Free Expert Advisors for MT5: We Tested the 34 That Ship With MetaTrader

If you search for a free Expert Advisor for MT5, the first 34 are already installed. MetaTrader 5 ships them with every terminal, with their source code, written by MetaQuotes.

We tested them. All 34 ran in the Strategy Tester with the inputs they come with, on the same symbol, timeframe and period. This guide shows the result of each one, explains what went wrong in the worst cases and lists the other places where free EAs come from.

Disclosure: RoboXpert develops trading software and has a commercial interest in this subject. The tests ran on 7 October 2026 in MetaTrader 5 build 6230 on a demo server. They are backtests of example programs with default inputs. They are not a forecast and not a recommendation. AI assisted with research, calculation and drafting. See our editorial methodology.

Where to find a free Expert Advisor for MT5

There are four sources. They differ in whether you can read the code and in who checked the file.

The first two give you the source. You can read what the EA does, compile it yourself and change it. The quote is from the MetaTrader 5 help.

A free product in the Market is a compiled file. According to the Market rules it goes through the same automatic check as a paid one, and users who downloaded the full version may review it. Our guide on reading an MQL5 Market page explains what that check covers.

A paid EA that someone offers “free” or “cracked” on another website belongs in none of these groups. The Market rules forbid reselling and decompiling products. A file that has been through a decompiler is no longer the seller’s file, and nobody can tell you what was changed in it. An EA runs inside your terminal with access to your trading account.

The 34 Expert Advisors that ship with MetaTrader 5

In build 6230 they sit in three folders of the Navigator:

The seven patterns of the Free Robots are Black Crows and White Soldiers, Engulfing, Harami, Meeting Lines, Dark Cloud and Piercing Line, Hanging Man and Hammer, and Morning and Evening Star with Doji. The Examples folder holds five more programs that demonstrate charts, controls and 3D graphics. Their source contains no trade calls, so we left them out.

Every file carries the line “Copyright 2000-2026, MetaQuotes Ltd.” None of the 34 source files names a symbol or a fixed timeframe.

One of the 28 Free Robots is a copy of another. The file “BullishBearish MeetingLines Stoch” contains the pattern check of “BlackCrows WhiteSoldiers Stoch”: its code looks for “3 Black Crows” and “3 White Soldiers”, and no Meeting Lines. Apart from the file name, the magic number and spacing, the two source files are the same. In our installation there are therefore 27 different Free Robots, and the two files produced the same trades in every test.

How we tested them

Every EA ran once under the same conditions. The table lists them as the reports print them:

The symbol and the timeframe are our choice, because the files name none. We took EURUSD on H1 as a common setup; the M15 run further down changes the timeframe.

“18% real ticks” means that the server supplied real ticks from April 2026 only. For the 27 months before, the tester generated ticks from minute bars. Our glossary explains what that changes.

The Free Robots trade a fixed 0.10 lots with a stop and a target of 200 points each, which is 20 pips on EURUSD. Two of the other EAs size their positions from the free margin, so the leverage of the test account decides how large they trade. That matters for two results below.

Results of the 34 free EAs on EURUSD

Seven of 34 ended with a profit. The best one made 165.29 USD in 33 months, which is 1.65% of the deposit. The chart shows every result:

The table lists all 34 with the figures from their reports, sorted by net profit:

Three things stand out before any single EA:

• The profits are small and the losses are not. The seven winners made between 14 and 165 USD. Six EAs lost more than 700 USD.
• Many results rest on a handful of trades. Three EAs never traded. Five more made fewer than 30 trades in 33 months; one of the “winners” made a single trade.
• None of the frequent traders won. Six EAs made more than 500 trades, among them the four Dark Cloud and Piercing Line robots with up to 1,699. All six lost. Every trade pays spread and commission, and a strategy without an edge pays them for nothing.

Five results to look at closely

ExpertMAPSARSizeOptimized lost the whole deposit

The EA trades the same signal as ExpertMAPSAR, which ended with a small profit. The difference is one module: it sizes each position with 10% of the free margin. With 10,000 USD at leverage 1:500, its first position was 4.53 lots, about 450,000 EUR. The first two trades won more than 5,600 USD. Then the same size worked the other way. By July 2025 the account could no longer open 0.01 lots, and the report closes with a net loss of 9,998.31 USD and an equity drawdown of 99.99%.

The default input of an example file is no recommendation, and nothing in the file says so.

Moving Average ended near zero after a 56% drawdown

The best-known example EA closed 777 trades and ended 48.06 USD down. Its report shows the largest equity drawdown of all EAs that survived: 56.08%. It lost 4,618.74 USD in 2024 and won back 1,711.07 USD in 2025 and 2,859.61 USD in 2026. This EA also sizes from the free margin, with a default of 2%, and opened about 0.9 lots on our account. Net profit alone would have hidden the path; our guide on equity and balance drawdown explains the two figures.

The same two EAs at leverage 1:100

ExpertMAPSARSizeOptimized and Moving Average size from the free margin, so we ran both again with one change: leverage 1:100. The table sets the two runs of each EA side by side:

At a fifth of the leverage both EAs opened a fifth of the size. Moving Average made the same 777 trades and ended with a small profit after a drawdown of 15.43% where the first run showed 56.08%. ExpertMAPSARSizeOptimized still lost three quarters of the deposit. A backtest of an EA that sizes from margin belongs to the leverage it ran on. A report made at 1:500 says little about an account with lower leverage.

Three robots never traded

The Meeting Lines robots with CCI, MFI and RSI made no trade in 33 months of EURUSD H1. We did not examine why. An EA that does nothing shows no loss, and no information either.

One winner had one trade

“BlackCrows WhiteSoldiers RSI” shows a profit factor of 56.43. It made one trade, which won 19.40 USD. The profit factor divides gross profit by gross loss; here the gross loss was the commission of the entry deal. Read the number of trades before any ratio. Our guide on the Strategy Tester report shows where each figure comes from.

Do the winners stay winners?

Seven of 34 made a profit. With 34 attempts and no edge at all, some will end above zero. Two checks show whether the seven are more than that.

Year by year. Only one of the 34 EAs was profitable in each of the three calendar years: BullishBearish Harami RSI, with 40.43, 22.63 and 102.23 USD. All others had at least one losing or empty year.

Another timeframe. We ran all 34 again on M15, with everything else unchanged.

Two of the seven H1 winners also made a profit on M15: BullishBearish Harami Stoch with 10.59 USD and ExpertMAPSAR with 52.73 USD. The best EA on H1, Harami RSI, ended slightly below zero. Across all 34, three ended with a profit on M15, three never traded, 27 lost and one lost the deposit. The largest gain of any EA on M15 was 52.73 USD.

Moving Average, which had ended near zero on H1, made 3,402 trades on M15 and lost 9,374.71 USD, with an equity drawdown of 93.84%.

Another symbol. As a third check we ran the 34 on XAUUSD H1 in the mode “1 minute OHLC”, on the same gold data as our gold EA benchmark. Six ended with a profit, the best with 144.60 USD. Two lost the whole deposit: MACD Sample and ExpertMAPSARSizeOptimized. None of the seven EURUSD H1 winners was profitable in all three tests. The stops of the Free Robots are 200 points, which is 2 USD in the gold price, which is very close for gold. The run shows what happens when an EA is moved to another market without adapting its inputs.

Our guide on whether Expert Advisors work explains the selection effect with numbers: among enough candidates without an edge, a few always look good.

What a free EA is good for

Learning how an EA is built. The bundled files are short and commented. MACD Sample has about 450 lines; you can follow how it reads an indicator, checks for an open position and sends an order. If you want to write your own, start there or with our guide on building an MT5 EA with AI.

Practising the Strategy Tester. Every reader has these 34 files. You can repeat our runs, change one input, switch the test mode and watch what moves. That is the fastest way to learn how to read a report.

A base for your own rules. The source may be changed. A stop, a time filter or another position size is a few lines.

A free EA is no shortcut to a trading result. The check is the same as for a paid one: the rules, the loss control, a test on your own broker’s data and a forward period. Price says nothing about the edge in either direction. Our checklist for choosing an Expert Advisor applies to both, and the guide on installing an EA in MT5 covers files from the Code Base.

Download results and script (ZIP). The figures of every run as JSON, the script that reads them from the reports, and two original MT5 reports. No executable, no account data.

What we did not test

• Two symbols, two timeframes on EURUSD, one period, one demo server. Another broker’s price history, spread and commission will give other figures.
• Default inputs only. We did not optimise anything. An optimised version of any of these EAs would show a better backtest and would prove less.
• On EURUSD, 27 of the 33 months ran on generated ticks, because the server supplied real ticks from April 2026 only. The gold runs used four prices per minute.
• Leverage was 1:500, as the reports print it. Only the two EAs that size from the free margin were repeated at 1:100.
• We tested no EA from the Code Base or from the free section of the Market, and we name none. The counts for the Market are from its category pages on 7 October 2026.
• No forward test and no live account. These are backtests.

Frequently asked questions

Does MetaTrader 5 come with free Expert Advisors?

Yes. Build 6230 installs 34 trading EAs with source code: MACD Sample and Moving Average in the folder Examples, four in the folder Advisors and 28 in the folder Free Robots. You find them in the Navigator under Expert Advisors.

Is there a profitable free EA for MT5?

In our backtest, 7 of the 34 bundled EAs ended with a profit on EURUSD H1, the best with 165 USD on 10,000 USD in 33 months. Only one was profitable in each of the three years. A backtest with default inputs on one symbol does not show that an EA will be profitable.

Are free Expert Advisors safe?

It depends on the source. Files that come with MetaTrader and files from the MQL5 Code Base include their source code, so you can read and compile them yourself. A compiled file from an unknown website cannot be inspected and runs with access to your trading account.

Can I use the bundled EAs on a live account?

Technically yes. They are examples: in our test one of them lost the whole deposit with its default inputs and another went through a 56% drawdown. Test any EA on your own broker’s data and on a demo account first.

Is a free EA worse than a paid one?

The price does not tell you. A free EA with source code can be inspected, which most paid products cannot. A paid product may come with support and updates. Neither is evidence of a trading edge; apply the same checks to both.

Where is the source code of the bundled EAs?

In the data folder of the terminal under MQL5/Experts, in the subfolders Advisors, Examples and Free Robots. Each EA is an MQ5 file that opens in MetaEditor, next to the compiled EX5 file the terminal runs.