372 free Expert Advisors from the MQL5 Code Base, every one run on the same conditions. 32 finished in profit. 11 did it without a drawdown over 20% — those are below. Each row links straight to the author's file: free, no signup.
Positive return, maximum drawdown under 20%, at least 30 closed trades. Sorted by return. Row numbers are the file's position in the full 372-row table.
| # | EA | Return 32 mo |
Per mo | Max DD | Trades | Win % | |
|---|---|---|---|---|---|---|---|
| 310 | Daydream01 | +16.71% | +0.52% | 15.9% | 43 | 81.4 | Download ↗ |
| 252 | Currencyprofits_01.1 | +4.82% | +0.15% | 5.3% | 44 | 15.9 | Download ↗ |
| 48 | MACD EA | +4.69% | +0.15% | 6.7% | 46 | 61.5 | Download ↗ |
| 169 | Eugene | +4.69% | +0.15% | 16.2% | 32 | 37.5 | Download ↗ |
| 61 | х1 | +3.61% | +0.11% | 6.0% | 76 | 73.7 | Download ↗ |
| 118 | MA2CCI same code, posted 3× | +3.31% | +0.10% | 6.4% | 261 | 28.7 | Download ↗ |
| 316 | SMC MaxMin at 1200 | +2.59% | +0.08% | 5.7% | 983 | 50.2 | Download ↗ |
| 4 | VR Locker Lite - positive-lock strategy | +2.56% | +0.08% | 3.4% | 145 | 38.6 | Download ↗ |
Return is the closing balance against the $10,000 start. Max DD is the deepest peak-to-trough of equity, floating losses included — that is the number that decides whether you can hold the position. Trades are closed trades.
Every row here was tested on $10,000. On a smaller account the same dollar drawdown is a bigger share of it, and some of these stop being runnable at all: put your account size in and this list rebuilds →
Every download on this site is a file, not an app. Once it is on a chart and doing nothing, the reason is almost always one of nine things and none of them is the file: the nine checks, cheapest first →
These show a 90–100% win rate because they never closed a losing trade. Read the next section before you download one.
| # | EA | Return 32 mo |
Per mo | Max DD | Trades | Win % | |
|---|---|---|---|---|---|---|---|
| 145 | The new moving-average trading system no losing trade closed | +12.88% | +0.40% | 17.7% | 64 | 100.0 | Download ↗ |
| 137 | Virtual trailing stop no losing trade closed | +5.41% | +0.17% | 10.3% | 103 | 100.0 | Download ↗ |
| 274 | Channels no losing trade closed | +0.71% | +0.02% | 10.4% | 50 | 98.0 | Download ↗ |
A 100% win rate is not a win rate. It is an EA with no stop loss. On those three rows the average losing trade is exactly 0.00 — nothing was ever closed at a loss, because the position was held until price came back. The 10–18% drawdown on those rows is the size of the hole it was sitting in while it waited. That works until the one time price does not come back, and then the account is gone in a single trade. If you run one of these, set your own stop.
And the best row on this page is +16.7% in 32 months. That is +0.52% a month, about 6% a year, before swap and before the spread widening you will actually meet. It is not a living. It is a real result, which is rarer than it sounds — and it is why the other 340 files are not on this page.
Daydream01 — #310 on the list.
81.4% of its 43 trades won, and its average win ($69.40) is smaller than its average loss ($91.62) — so it wins often and small, loses rarely and bigger. That is the profile that keeps working when the market stops trending; the opposite shape (rare big wins, constant small losses) is what empties accounts quietly. 43 trades in 32 months is about one every three weeks, so it needs patience, not attention. Its 15.9% drawdown is the part you have to be able to sit through.
Lowest drawdown on the whole page is VR Locker Lite - positive-lock strategy at 3.4% — +2.56% over the period. Small return, but it barely wobbled getting there, which is what you want from the first EA you ever put on a live chart.
The filter is not strict. The pile is just that bad.
| Stage | Files | What happened |
|---|---|---|
| Downloaded from the Code Base | 372 | Every free EA I could get over three months |
| Compiled and opened a trade | 190 | The rest never traded, or would not compile |
| Ended the period in profit | 32 | 158 of the 190 that traded lost money |
| Profit, drawdown under 20%, 30+ trades | 11 | After collapsing 2 rows that were the same code reposted |
The other 361 are not all scams. Most of them did something more ordinary: never traded because the defaults were wrong, or lost 20% a year in small pieces, or made a fortune and gave 96% of it back. The full written analysis is lab note 005, and every single row is public in the 372-row table with the raw CSV behind it.
Running live prices since 29 October 2026, on the same $10,000.
I picked Daydream01, SMC MaxMin and VR Locker Lite from this list — three different structures, not the three highest numbers — and put them on a MetaTrader 4 demo to see whether the backtest was describing anything real. The trades are published as they fill, including the ones that lose.
Lab note 007 — the forward test, and the four ways it could still be wrong →
The same harness, the same published conditions, the same willingness to print a loss. What a test costs and what you get →