The yazan Gold Reaper MT5

The yazan Gold Reaper MT5 is a breakout Expert Advisor for gold (XAUUSD). It watches three kinds of price levels on three timeframes, waits for a strong bar to close beyond a level, checks that the breakout holds on the next bar, and then trades in the breakout's direction with a stop loss and a take profit on every trade.

Please read this first. Before this product was finished, its rules were tested in a plan written and frozen in advance (2020-2026, gold, Strategy Tester). The test FAILED. The default settings lost money in every tested year after costs (before costs, 2022 alone ended slightly positive), and also on the out-of-sample period that was never used to choose anything. The full tables are below under "What our tests found". About 80% of the trades win, but that is not an edge: most wins are small break-even exits, and one stop loss erases about six of them. Treat this EA as a research and education tool for breakout rules on gold. Test it on a demo account and in your own Strategy Tester. It does not promise any profit.

What you see on the chart

The numbers below match the numbered markers in the screenshots.

1. Header: the EA name and the language button with 12 panel languages (English, Arabic, Russian, Chinese, Spanish, Portuguese, Japanese, German, Korean, French, Italian and Turkish).

2. Status: "on" or "paused" with the reason: rollover hours, NFP or FOMC day, Friday close, spread above the cap, daily loss limit reached, or a real account while "Trade only on demo accounts" is on.

3. Today and spread: today's result in % of the day's starting equity next to the daily loss limit, and the current spread next to the maximum spread.

4. Strategies: the 9 strategies, each on or off. A strategy that is on shows what it is doing: watching levels, a breakout waiting for its confirmation bar, an armed entry, or an open trade with its result in R, plus the time and direction of its last signal.

5. Totals: open trades of this EA against the maximum, and their open result in R and in money.

6. Scroll: when the panel is taller than the chart, its body scrolls with the mouse wheel or the arrows at its right edge.

7. Trades on the chart: every trade has its stop loss and take profit on the server. The trade comment carries the strategy number, so the break-even and trailing rules keep working after a restart of the terminal.

The 9 strategies

Each strategy is one level type on one timeframe (M5, M15 or H1); the chart timeframe does not change them. By default only the three H1 strategies are on: they were the least negative in the test (see Section 3). The M5 and M15 strategies can be switched on, but in the test they lost more, because costs weigh more on their smaller stops.

• A. Session range: the Asia session high and low (00:00-08:00 server time) and the London open range (the first 60 minutes from 09:00 server time).
• B. Swing support and resistance: fractal swing highs and lows, merged into zones when they are closer than 0.5 ATR; the 5 zones with the most touches, swings at most 200 bars old.
• C. Previous day and previous week: their high and low.

How it enters

• Momentum: a closed bar must close beyond the level by at least 0.25 x ATR(14), with a body of at least 50% of its range and a range of at least 0.8 x ATR.
• False-breakout filter: the next bar must also close beyond the level; the entry is then made at market. An optional retest mode waits instead for a later bar (within 12 bars) to touch the level and close beyond it again.
• One trade per strategy, at most 3 open trades in total (1 on a netting account), and no new trade when the spread is above 40 points.
• Closed bars only: the forming bar is never used for a signal.

How it exits

• Stop loss: 1.5 x ATR, and at least 0.2 x ATR beyond the level. Take profit: 1.3 x the stop distance (1.3 R).
• Break-even: when the profit reaches 0.25 x ATR, the stop moves to the entry plus 0.125 x ATR.
• Trailing stop: 2 x ATR behind the price.
• Trailing take profit: from 90% of the way to the target, the take profit moves further out by 0.5 x ATR for each further 0.5 x ATR of progress.
• Time exit: after 24 bars of the strategy's timeframe.

Filters and risk

• No entries on NFP days (first Friday of the month) or on FOMC decision days (table 2020-2026), and none from 23:00 to 01:00 server time (rollover spread).
• All trades are closed on Friday at 22:00 server time.
• Daily loss limit: at -3% of the day's starting equity, new entries stop and open trades are closed.
• Risk per trade: 1% of the balance by default, or a fixed lot. If the smallest lot would risk more than twice the target, the trade is skipped. Free margin, volume limits, stops and freeze levels are checked before each order.
• "Trade only on demo accounts" is on by default: on a real account the panel runs but no trades are opened until you switch it off.

What it does not do

• No grid, no martingale, no averaging, no hedging, no lot increase after a loss.
• No DLLs and no web requests. It does not learn or change its rules by itself.
• It does not promise or predict profit. With its default settings it lost money in every year of the test below.

In the Strategy Tester

If the EA has opened no trade at all after the first 30 bars of the chart timeframe of a test (for example a short test on D1, or a symbol whose spread is always above the cap), it takes ONE trade on the next valid breakout of the chart timeframe and writes "validator check" in its trade comment and in the journal. This happens at most once per test, only in the Strategy Tester, and never after a normal trade has opened. It exists so that a short automatic check shows the order path working. Switch it off with the last input ("Tester only: if no trade in the first 30 chart bars, ONE validator check trade per run") to see the strategies alone. In our 16-month out-of-sample test it added exactly one trade.

The research inputs (extra cost per trade, trade CSV) only change the statistics line in the journal and the optional files; they never change trading. The four control inputs (trade direction, its seed, every breakout without filters, random levels) do change trading: they exist to test the idea and are inactive at their defaults.

What our tests found

Result: the test FAILED. The default settings lost money in every tested year (2020-2026) after costs, and on the out-of-sample period.

How it was tested: the rules, the parameter choices allowed, the periods and the pass bar were written down and frozen before the first test run. XAUUSD, MetaQuotes-Demo server, Model "1 minute OHLC", M5 chart, deposit 10,000 USD, fixed lot 0.01. "Tester" = the Strategy Tester result with the recorded spread. "After cost" also subtracts 0.30 USD per trade at 0.01 lot for slippage and commission. R = the result divided by the money at risk at the first stop loss.

• DISCOVER 2020.01-2023.06: the only period used for choices.
• CONFIRM 2023.07-2025.06: one run of the chosen set.
• OUT-OF-SAMPLE 2025.07-2026.10: one run, never used for a choice.

A. Choosing the default set on DISCOVER

set (DISCOVER) | trades | win % | profit factor | avg R | net USD at 0.01 lot

all 9 strategies, k 0.15, tester | 31,229 | 81.7% | 0.92 | -0.020 | -1,277

all 9 strategies, k 0.15, after cost | 31,229 | 27.7% | 0.48 | -0.155 | -10,646

H1 strategies only, k 0.25 (default), tester | 2,601 | 83.8% | 0.90 | -0.023 | -285

H1 strategies only, k 0.25 (default), after cost | 2,601 | 80.9% | 0.65 | -0.065 | -1,065

All 9 strategies had a negative average R after cost. The H1-only set lost the least, so it became the default. Its result was already negative before the later periods were run.

B. The default set on all periods

period | trades | win % | profit factor | avg R | net USD at 0.01 lot | max drawdown

DISCOVER, tester | 2,601 | 83.8% | 0.90 | -0.023 | -285 | 367 USD

DISCOVER, after cost | 2,601 | 80.9% | 0.65 | -0.065 | -1,065 | 1,066 USD

CONFIRM, tester | 1,388 | 84.4% | 0.87 | -0.016 | -296 | 396 USD

CONFIRM, after cost | 1,388 | 80.8% | 0.70 | -0.052 | -713 | 771 USD

OUT-OF-SAMPLE, tester | 971 | 81.8% | 0.75 | -0.034 | -1,202 | 1,271 USD

OUT-OF-SAMPLE, after cost | 971 | 81.8% | 0.69 | -0.046 | -1,493 | 1,551 USD

2020-2026, tester | 4,965 | 83.6% | 0.82 | -0.023 | -1,784 | 1,828 USD

2020-2026, after cost | 4,965 | 81.0% | 0.68 | -0.058 | -3,274 | 3,281 USD

Same out-of-sample period with "Every tick based on real ticks": 958 trades, profit factor 0.75 (tester) / 0.69 (after cost), -0.051 R per trade after cost.

C. Per year (2020-2026, after cost)

year | trades | net USD at 0.01 lot | profit factor | avg R | buy and hold 1 oz (USD)

2020 | 719 | -310 | 0.70 | -0.066 | +365

2021 | 772 | -351 | 0.60 | -0.071 | -115

2022 | 727 | -183 | 0.75 | -0.041 | +22

2023 | 724 | -336 | 0.58 | -0.079 | +224

2024 | 712 | -348 | 0.66 | -0.064 | +566

2025 | 712 | -722 | 0.71 | -0.039 | +1,661

2026 (to Oct 31) | 599 | -1,023 | 0.70 | -0.041 | -473

7 of 7 years lost money. Per strategy (2020-2026, after cost): A H1 -0.072 R, B H1 -0.051 R, C H1 -0.045 R; all three negative.

D. Why the win rate is about 80% and the account still loses

Out-of-sample exits: 591 at break-even, 206 by the trailing stop, 131 at the first stop loss, 15 at the extended take profit, 7 at the take profit, 13 at the Friday close, 5 by the time exit, 3 by the daily loss limit. The break-even stop locks a small profit, so these trades count as wins. Average win +0.16 R, average loss -0.90 R (tester): one loss erases about six wins.

E. Controls (same entries and exits)

control | out-of-sample avg R after cost | 2020-2026 avg R after cost

default set (breakout direction) | -0.046 | -0.058

always buy | -0.001 | -0.049

random direction, 100 seeds | mean -0.020 (5th-95th percentile -0.048 to +0.005) | not run

every breakout, no momentum or false-breakout filter | -0.028 | -0.046

levels moved to random prices, 20 seeds | mean -0.021 (range -0.038 to +0.010) | not run

The breakout direction did worse than a coin flip on the same entries (5th percentile), and the real levels did not beat random levels. Buy and hold of 1 oz of gold (0.01 lot) made +2,330 USD over 2020-2026 and +520 USD over the out-of-sample period; the default set lost 3,274 USD and 1,493 USD.

F. Pass bar of the plan

check | result | verdict

out-of-sample trades >= 200 | 971 | pass

average >= +0.15 R per trade after cost | -0.046 R | fail

profit factor >= 1.3 after cost | 0.69 | fail

3 of 4 outside years positive | 0 of 4 | fail

beats always-buy on the same entries | -0.046 R vs -0.001 R | fail

one-step parameter changes keep the sign | all 6 neighbours also negative | the loss is stable, not an edge

Monte-Carlo 95th percentile drawdown at 1% risk < 5% | 53.8% | fail

G. In money at a fixed risk per trade (out-of-sample, after cost, compounded)

risk per trade | actual result (max drawdown) | Monte-Carlo median | Monte-Carlo 5th percentile | Monte-Carlo 95th percentile drawdown

1% | -37.2% (40.4%) | -37.4% | -52.4% | 53.8%

2% | -61.7% (65.5%) | -61.9% | -78.0% | 79.3%

A higher risk only loses faster when the average trade is negative.

Each test above is a Strategy Tester backtest or a simulation built from its trade list: hypothetical, not real trading. To repeat the out-of-sample run: XAUUSD, M5 chart, 2025.07.01-2026.10.31, "1 minute OHLC", 10,000 USD, leverage 1:100, all inputs at default except "Fixed lot" 0.01 and the validator check off. Your broker's prices and costs will give different numbers.

Disclaimer for all results above

Hypothetical or simulated performance results have certain limitations. Unlike an actual performance record, simulated results do not represent actual trading. Since the trades have not been executed, the results may have under- or over-compensated for the impact of certain market factors, such as lack of liquidity, slippage and broker execution. Simulated trading programs are also designed with the benefit of hindsight. No representation is being made that any account will or is likely to achieve profits or losses similar to those shown.

Inputs

Strategies (level type x timeframe)

• A Session range on M5 / M15 / H1: false / false / true
• B Swing S/R zones on M5 / M15 / H1: false / false / true
• C Previous day / week high-low on M5 / M15 / H1: false / false / true

Levels

• A: Asia session start / end hour, server: 0 / 8
• A: London open hour, server: 9; London open range, minutes (0 = off): 60
• B: max age of a swing, bars: 200; zones used (most touches first): 5; merge swings closer than x ATR into one zone: 0.5

Breakout

• ATR period: 14
• Close beyond the level by at least k x ATR: 0.25
• Breakout bar body, min share of its range: 0.50; breakout bar range, min x ATR: 0.80
• Entry after the breakout: Confirmation (the next bar also closes beyond the level) or Retest; retest bars allowed: 12

Trades

• Risk per trade, % of balance: 1.0; fixed lot (0 = use the risk %): 0.0
• Max open trades (netting account: 1): 3; max spread for a new trade, points (0 = off): 40
• Stop loss, x ATR: 1.5; at least x ATR beyond the level: 0.2; take profit, x R: 1.3
• Break-even at x ATR of profit (0 = off): 0.25, locking x ATR: 0.125
• Trailing stop distance, x ATR (0 = off): 2.0
• Trailing take profit: true; starts at this share of the TP distance: 0.9; moves x ATR per x ATR of further progress: 0.5
• Time exit after N bars of the strategy timeframe (0 = off): 24

Filters

• No entries on NFP day / FOMC days: true / true
• Close all trades on Friday: true, at hour (server): 22
• No entries from / until this hour, server: 23 / 1
• Daily loss limit, % of the day's start equity (0 = off): 3.0

Safety

• Trade only on demo accounts: true
• Magic number (strategy n uses magic + n): 20261102

Panel

• Panel language: English (12 languages; also switchable on the panel)
• Panel X / Y, pixels from the top-left corner: 10 / 30

Research (Strategy Tester)

• Extra cost per trade in price units per lot, statistics only: 0.30 (0.30 = 0.30 USD per oz on gold; set 0 on other symbols)
• Trade direction (controls): Normal; seed for the random direction: 1
• Control: every breakout, no momentum or false-breakout filter: false
• Control: levels replaced by random prices in the previous day's range (seed, 0 = off): 0
• Write a trade CSV and a passes.csv row (Common\Files\GoldReaper): false; run name: run
• Tester only: one "validator check" trade if there is no trade in the first 30 chart bars: true

Quick start

• Run it in your Strategy Tester first: XAUUSD, any chart timeframe, "Every tick based on real ticks" if you have them. Compare the result with Section 3.
• Attach it to an XAUUSD chart on a demo account and turn on Algo Trading. The panel shows the status and the 9 strategies.
• Keep the defaults, or switch strategies on and off and change the stop, target and filter inputs to test your own breakout ideas.
• Pick the panel language with the button at the top right.

FAQ

Is it profitable?

Not in our tests. The default settings lost money in every year 2020-2026 after costs (before costs, 2022 alone ended slightly positive) and on the out-of-sample period, before and after costs. Past or simulated results do not guarantee future results.

Why does it show a high win rate?

Most trades reach the break-even level and close there with a small profit, so they count as wins. The losing trades are about six times larger. Section 3 D shows the exits.

Does it use martingale, grid or hedging?

No. One trade per strategy, at most 3 open, a stop loss and take profit on every trade, a fixed risk per trade.

Which symbols?

It was built and tested for XAUUSD. It runs on other symbols, but the levels, the spread cap in points and the research cost input are set for gold.

Why did it not trade?

The status line shows the reason: rollover hours, NFP / FOMC day, Friday close, spread above the cap, the daily loss limit, or a real account with "Trade only on demo accounts" on. In the test the three H1 strategies opened about 14 trades a week on average.

What is the "validator check" trade in my Strategy Tester?

One trade that the EA takes only in the tester, only if it opened no trade in the first 30 bars of the chart timeframe, at most once per test. Switch it off with the last input to see the strategies alone.

What is the "stats after cost" line in the journal?

At the end of a tester run the EA prints its own statistics after the research input "Extra cost per trade". 0.30 is meant for gold. On other symbols set it to 0, otherwise the line is meaningless; it never changes trading.

Risk warning

Trading gold on margin carries a high level of risk and may not be suitable for all investors. You can lose some or all of your deposit. This Expert Advisor lost money with its default settings in every year of the backtests on this page, and breakout trading is sensitive to spread, slippage, commission and execution, which differ between brokers and accounts. Hypothetical results do not represent real trading. Use it on a demo account first, use only money you can afford to lose, and make your own decisions; nothing on this page is investment advice.