behalf.

Why does this setup use 9, 34 and 72 instead of the standard values?

The platform ships this indicator with 9, 26 and 52, the periods Goichi Hosoda published in 1969. Lengthening the second and third values to 34 and 72 produces slower lines that react less to brief moves, which suits a fast execution chart. The trade-off is responsiveness: slower lines confirm later. Values that far from the defaults usually emerge from an optimisation process, so they deserve the sensitivity testing described above before anyone trusts them.

Does this EA work on instruments other than USDJPY?

The configuration here was built for one instrument, and running it unchanged elsewhere is a test rather than a deployment. Values tuned on USDJPY may transfer to related currency pairs, which is worth checking as evidence, though failure elsewhere does not automatically condemn a system designed for a single symbol and traded only there. Gold and index instruments behave differently enough that Ichimoku values calibrated on a currency will rarely carry across without recalibration.

What timeframe should an automated system like this trade?

This build executes on M5 and reads M30 and H1 as filters, but no timeframe is correct in general. Faster charts generate more signals and suffer more from spread and slippage; slower charts produce fewer opportunities and absorb execution costs more comfortably. The sensible approach is to pick the chart that matches the behaviour you are trying to capture, then confirm through testing that the results are not an artefact of that particular choice.

Is a free EA download safe to run?

Treat any downloaded program of this kind as unverified until you have tested it yourself. Compiled .ex4 files cannot be inspected, so you cannot see what conditions they encode or what position sizing they apply. Run anything new on a demo account first, watch how it behaves around news and session boundaries, and check whether it uses averaging methods such as martingale that can hide losses for a long time before revealing them.

How do I know whether the values are over-optimised?

Watch for four signals. Performance falls apart when you add more history. Small changes to the indicator values cause large changes in output. Results depend heavily on where the evaluation started. And behaviour does not carry to comparable instruments. Any one of those is a caution; two together usually mean the values describe the sample they were fitted on rather than something the market repeats, at which point rebuilding beats adjusting.

Do I need to understand the indicator to run an automation built on it?

You can run one without understanding it, and you will regret that the first time it behaves unexpectedly. Knowing what each line measures tells you which conditions the program is likely reading and what kind of price behaviour will confuse it. That knowledge is what lets you decide whether a losing run reflects normal variance in live trading or a genuine breakdown, which is a judgement no software will ever make on your behalf.

Working through a configuration like this one and unsure whether the values hold up, or whether an automation you have been handed is worth running live? Get in touch with Algo Trading Space and tell us what you are testing. We will point you toward the checks that matter for your setup rather than toward another download.

Trading foreign exchange carries a substantial risk of loss and is not suitable for every investor. Nothing here is financial advice, and past behaviour of any automated system is not a reliable indicator of future outcomes.