Strategy and research

• Rule-based: every entry, exit and sizing decision follows a fixed, documented rule set. There is no discretionary override.
• No martingale: position size is never increased to recover a previous loss.
• No grid: one position at a time; the systems never add positions at fixed intervals against an open trade.
• Defined risk: every order is sent with a stop-loss and a take-profit; the code refuses any order without both. Risk per trade is known before the order is sent.
• Execution guards: NOVA, QUANTIS and VECTOR X1 only buy while the daily close is above its 200-day average (trend filter) and skip entries when the spread is too wide; APEX only trades during the US cash session, closes everything before its close and skips entries when the spread is too wide.

NOVA is a trend-following system on Ethereum (ETHUSD), buy only, built for accounts from 500 USD. It uses the same engine as QUANTIS, with its own validated settings. It buys when an H4 close breaks above the highest high of the last 50 H4 bars, and only while the daily close is above its 200-day simple moving average. The initial stop-loss (3 × ATR) and take-profit (20 × ATR) are sent with the order; a trailing stop placed 4 × ATR below the highest high since entry only moves up, and the position is closed when the daily close falls below the 200-day average. One position at a time, never a short. Default risk is 2% per trade. The minimum Ethereum position of our test account is 0.01 ETH (about 25 USD): in our simulated backtests from January 2018 to September 2026, NOVA took all its signals from 500 USD at the default setting (55 trades out of 55, including 14 out of 14 since 2025), with +42.5%, a profit factor of 1.90 and a maximum drawdown of 12.0% at 500 USD; the percentage results stay practically the same up to 100,000 USD. On a 500 USD account, the license price can exceed the possible gains. The system trades rarely: 55 trades from January 2018 to September 2026. By calendar year, in the full backtest started with 10,000 USD (maximum drawdown 13.1% over the period): 2018, 2021, 2024, 2025 and 2026 (to 26 September) were positive, 2023 was a losing year (−4.3% of the balance at the start of the year); no trade in 2019, 2020 and 2022. In 2019 and 2020, Ethereum was below 750 USD and the average spread of our test account (about 4.1 USD in 2019, 2.7 USD in 2020) exceeded the 0.15% spread filter: NOVA did not enter any position (see Risk management > Maximum spread guard). In 2022, Ethereum was mostly below its 200-day average.

APEX is an intraday momentum system on the Nasdaq 100 (NAS100), buy only. It is based on the intraday momentum (noise area) model published by Zarattini, Aziz and Barbon (2024) for US index ETFs. Each day, the noise area measures how far the index usually moves from its open at each time of the session, over the last 14 days. Every 30 minutes during the US cash session (9:30-16:00 New York time), APEX buys if price has left the noise area upwards. The trailing stop sits at the noise boundary or at the session VWAP, whichever is tighter. Everything is closed 5 minutes before the cash close: no position is held overnight, so there is no swap. With the Standard profile (default), position size targets a 2% daily volatility, capped at 4× leverage and at 2% risk at the stop; the Dynamic profile (InpRiskProfile = 2) doubles these three values (4%, 8×, 4%) for 500 to 1,000 USD accounts, with a higher risk. The Max profile (InpRiskProfile = 3), high risk, multiplies them by four (8%, 16×, 8%): in our simulated backtests, its max drawdown is about 24%, against 6 to 7% with the Standard profile (details in the methodology below). Every order carries a stop-loss and a take-profit (3% of the entry price). In the default backtest (Standard profile, 10,000 USD), every calendar year was positive, 2025 only just (+1%); with the Dynamic profile, in the test started with 500 USD, 2025 was slightly negative (−12 USD, or −0.8% of the balance at the start of the year). Transparency point: the buy-only version was chosen after the validation period had been seen (the long and short version was flat over 2025-2026); the demo account follow-up will serve as the final validation.

QUANTIS is a long-term system on Bitcoin (BTCUSD), buy only. It buys when an H4 close breaks above the highest high of the last 20 H4 bars, and only while the daily close is above its 200-day simple moving average. Stop-loss at 5 × ATR, take-profit at 16 × ATR, trailing stop at 3 × ATR; the position is closed when the trend is lost. Default risk is 2% per trade, with about 5,000 USD of capital needed; at 4% (InpRiskPct = 4), about 2,000 USD is enough, with deeper drawdowns. While Bitcoin is below its 200-day average, QUANTIS does not trade (for example in 2019 and 2022).

VECTOR X1 is a trend-following system on gold (XAUUSD), buy only. It buys when an H4 close breaks above the highest high of the last 80 H4 bars, and only while the daily close is above its 200-day simple moving average. The stop-loss (5 × ATR) and take-profit (20 × ATR) are sent with the order; a trailing stop placed 3 × ATR below the highest high since entry only moves up, and the position is closed when the daily close falls below the 200-day average. One position at a time, never a short. The aggressiveness level, from 1 to 5, sets the risk per trade at 1, 2, 3, 4 or 5% of the balance (level 3 by default). The system trades rarely: 109 trades from December 2018 to September 2026, about 13 a year. Gold is expensive per lot: with the minimum position of our test account (0.01 lot = 1 ounce), it needs about 5,000 USD at level 3, 3,000 USD at level 5 and 25,000 USD at level 1; below that, the EA skips the signal rather than oversizing the position. In the backtest, 2021 and 2022 were losing years.

All four systems send a stop-loss and a take-profit with every order (the code refuses any order without both), hold one position at a time and use neither a grid nor a martingale: lot size is calculated from the risk and the stop distance, and is never increased to recover a loss. A daily loss limit applies to each of them.

1. 1

   Explicit rules

   NOVA, QUANTIS and VECTOR X1 use an H4 trend breakout filtered by the 200-day simple moving average (same engine, settings specific to each system); APEX follows the intraday momentum model published by Zarattini, Aziz and Barbon (2024). The rules of each system are described above and on its product page.

2. 2

   Data and costs

   Backtests use data from an MT5 broker (standard account, no commission, 1:100 leverage), with the variable spread from that data and swaps included. Slippage is not simulated.

3. 3

   Real and reconstructed ticks

   On this server, real ticks are available from June 2025 only. Before that date, MetaTrader reconstructs ticks from 1-minute bars. 2026 results are 100% real ticks.

4. 4

   In-sample and out-of-sample periods

   Settings are chosen on the in-sample period only, up to 31 December 2024. The period from 1 January 2025 to 26 September 2026 is the out-of-sample validation and was never used to choose settings, with one exception disclosed above: the buy-only version of APEX was chosen after this period had been seen.

5. 5

   Acceptance criteria set in advance

   Three criteria are fixed before seeing the results: net profit above zero, profit factor of at least 1.10, and maximum drawdown no more than twice the in-sample drawdown.

6. 6

   NOVA: validation window and caveats

   NOVA's settings were chosen on 2018-2024 only, then tested once on the validation window, January 2025 to September 2026 (real ticks from June 2025), with an additional criterion set in advance: with 500 USD, take at least 95% of the trades taken with 100,000 USD. Simulated results over that window: at 1% risk per trade, +5.5%, profit factor 1.97, maximum drawdown 3.2%, 14 trades; with 500 USD at 2% (default setting), +10.4%, maximum drawdown 5.9%, 14 trades out of 14. Two caveats: the validation window holds only 14 trades, too few to draw a firm conclusion; and the QUANTIS (Bitcoin) settings did not work on Ethereum (slightly negative result over 2018-2024): NOVA has its own settings, chosen on 2018-2024. Past performance, real or simulated, is not indicative of future results.

7. 7

   Capital tests

   Each system is also tested with deposits from 500 to 100,000 USD, over the same periods and with the same strategy: only the deposit and the risk setting change. This measures the share of the signals since 2025 that the EA still takes at each deposit (capital table in Risk management).

8. 8

   APEX: Max profile (high risk)

   The Max profile (InpRiskProfile = 3) runs the same APEX strategy with four times the exposure and the risk of the Standard profile. It is a high-risk setting. Simulated MetaTrader 5 backtests, January 2019 to September 2026: with 500 USD, +1,653.4% over the period (gains reinvested), profit factor 1.37, maximum drawdown 24.1%, 1,248 trades out of 1,257 (266 out of 267 since 2025); with 10,000 USD, +1,928.6%, maximum drawdown 24.5%. The profile stops at four times: with 8 times the exposure of the Standard profile, the maximum drawdown reached 50.8% (about half the account, lost between mid-2024 and early 2026). Slippage is not simulated, and its effect grows with position size. Past performance, real or simulated, is not indicative of future results.

9. 9

   Parameter stability

   Settings are picked on a plateau, where neighbouring settings also perform well, rather than from the single best optimization pass.

10. 10

    Many ideas rejected

    Many ideas are discarded. For example, about 1,100 Forex scalping variants were tested and rejected: none remained robust once real costs were included.

11. 11

    What is not done

    No walk-forward analysis, no Monte Carlo resampling, no commission or slippage simulation. No live or demo track record is published yet: a demo account follow-up, published on Myfxbook, is planned but has not started.

• Regime changes: market structure, volatility and correlations change. Rules that worked in one regime can underperform in another.
• Costs: spreads, commissions and swaps differ between brokers and account types, and directly affect results. Our backtests include no commission.
• Slippage and execution: live fills can differ from tested prices, especially around news and at session opens. Our backtests do not simulate slippage.
• Data and modelling: backtests depend on the quality of historical data and on modelling assumptions. Before June 2025, our backtests rely on ticks reconstructed from 1-minute bars.
• Drawdowns: losing periods will occur, and future drawdowns can be deeper or longer than any observed historically.