Scalper XAUUSD EA Free Download: 7 Truths to Check

You are about to install a Scalper XAUUSD EA free download, and the first useful thing I can tell you is that the download itself is not the decision. The decision is whether seven specific things are true of the file and your account, and most of them can be checked in the time it takes to read this page. If they are true, you have a known object worth testing. If any of them is false, you have saved yourself a live account’s worth of learning.

Gold scalping attracts free EAs because it looks simple: small moves, high frequency, automation. In practice it is the most cost-sensitive style in retail trading. A scalper does not need to be right very often; it needs its average win to clear the spread and commission it pays on every single trade. On XAUUSD, that cost is large and it is not constant — it widens exactly when volatility rises and the strategy most wants to trade.

So this page is not a list of seven robots. It is the seven truths that decide whether any gold scalper deserves your account: the file’s provenance, its stop logic, the spread it needs, the drawdown it produces, the timeframe it suits, how its position size is chosen, and what it does when the market gaps. Then it gives you a free, MIT-licensed gold EA for MT5 whose source you can read, so the warning about adding to losers stops being abstract.

Truth one: the file’s provenance matters more than its strategy

Start with who made it, because everything else depends on being able to trust the description. A real download names a publisher, states a licence with a working link, and either ships source or lives in a public repository. An anonymous compiled binary gives you none of that, and an expert advisor is a program that can open, modify and close positions on your live account.

Run four checks and stop if any fails:

Check	Pass	Fail
Publisher	A named author or organisation	No name, or a throwaway handle
Licence	MIT, GPL-3.0, Apache-2.0, or stated freeware terms, with a link	“Free” and nothing else
Source	Readable MQL4 or MQL5 code, or a public repository	A compiled binary only
Drawdown	A maximum equity drawdown over a stated trade count	A win rate and a rising curve

This takes five minutes and it eliminates most of what a search engine will show you for a “free XAUUSD scalper”. The reason to check provenance first, rather than the strategy, is that you cannot evaluate a strategy you cannot read. A robot whose logic is hidden is not a candidate; it is an unknown.

Truth two: the stop logic is the strategy

On a gold scalper, how it loses is more important than how it wins, because losing is what ends accounts. Open the source or the inputs and confirm two things: that a stop-loss is set on every position, and that there is no recovery logic of any kind.

You do not need to read MQL to screen for this. Search the source for the order function that opens positions, and for words like LotMultiplier, GridStep, Martingale or Averaging. A stop that is genuinely set will appear as a stop-loss level applied after the order opens. Recovery logic appears as a growing position count or a multiplying lot size, and both mean the EA is hiding losses in floating exposure rather than accepting them.

Gold is the worst possible instrument for recovery logic, because it trends harder and further than a currency pair. The LBMA and CME price drivers — inflation data, employment data, rate decisions, the dollar index — can push gold sharply within a single session. A grid built for small steps meets a large directional day, and the arithmetic that looked fine in a range becomes the mechanism of the loss. Our best gold scalper EA review treats the absence of grid and martingale as the first test a scalper must pass.

Truth three: the spread it needs decides whether it works

A scalper’s edge is not the signal; it is the signal minus the cost of taking it. On gold, that cost is the round-trip spread plus commission, charged on every trade, and it is the reason the same robot can be positive on one broker and negative on another.

Here is the arithmetic as it actually plays out. On a raw account in a calm London hour, XAUUSD might quote a round-trip spread around 15 to 30 points. Around a US inflation release, the same symbol can quote several times that while price is still moving, because liquidity thins when everyone wants the same side. A scalper targeting a modest range pays that cost on every trade, so the number of trades multiplies the tax. A robot that fires fifty times a month pays it fifty times.

That is why the maximum spread input is a risk feature, not a nicety. Set it to a value your broker genuinely quotes for most of your intended hours, and accept that it blocks trades. A blocked trade you had no edge on costs nothing. A trade filled through a 100-point spread on a 200-point target is already behind when it opens. And setting the cap to zero does not widen your tolerance; it switches the filter off, which is the opposite of what the input is for. The XAUUSD scalping robot settings guide goes into the spread cap in more detail.

Truth four: the drawdown figure is the one that decides survival

A win rate tells you how often trades worked; a drawdown tells you how bad the bad weeks get. On gold, the second number is the one that determines whether you are still trading next year.

When you look at any drawdown claim, ask three questions: is it measured on equity or balance, over how many trades, and was it live or demo. Equity drawdown includes open positions, which is the number that reflects what you could actually withdraw. Balance drawdown ignores them, which is how a grid robot shows a calm curve until it collapses. A figure without a trade count is meaningless, because a short sample flatters almost any strategy. And a demo record is a description of a demo.

Tested maximum equity drawdown	Meaning for a retail trader	Action
Under 10 percent	Very low, if the sample is long and honest	Still forward-test it through a news week
10 to 20 percent	Moderate; realistic for a sound gold scalper	Size so it is survivable, and expect more live
20 to 30 percent	Outer limit for most traders	Only with low risk per trade
Above 30 percent	Too deep to recommend	Reduce risk, or choose a different system

Plan for live drawdown to be deeper than the backtest, often by a substantial margin, because spread, slippage and swap are rarely modelled fully.

Here is a worked check you can run on any published figure. Suppose a sheet reports a 12 percent maximum equity drawdown over 40 trades. Forty trades is a short sample. At 1 percent risk per trade, a run of five consecutive losses alone costs about 5 percent, so the reported number has not yet met the kind of streak a scalper meets in a typical year. Ask for the trade count, the period and whether the figure is equity or balance, and treat the answer as the start of your own test rather than its conclusion. A figure built from one good month is an opinion, not a measurement. The best gold robot picks page uses the same ranking — drawdown and exposure before returns — across a shortlist.

Truth five: the timeframe changes what the EA can be

The marketing implies M1; the honest starting point for most traders is M5 or M15. The lower the timeframe, the more of gold’s spread and latency is compressed into every decision, and the smaller the margin for error.

There is nothing wrong with M1 gold scalping as a discipline — it is a real style with real operators. But it demands that the two variables you cannot control, spread and latency, are both good, and that your settings are measured rather than inherited. Beginners are better served by fewer trades on a higher timeframe, because lower frequency gives you time to notice when something is wrong, and because the cost per trade is a smaller share of a bigger target. If you want the deeper strategy view, the gold scalping strategies guide covers the session and range logic behind it.

The timeframe also determines your expected trade count, which in turn determines whether your sample is large enough to mean anything. A higher timeframe produces fewer trades, which is honest but slow to test; a lower timeframe produces many trades, which tests faster but costs more. Match the timeframe to your patience and your broker, not to the screenshot.

Truth six: position size should be derived, never chosen

This is the truth that saves accounts, and it is the one free-EA pages skip. A gold scalper will have losing streaks; the size of each loss is what turns that streak into a survivable week or a closed account.

Derive the lot from the stop and a fixed risk percentage, never the other way round. If gold needs a 2.50-dollar stop on your timeframe and you are willing to risk 0.5 percent of a $1,000 account, the lot size solves to roughly 0.02 on a 100-ounce contract. That is the number, and it does not change because a bigger lot would make the returns look better. Choosing the lot first inverts the relationship, so your risk quietly grows exactly when the market is more volatile and the stop is wider.

The arithmetic of a losing streak is worth stating plainly. At 0.5 percent risk per trade, a run of five losses costs about 2.5 percent. At 2 percent, the same run costs about 10 percent. On a scalper, five losses in a row is an ordinary event, not a crisis. The strategy did not change between those two accounts; the size did, and the size decided the outcome. A practical habit makes this concrete: before you accept any lot the EA proposes, write the dollar amount you would lose if the stop is hit, then ask whether you would still leave the robot running after three of those losses in a row. If the answer is no, the size is wrong, whatever the backtest says. This is why the featured EA is one you can read rather than another black box: you can confirm its stop logic and its sizing before you trust a single number it produces.

Truth seven: what the EA does when the market gaps

Gold gaps over weekends and jumps through stops during news, and a scalper’s stop is not a promised price — it is an instruction that becomes a market order when touched. In fast conditions you can be filled well beyond the level you set, which means the risk you calculated is the risk you hoped for, not the risk you always get.

Three practical responses, none of them optional on gold. First, cap the total daily loss so one bad session cannot compound into a bad week. Second, stand aside inside the window around scheduled releases rather than trading through them. Third, keep the position small enough that a gap of several times the normal spread is survivable, because on gold that is a realistic single-event move, not a tail case. ESMA’s product intervention measures cap retail leverage on gold at 20:1 and require negative balance protection, which limits the worst case but does not remove it. That is why the sizing decisions above matter more than any setting the EA ships with.

What exactly do you get, and what does it not do?

The EA featured here is Gold Bakome EA, a free MIT-licensed expert advisor for MetaTrader 5 with its source published by BAKOME-Hub. It is a gold EA you can read, and it is deliberately built as the opposite of the family truth two warns about: no grid, no martingale, no averaging, and a fixed stop loss as a safety net.

Why this instead of a seventh robot recommendation? Because the seven truths above keep arriving at the same conclusion: the variables that decide your result are spread, stop, timeframe and size, not the entry rule. Here you get a named publisher, a stated licence, readable MQL5 source and a stop you can point at — the checks from truth one, all of them answerable because the file is in the open. Run it on demo and the same truths apply; you are judging a known object rather than a screenshot.

What	Detail
Name	Gold Bakome EA (gold_bakome)
Platform	MetaTrader 5 (MQL5 source)
Licence	MIT — use, modify and distribute, keeping the copyright notice
Author	BAKOME-Hub
What it does	Trades XAUUSD with ATR-based adaptive logic, a fixed stop loss and session filters
What it does not do	It does not promise any particular return, and it does not size positions for you until you set its risk inputs

What MIT means for you. MIT is the most permissive of the common open licences: you may use, modify and redistribute the software, including commercially, as long as the copyright notice and licence text travel with any copy you pass on. For a trader, the useful part is the same as with any open licence — you can read the source and confirm what the file does with your account before it does anything at all.

What it does not do, stated plainly. It does not know your broker’s spread, commission or fill quality, and it does not carry a verified live track record. Its logic is adaptive rather than fixed, so two runs on two feeds will diverge. The behaviour comes from inputs you supply and a market you do not control, not a prediction — and it will not turn a bad configuration into a good one.

How to use it on a gold scalper. Install it on a demo XAUUSD chart and read the input block before you change a thing, exactly as truth two asks: find the stop, confirm there is no lot multiplier or averaging routine, and note the session filter. Then apply the sizing rule from truth six to whatever lot it proposes, and set a spread cap from truth three. Doing those steps on a file you can read is the whole exercise. For the wider robot context, the best MT4 EA hub applies the same standard to the catalogue.

How should you record a demo test so the result means something?

Use one row per closed trade and record the facts the terminal will not summarize for you: date and session, intended entry, actual fill, spread at entry, stop distance, calculated risk, exit reason and a screenshot. Add the EA version, preset name, broker server and symbol suffix at the top of the sheet. Without those details, you cannot tell whether a changed result came from the strategy, the settings or the feed.

Review the record after thirty trades, not after every loss. Calculate the average spread, average slippage, longest losing run, maximum equity drawdown and average result after costs. Then compare those figures with a backtest covering the same calendar dates. The two records do not need to match trade for trade, but they should tell the same broad story. If the backtest is positive while the demo is consistently negative after costs, do not optimize the demo until it agrees. Find the source of the difference first.

Keep the test configuration fixed for the full sample. Changing a stop after five trades, a session after ten and a risk level after fifteen creates three tiny experiments rather than one useful one. If you discover a setting that must change, close the test, label the result, start a new version and collect a new sample. This feels slow, but it is much faster than funding a live account with evidence that cannot answer a basic question.

Why does a two-decimal quote change your stop?

A gold-specific note that saves confusion: if your broker quotes XAUUSD to two decimals, set the stop in price terms rather than in “pips”. A one-point move on gold is not the same dollar value as on a five-decimal currency pair, and mixing the two is how a risk calculation silently becomes wrong.

How do you stress-test a preset before it trades live?

A preset is a set of assumptions about the market. Stress-testing finds out which assumptions break first, and it belongs before the account is funded rather than after.

Test the costs, not just the strategy. Re-run the backtest that looked good with the spread, commission and swap your broker actually charges. A gold scalper can win the entries and still lose the account to costs, because a strategy targeting a few dollars per trade hands a fixed toll to the broker on every round trip.

Change one assumption at a time. Widen the spread by fifty percent and re-run. Add the slippage you measured on demo and re-run. Shift every entry by one bar and re-run. When performance collapses after a single change, you have found the dependency that matters, and it is usually spread or timing rather than the entry rule.

Read the worst week, not the average month. Sort the equity curve into weekly buckets and study the bottom five. That distribution describes what the account has to survive, and it is exactly what a monthly return conceals.

The three failure modes worth naming before you install

• The stop sits inside the noise. Gold moves several dollars without changing anything. A stop tighter than that turns ordinary variation into a loss, and no exit rule repairs it.
• The position is chosen rather than derived. When the lot is fixed while the stop moves, risk varies silently with volatility, and it is largest exactly when the market is fastest.
• The session is wrong. Running a London-to-New-York scalper through the Asian range, or holding it through a policy announcement, is a different strategy from the one that was tested.

Why one clean stress test beats ten optimised backtests

Optimisation is good at finding settings that fit the past, and it says nothing about whether those settings sit on a plateau or a spike. A stress test answers the more useful question: if the market is a little worse than the sample, does this survive? Two or three deliberate perturbations will describe the shape of the risk far better than another hundred parameter combinations describe the height of the curve.

Keep the results. Write the spread, the slippage, the dates and the version into a file beside the preset. When the live account diverges from the demo, that record is the only thing that lets you find out why — and if the preset survives the cost increase and the timing shift, run it on demo for a defined sample and then leave it alone.

What the stress test cannot tell you

It cannot tell you what will happen, only what your assumptions have to survive. A preset that holds up under a wider spread and a shifted entry has demonstrated robustness, not a result. Keep that distinction, because it is the difference between evidence and reassurance.

Nor can it rescue a strategy whose logic you have not read. If the entry rule depends on something that exists only in the past — a repainting indicator, a value that looks ahead, or a signal that recalculates its last few bars — the numbers will look fine in every test and change their answer in live conditions. Reading the source is the only check for that, and it costs minutes rather than a testing cycle.

Finally, it says nothing about execution when volatility spikes. Stress-testing inside a backtest is a simulation of costs; the live version arrives as slippage on the entry you believed you had. Measure it on demo, compare it against what the backtest assumed, and adjust the preset before it runs on real money. Record the outcome either way, because the next scalper you evaluate will raise exactly the same three questions.

Your next step before installing any gold scalper

Do two things, in this order.

First, run the seven checks on the Scalper XAUUSD EA free download you have in hand: publisher and licence, source, stop logic, spread need, drawdown, timeframe, and how its position size is chosen. Any one of them can be a reason to walk away, and together they take less time than reading a sales page.

Second, get the MIT-licensed Gold Bakome EA from its source repository, read the single MQL5 file, and confirm for yourself that it carries a real stop and no averaging routine. Then run it on demo and size it by the rule from truth six. Doing that on a file you can inspect lowers your drawdown more than switching robots ever would.

Be clear-eyed about what you are doing. Gold is volatile, leverage works against you as easily as for you, and losses happen on any system. Everything here is educational, none of it is financial advice, and no result is promised or implied. Test on demo, size every position as though the worst backtest day will happen again, and only risk money you can afford to lose. If you would rather start from a curated set of downloads with their platforms and licences stated, the free download library is the place to begin.