RoboXpert BTC Weekend · BTCUSD Expert Advisor for MetaTrader 5
Max. equity drawdown in this test 30.74%
A backtest, not a live account. Past drawdown is not a loss limit.
MT5 Strategy Tester · 1 July 2022 – 16 October 2026 · Binance second-by-second prices with a fixed $6 spread (what a broker with a $6 spread charges since August 2026) · tested 28 October 2026. Risk level high: one stopped series costs about 11% of the balance; low +768% (drawdown 8.9%), medium +6,203% (drawdown 17.3%) in the same test.
Weekend of 10 Oct 2026: +34,931% ($3,503,143); lowest equity $3,492,531; BTC $83,639
Our chart of the MT5 backtest from 1 July 2022 – 16 October 2026: the blue line is the balance after each weekend, red marks show how far open positions pulled equity below it during that weekend, the faint line is Bitcoin’s price when trading stopped. Rebuilt by us from the report’s deal list and the price bars of the test data (Binance prices with a $6 spread). Backtest, not a live account; not a forecast.
In the Strategy Tester
One run at the high level in the MetaTrader 5 Strategy Tester. The video starts on Saturday 21 January 2023 at $13,690, as the EA builds a short series of nine positions and turns long right after, then fast-forwards through the year to 6 January 2024 and ends with $78,032 on the account.
Risk levels
Only the position size changes: one stopped series costs about 3%, 6%, 11% of the balance. The EA works out the start lot from that and the current Bitcoin price. You pick the level in the EA’s settings. Larger positions grow faster on good weekends and lose more on bad ones.
The settings you get. Each card is the level as it is built into the EA: same rules, same setting.
3% per stopped series
First start lot on $10,000: 0.05 BTC
Minimum capital about $8,500; with less, a cent account
Largest order 1.15 BTC.
6% per stopped series
First start lot on $10,000: 0.10 BTC
Minimum capital about $4,300; with less, a cent account
Largest order 16.43 BTC. Some brokers take at most 10 BTC per order; with that limit the same test made +25,905%.
11% per stopped series
First start lot on $10,000: 0.19 BTC
Minimum capital about $2,300; with less, a cent account
Largest order 20 BTC. Some brokers take at most 10 BTC per order; with that limit the same test made +25,905%.
Each level is its own Strategy Tester run with the same rules, data and period as the detailed test below (1 July 2022 – 16 October 2026, $10,000 start). Lots grow with the balance, so the gap between the levels widens over time. In fast moves a stopped series costs more than its level: the costliest series on each card shows by how much. Large weekend orders can fill worse live than in the tester.
The cards above use a $6 spread. Here are the same three levels at wider spreads. Every entry pays the spread, so at high risk +34,931% at $6 becomes +1,601% at $25.
Spreads we measured per BTC: $5–6 and $15 at two brokers on 21 October 2026, $18.50 at a third on the live weekend of 24–25 October 2026; $25 stands for an expensive broker. Each card uses one fixed spread for the whole period, with the same rules, Binance prices and period as the cards above (1 July 2022 – 16 October 2026, $10,000 start, without the crash weekend of 11–12 October 2025). The wider the spread, the less is left, because every position of a series pays it. At $25 the EA opened its first series only on 28 October 2023: until then Bitcoin was mostly below $31,250, where $25 is above its spread limit; the figure per year still counts the whole period. Spreads widen in fast moves; the EA opens nothing while the spread is above 0.08 % of the price ($80 at $100,000).
The backtest
One run of RX_BtcWeekend 3.15 on BTCUSD M5 in the MetaTrader 5 Strategy Tester, mode “Every tick based on real ticks”, $10,000 deposit, leverage 1:500, 50 ms execution delay. The ticks are Binance BTCUSDT prices second by second in trade-server time, widened by a fixed $6 spread: what the cheapest broker we measured has quoted for Bitcoin since August 2026 (before that $12). The test assumes today’s spread for the whole period; the cards under the risk levels show the same test at $10, $15, $18.50 and $20. Run by us on 28 October 2026, with exactly the settings you get.
Why the test starts on 1 July 2022: since the second half of 2022 Bitcoin’s weekends have swung about half as much as in 2021 and early 2022, and deep one-way moves have become much rarer (our count on Binance second-by-second prices since 2020). We decide on the market as it has been since then; the older years stay with us as a stress test.
Why not a broker’s own ticks: none of the brokers we tested holds complete real Bitcoin weekend ticks since 2022 (one only since April 2026, the others miss April 2025 to March 2026 or May 2024 to January 2026), and the tester silently fills missing days with generated ticks. On the brokers’ own ticks the same high-risk settings made +65.8% at a broker with a $6 spread (25 Apr 2026 – 16 Oct 2026, drawdown 16.3%) and +25.1% at a broker with a spread of about $18.50 (29 Jan 2026 – 16 Oct 2026, drawdown 38.7%). That broker’s prices moved differently from Binance’s in February and March 2026 and set off more stops, so a real account can do clearly worse than this test.
The headline includes the spread but no extra costs. With 0.01% of the price added to every fill (about $10 per BTC at $100,000) the same period made +10,407% with a 42.3% drawdown; with 0.02% it made +3,834%. The grid trades about 6,200 series, so your broker’s costs decide a large share of the result.
One weekend is left out: 11–12 October 2025, when Bitcoin swung by about $15,000 within hours of the Saturday open. With a fixed spread the grid earned so much on it that the test reaches +145,097% with that weekend included. A broker would widen spreads in such a move, so we treat that weekend as a stress test, not as the headline.
Monthly results are the change of the balance within each month, rebuilt from the report’s deal list. A backtest is not a live or demo-forward result: live trading adds slippage, outages and your broker’s own crypto conditions.
Since 23 October 2026 a public live account has run the EA at the high level, linked at the top of the page. Its first weekends say little next to 51 months of backtest.
Why the weekend
On weekends Bitcoin usually swings back and forth in a narrow band. We tested each part of the weekend on its own with the same rules, and the weekdays as well: weekday hours, above all the US session, broke the grid.
Net+1,865%
Max. equity drawdown33.4%
Net+5,770%
Max. equity drawdown25.2%
Net+34,931%
Max. equity drawdown30.7%
Same rules in every test (stop at level 10), $10,000 start, 1 July 2022 – 16 October 2026. Each card uses the same trading window as the EA: Saturday from 00:00, Sunday until 23:30, without a break at midnight. The EA stops at 23:30, half an hour before the day change of many brokers. Until version 3.14 it stopped at 20:00; in our tests since mid-2022 the later end earned more at the same drawdown at low, medium and high risk (fixed spreads from $6 to $20), together with closing deep series at ±0.
The strategy
Once Bitcoin has moved a set share of the price away from its reference, the EA opens a position against the move. If the move goes on, it adds larger positions at set distances. It closes the whole series once price is a little past the series’ average entry, and the next reference starts there. From the 8th position on, a series closes as soon as price is back at its average entry (±0) instead of waiting for the full target; that is on by default and can be switched off. All distances are percent of the price, so the EA behaves the same at $30,000 and at $200,000.
BTCUSD from the test data (Binance prices with a $6 spread), 5-minute bars (range and close), trade-server time. Each dot is one position the EA opened; its size shows the lot size. The dashed line is the series target after each entry, a little past the average entry price.
3,269 of 6,232 series closed with their first position. Only 34 reached level 10.
Crash and stop
On 7 Feb 2026 Bitcoin fell more than $800 in about 10 minutes. The series reached 6 positions, about 107 BTC; the test symbol then refused the next add (volume limit), the EA counted the series as full, and one more step down the stop closed it at −$54,638. The next series traded the rebound, and the weekend still ended at +3.9%.
BTCUSD from the test data (Binance prices with a $6 spread), 5-minute bars (range and close), trade-server time. Each dot is one position the EA opened; its size shows the lot size. The dashed line is the series target after each entry, a little past the average entry price.
On 2 Mar 2025 one series grew to 26 positions and lost $959,982. The account still ended at +46,552%, but at its deepest it had lost most of its value.
Same period with 0.01% of the price added to every fill: drawdown 42.3%. With 0.02% it was +3,834%. The spread is already in the headline.
The whole test including 11–12 October 2025, when Bitcoin swung about $15,000 within hours. With a fixed spread the grid profited hugely; a broker would have widened spreads.
On 26 Oct 2025 a series reached the stop and cost 5.5% of the balance. The stop fired 123 times in this test; nothing stays open into the week.
The stop fires when a series holds 10 positions and price moves one more step against it. Per 0.01 BTC starting lot that loses about 0.303% of the Bitcoin price: $253 at $83,639.
The original report
Charts exactly as MetaTrader 5 saved them, for each risk level from its own run.
Saved by MetaTrader 5 with the report of the high level. The x-axis counts deals, not days. About 1.8 times the position size of medium: the balance grows faster, and the stopped series of 2025 and 2026 cut deeper in dollars.
Every test run
Pick the spread closest to your broker's and your risk level: every year on its own and the whole test, with the exact settings to download.
Typical Bitcoin spreads we measured in October 2026: about $5–6 at the cheapest broker, about $15–20 at many others ($18.50 on our own live account). $25 shows an expensive broker. Pick the spread closest to your broker's.
Binance prices$6 spread
Low · 3%v3.15
Net result+768.4%$10,000 → $86,841
Largest drop 8.9%
Year by year
The crash weekend of 11/12 October 2025 is left out: with a fixed spread it gives the grid unrealistic gains, because a real broker widens the spread and the EA's spread limit blocks new series.
The years are parts of this one run: each year's figure is the change of the balance from 1 January to 31 December (2022: from 1 July, when the test starts; 2026: to the end of the test). The risk level sizes from the current balance, so each year stands on its own.
Binance BTCUSDT prices second by second, shifted to the broker's server time (GMT+2/+3) and widened by the spread shown. No broker has complete real Bitcoin weekend ticks since 2022.
EA version 3.15: The window ends on Sunday at 23:30 instead of 20:00 (UTC+2, UTC+3 in US summer time); in 3.15 the end could be set, since 3.16 it is fixed at 23:30. Close deep series at ±0 can no longer be switched off (from the 8th, 9th or 10th position).
2026-10-28-btcv315-ic6y-h22-lowInputs of this run
differs from the default
Settings file (.set)Binance prices$6 spread
Medium · 6%v3.15
Net result+6,203%$10,000 → $630,279
Largest drop 17.3%
Year by year
The crash weekend of 11/12 October 2025 is left out: with a fixed spread it gives the grid unrealistic gains, because a real broker widens the spread and the EA's spread limit blocks new series.
The years are parts of this one run: each year's figure is the change of the balance from 1 January to 31 December (2022: from 1 July, when the test starts; 2026: to the end of the test). The risk level sizes from the current balance, so each year stands on its own.
Binance BTCUSDT prices second by second, shifted to the broker's server time (GMT+2/+3) and widened by the spread shown. No broker has complete real Bitcoin weekend ticks since 2022.
EA version 3.15: The window ends on Sunday at 23:30 instead of 20:00 (UTC+2, UTC+3 in US summer time); in 3.15 the end could be set, since 3.16 it is fixed at 23:30. Close deep series at ±0 can no longer be switched off (from the 8th, 9th or 10th position).
2026-10-28-btcv315-ic6y-h22-mediumInputs of this run
Binance prices$6 spread
High · 11%v3.15
Net result+34,931%$10,000 → $3,503,143
Largest drop 30.7%
Year by year
The crash weekend of 11/12 October 2025 is left out: with a fixed spread it gives the grid unrealistic gains, because a real broker widens the spread and the EA's spread limit blocks new series.
The years are parts of this one run: each year's figure is the change of the balance from 1 January to 31 December (2022: from 1 July, when the test starts; 2026: to the end of the test). The risk level sizes from the current balance, so each year stands on its own.
Binance BTCUSDT prices second by second, shifted to the broker's server time (GMT+2/+3) and widened by the spread shown. No broker has complete real Bitcoin weekend ticks since 2022.
EA version 3.15: The window ends on Sunday at 23:30 instead of 20:00 (UTC+2, UTC+3 in US summer time); in 3.15 the end could be set, since 3.16 it is fixed at 23:30. Close deep series at ±0 can no longer be switched off (from the 8th, 9th or 10th position).
2026-10-28-btcv315-ic6y-h22-highInputs of this run
differs from the default
Settings file (.set)Binance prices$18.50 spread
Low · 3%v3.15
Net result+183.5%$10,000 → $28,351
Largest drop 12.1%
Year by year
The crash weekend of 11/12 October 2025 is left out: with a fixed spread it gives the grid unrealistic gains, because a real broker widens the spread and the EA's spread limit blocks new series.
The years are parts of this one run: each year's figure is the change of the balance from 1 January to 31 December (2022: from 1 July, when the test starts; 2026: to the end of the test). The risk level sizes from the current balance, so each year stands on its own.
Binance BTCUSDT prices second by second, shifted to the broker's server time (GMT+2/+3) and widened by the spread shown. No broker has complete real Bitcoin weekend ticks since 2022.
EA version 3.15: The window ends on Sunday at 23:30 instead of 20:00 (UTC+2, UTC+3 in US summer time); in 3.15 the end could be set, since 3.16 it is fixed at 23:30. Close deep series at ±0 can no longer be switched off (from the 8th, 9th or 10th position).
2026-10-28-btcv315-sp-s185y-h22-lowInputs of this run
differs from the default
Settings file (.set)Binance prices$18.50 spread
Medium · 6%v3.15
Net result+1,232%$10,000 → $133,202
Largest drop 21.1%
Year by year
The crash weekend of 11/12 October 2025 is left out: with a fixed spread it gives the grid unrealistic gains, because a real broker widens the spread and the EA's spread limit blocks new series.
The years are parts of this one run: each year's figure is the change of the balance from 1 January to 31 December (2022: from 1 July, when the test starts; 2026: to the end of the test). The risk level sizes from the current balance, so each year stands on its own.
Binance BTCUSDT prices second by second, shifted to the broker's server time (GMT+2/+3) and widened by the spread shown. No broker has complete real Bitcoin weekend ticks since 2022.
EA version 3.15: The window ends on Sunday at 23:30 instead of 20:00 (UTC+2, UTC+3 in US summer time); in 3.15 the end could be set, since 3.16 it is fixed at 23:30. Close deep series at ±0 can no longer be switched off (from the 8th, 9th or 10th position).
2026-10-28-btcv315-sp-s185y-h22-mediumInputs of this run
Binance prices$18.50 spread
High · 11%v3.15
Net result+7,129%$10,000 → $722,883
Largest drop 39.8%
Year by year
The crash weekend of 11/12 October 2025 is left out: with a fixed spread it gives the grid unrealistic gains, because a real broker widens the spread and the EA's spread limit blocks new series.
The years are parts of this one run: each year's figure is the change of the balance from 1 January to 31 December (2022: from 1 July, when the test starts; 2026: to the end of the test). The risk level sizes from the current balance, so each year stands on its own.
Binance BTCUSDT prices second by second, shifted to the broker's server time (GMT+2/+3) and widened by the spread shown. No broker has complete real Bitcoin weekend ticks since 2022.
EA version 3.15: The window ends on Sunday at 23:30 instead of 20:00 (UTC+2, UTC+3 in US summer time); in 3.15 the end could be set, since 3.16 it is fixed at 23:30. Close deep series at ±0 can no longer be switched off (from the 8th, 9th or 10th position).
2026-10-28-btcv315-sp-s185y-h22-highInputs of this run
differs from the default
Settings file (.set)Binance prices$25 spread
Low · 3%v3.16
Net result+126.2%$10,000 → $22,616
Largest drop 12.5%
Year by year
The crash weekend of 11/12 October 2025 is left out: with a fixed spread it gives the grid unrealistic gains, because a real broker widens the spread and the EA's spread limit blocks new series.
The years are parts of this one run: each year's figure is the change of the balance from 1 January to 31 December (2022: from 1 July, when the test starts; 2026: to the end of the test). The risk level sizes from the current balance, so each year stands on its own.
Binance BTCUSDT prices second by second, shifted to the broker's server time (GMT+2/+3) and widened by the spread shown. No broker has complete real Bitcoin weekend ticks since 2022.
EA version 3.16: Trades like 3.15 with its preset settings; the window end on Sunday is fixed at 23:30 (no setting).
2026-10-28-btcv316-sp-s25y-h22-lowInputs of this run
differs from the default
Settings file (.set)Binance prices$25 spread
Medium · 6%v3.16
Net result+304.4%$10,000 → $40,444
Largest drop 23.7%
Year by year
The crash weekend of 11/12 October 2025 is left out: with a fixed spread it gives the grid unrealistic gains, because a real broker widens the spread and the EA's spread limit blocks new series.
The years are parts of this one run: each year's figure is the change of the balance from 1 January to 31 December (2022: from 1 July, when the test starts; 2026: to the end of the test). The risk level sizes from the current balance, so each year stands on its own.
Binance BTCUSDT prices second by second, shifted to the broker's server time (GMT+2/+3) and widened by the spread shown. No broker has complete real Bitcoin weekend ticks since 2022.
EA version 3.16: Trades like 3.15 with its preset settings; the window end on Sunday is fixed at 23:30 (no setting).
2026-10-28-btcv316-sp-s25y-h22-mediumInputs of this run
Binance prices$25 spread
High · 11%v3.16
Net result+1,601%$10,000 → $170,145
Largest drop 42.4%
Year by year
The crash weekend of 11/12 October 2025 is left out: with a fixed spread it gives the grid unrealistic gains, because a real broker widens the spread and the EA's spread limit blocks new series.
The years are parts of this one run: each year's figure is the change of the balance from 1 January to 31 December (2022: from 1 July, when the test starts; 2026: to the end of the test). The risk level sizes from the current balance, so each year stands on its own.
Binance BTCUSDT prices second by second, shifted to the broker's server time (GMT+2/+3) and widened by the spread shown. No broker has complete real Bitcoin weekend ticks since 2022.
EA version 3.16: Trades like 3.15 with its preset settings; the window end on Sunday is fixed at 23:30 (no setting).
2026-10-28-btcv316-sp-s25y-h22-highInputs of this run
differs from the default
Settings file (.set)Backtest in the MetaTrader 5 Strategy Tester. Past results, no forecast; live trading adds slippage, outages and your broker's own costs.
Included
RoboXpert BTC Weekend 3.16 as an .ex5 file: the tested rules on this page, with the licence check and the status panel.
Low, medium or high in one setting: the same levels as the runs on this page.
Bound to your MT5 account number. The EA checks it online.
One line shows the state and the next step, in English or German. It warns when the balance is too small for the level.
A PDF with every step: installation, settings, VPS and what each status means.
We check spread, crypto margin and contract terms with you before you start.
The EA runs on desktop MT5 over the weekend. You choose the account and the risk level and keep the terminal connected from Saturday to Sunday evening.
Talk through the requirementsLow, medium or high, chosen in the settings; medium is the default. The EA sizes the start lot so that one stopped series costs about that share of the balance at the current Bitcoin price, and keeps it for the whole series. The detailed test ran high: $10,000, so 0.19 BTC at the start.
The smallest order is 0.01 BTC, and a stopped series costs about $253 per 0.01 BTC of start lot at a Bitcoin price of $83,639 (0.303% of the price). With less money than these amounts (low · medium · high) one stopped series costs more than the level says.
At level 10 a series holds 0.43 BTC per 0.01 BTC start lot. The tester used 1:500; brokers often set their own crypto margin. We check it with you.
A cent account counts the balance in US cents: $85 count as 8,500 cents. If the BTCUSD contract terms there fit, the EA sizes its series as on a larger account. Cent accounts can have wider Bitcoin spreads than standard accounts, and a wider spread means less return; the spread cards under the risk levels show how much. We check the contract terms with you before you start.
Algo Trading on, terminal connected over the weekend. Hosting is a separate cost. The phone app is for monitoring.
Your next step
You get the EA, the setup guide in English or German and the link to our partner broker.
Open your account through that link and send us your MT5 account number. We check that it was opened through our link, activate it and let you know.
Pick the risk level and switch on Algo Trading. The status panel on the chart shows Ready and when trading starts, in the time of your computer.
Direct message to @RoboXpertAI
Message @RoboXpertAI on Telegram. You get the EA, the setup guide and the link to our partner broker; open your account through it, send us the account number and we activate it. On a demo account at any broker the EA runs for 30 days without activation, so you can try it first.
No licence fee: access runs through our partner-broker arrangement, and we receive broker rebates on referred trading activity. Spreads, commissions, swaps and VPS hosting can still cost money.
Yes, since 23 October 2026: a public live account runs the EA at the high level, linked to Myfxbook at the top of this page. Its first weekends say little next to the backtest, so every other figure here comes from the MetaTrader 5 Strategy Tester. The backtest ticks are Binance prices second by second with a fixed $6 spread, what the cheapest broker we measured has quoted since August 2026, because no broker we tested has complete real Bitcoin weekend ticks since 2022. On one broker’s own ticks from February to October 2026 the high level made far less than here and went through a deep drawdown; the details are under “What this backtest does and does not establish”. Live trading adds slippage, outages and your own broker’s spreads.
On Saturdays and Sundays Bitcoin mostly swings back and forth in a narrow band, which is what the grid needs. We tested weekday hours on six years of data as well: every weekday window we added made the result worse, and nights and the US session wiped out the test account in 2024–2026.
Yes, including your entire deposit. A series loses about 0.303% of the Bitcoin price per 0.01 BTC starting lot when the stop fires ($253 at $83,639), and a series still open on Sunday at 23:30 is closed at a loss. The costliest series in this test lost $103,583 on 26 Oct 2025, 5.5% of the balance.
On 7 Feb 2026 Bitcoin fell by more than $800 in about 10 minutes. The series reached 6 positions and the stop closed it at −$54,638. Without the stop, the same data built a series of 26 positions on 2 Mar 2025 that lost $959,982. That is why the stop at level 10 is always on.
Start with the level whose costliest series you could accept. In the backtest that series took 4.3% (low), 8.5% (medium), 15.7% (high) of the account. Higher levels made more in this test because lots grow with the balance; they also trade larger orders that can fill worse live. You pick the level in the EA’s settings; we talk it through with you.
The EA sizes each series so that one stopped series costs about 3%, 6%, 11% of the balance (low, medium, high). The smallest order is 0.01 BTC, so at a Bitcoin price of $83,639 that works from about $8,500, $4,300, $2,300; with less money a stopped series costs more than the level says, so smaller accounts run on a cent account (balance in US cents). Crypto margin is set by your broker and can differ from your account leverage; we check both with you.
No. The broker account stays in your name. You fund it, withdraw and decide whether algorithmic trading runs. The EA closes its own positions by Sunday 23:30 (UTC+2, UTC+3 in US summer time) at the latest; switching it off does not close open positions.
Versions
Newest first. The EA you get is always the newest version.