Search for the best gold trading bot for MT5 and you will find lists: ten names, a star rating each, an affiliate link under every one. What those lists rarely give you is a way to judge the next bot yourself. This guide is that: the questions to put to any gold expert advisor before you pay for it — ours included — and what a good answer looks like.
Why the 'best bot' lists cannot answer this for you
Most ranking pages are compiled from submissions and paid placements, not from testing. We documented how that works in how the 'best signal channel' lists are actually ranked, and bot lists are built the same way. Use them to find names. Use the checks below to decide.
The checks, in the order we would run them
1. The conditions behind the result
Every equity curve was produced on some account, over some period, with some cost model. Ask for all of it: the dates, the account type, whether spread and commission were included, whether lot size grew with the balance. A curve that compounds will always look steeper than one that does not, and both can be honest — but you need to know which one you are reading.
2. The worst parts before the best
Read the maximum drawdown, the worst month and the longest stretch without a new high before you look at the total. Then ask yourself whether you would still be running the bot at the bottom of that drawdown, with your own money. If the honest answer is no, the total return is irrelevant — you will switch it off at the worst moment. Our piece on what a drawdown does to your position size shows why this matters more than it looks.
3. Tester or live — and how the two were made
A strategy tester result and a live account answer different questions. The tester tells you how the rules behaved over a long history; a live account tells you how they survive real fills. Both are useful. What is not useful is a test with no stated cost model, or a 'live' screenshot with no account history behind it. The common ways a backtest flatters itself are in why your backtest of a news strategy was wrong.
4. Your broker is part of the result
Gold is expensive to trade at the wrong hour. The spread widens around releases and at the daily rollover, slippage appears when liquidity thins, and swap is charged on anything held overnight. A bot tested on one account type and run on another is a different bot. Ask which account the published numbers came from and run it on the same kind.
5. What you control
You should be able to set the risk yourself and see it respected: risk per trade, a maximum daily loss, a stop on every position. Ask what the minimum balance is and why. A bot that needs a larger account than you have is not 'almost right' for you — it is the wrong size.
6. What you are actually buying
- The licence: which accounts it runs on and how you move it to a new one.
- Updates: who maintains the bot when the broker or the platform changes.
- Support: a named way to reach a person, and how fast they answer.
- Terms: what happens when you stop paying, and what the refund policy says in writing.
- Setup: whether you need a VPS to keep MT5 running around the clock, and who helps you set it up.
- Results come with no dates, no account type and no cost model.
- Drawdown is missing, or only shown as a 'recovered' curve.
- The seller promises returns or calls the bot risk-free.
- You cannot set or cap the risk yourself.
- Support is a chat bubble with no name behind it.
- The period, account type and costs are stated next to the result.
- Maximum drawdown and the worst month are printed as plainly as the total.
- The minimum balance is stated and explained.
- Risk per trade and a daily loss limit are yours to set.
- Licence, updates and refund terms are written down.
A short routine before you subscribe
- Write down your limits
The largest drawdown you would sit through, the account size you will fund, the account type you trade on.
- Read the worst numbers
Drawdown, worst month, longest flat stretch. Compare each with your limits.
- Match the conditions
Same kind of account, same cost model, same lot logic as the published result.
- Start at the minimum
Run it at the smallest size the seller supports and log every trade in R multiples, so you can compare like with like.
- Review after a month
Not the profit: whether the trades, the drawdown and the costs look like what you were shown.
Where DM's bots stand on these checks
Our trading bots page puts the conditions next to every result: the test period, the account type, costs included, compounding stated, maximum drawdown, the worst month, the minimum and recommended balance. You set the risk. Read the Gold Pro Bot page against the checks above and decide for yourself — that is what the checks are for. For gold's day-to-day behaviour, the gold market page shows the current range and levels.
FAQ
Is there a gold bot that cannot lose?
No. Every bot that trades has losing trades and losing months. A seller who says otherwise has told you the one thing you need to know.
Do I need a VPS for an MT5 gold bot?
MT5 has to be running for the bot to trade. If your computer is not on around the clock, a VPS keeps the terminal running. Ask the seller what they recommend and whether they help with the setup.
Backtest or live record — which should I trust?
Both, for different things, and neither without its conditions. A long backtest with stated costs shows how the rules behave across markets; a live account shows how they survive real execution.
Read next
How Much Does Gold Move on US Retail Sales Day? Every Release of the Year, Measured
Every US Retail Sales release from November 2025 to October 2026, measured on gold: the hour before, the first 15 minutes, the first hour and the day.
Read →How Wide Should a Stop Loss Be on Gold? Six Months of Candles, Measured
How often a $5, $10 or $15 stop on gold gets touched by ordinary noise, measured on six months of real XAUUSD bars — and how to choose a stop that fits.
Read →How Much Does Gold Move on Fed Day? Every FOMC Decision of the Year, Measured
All eight Fed rate decisions from November 2025 to October 2026, measured on gold: the hour before, the first 15 minutes, the press conference, the day and the day after.
Read →