Why I open a new account every time I release a version

Live account #37376132 — RoboForex Ltd — real money — opened 17 July 2026 — report generated 28 August 2026.

Why I open a new account every time I release a version

Every time I ship a new version of an EA, I open a brand-new live account and fund it from zero.

I could keep one account running for three years, swap the EA on it four times, and present the combined curve as a single track record. That curve would look far better than this one. It would also be describing four different pieces of software.

So I don't do that. The rule is simple:

One EA version = one fresh account = one track record that belongs to that version alone.

The account below has never run anything except Apex Drawdown Zero V9. Not a signal. Not a manual trade. Not a second robot. The MetaTrader report proves it in one line:

Trades by robots: 43 — Trades placed manually: 0 — Trades from signals: 0

Whatever is in these numbers, the EA did it. The good and the bad.

The numbers: 17 July to 28 August 2026

Gross profit was $475.93. Gross loss was $210.64. The average winning trade was $20.69 and the average losing trade was $10.53 — a realised reward-to-risk ratio of 1.96 to 1.

That ratio is the mechanism of the whole system. The design target is not a high win rate. It is winners around twice the size of losers, with a hit rate slightly above a coin flip. Win rate is the number people quote. Expectancy is the number that determines the outcome.

Month by month:

• July 2026, from the 17th: +0.69%
• August 2026: +52.01%

The part that usually gets left out

Here is the losing side of the same account.

Result by symbol:

Gold produced effectively all of the result. EURUSD finished flat. EURJPY finished negative.

My reading of that, which you are free to disagree with:

Gold is where this system's logic fits best. The entry model is built on session ranges and price behaviour at range boundaries, and gold provides the volatility and session expansion that model depends on. It is the symbol I test most and the set file I would run first.

EURUSD and EURJPY behaved the way the risk controls are designed to make them behave in a market that offers nothing. Six weeks of summer FX was a low-volatility, range-bound period. Minus $5.01 across 19 EURUSD trades is not a strategy failing; it is a strategy declining to force trades and paying almost nothing to wait.

Six weeks is not a verdict on any of the three symbols. Eleven EURJPY trades is not a statistically meaningful sample and I am not going to present it as one. If EURJPY is still negative over a much longer period, that is a real result, and I will report it here the same way I am reporting this.

The equity curve was not a straight line.

July closed at +0.69%. The first three weeks were flat to slightly negative and equity dipped to $486.97, below the starting deposit. The move happened in August.

At the time of writing, the account is in a losing week: -3.35%, five trades, one winner.

The longest losing streak so far was 5 trades for -$31.78. The largest losing trade was -$53.10 on 31 July. The largest winning trade was +$153.84 on 21 August.

I am publishing this during a down week deliberately. An EA that is only written about after good months tells you nothing about how it behaves during bad ones.

Anyone running this should expect flat periods and losing weeks. A user who switches the EA off during the first flat stretch experiences the drawdown without the recovery.

How the system works

One qualified trade per day, maximum

The EA evaluates the session range, waits for price to interact with the boundary of that range, and opens a position only when behaviour at that boundary passes its filters. If a day produces no qualified setup, it does not trade.

That is why the account shows 43 trades across 42 days, with a number of zero-trade days in between.

Average trade duration is 15 hours 35 minutes. It is not a high-frequency scalper. It takes a position and holds it.

Stop loss attached at entry, on every position

A broker-side stop loss, placed when the position opens. Not a virtual stop held in the EA. Not a stop applied later.

What the system does not do

• No martingale.
• No grid.
• No averaging into losing positions.
• No recovery mode that increases lot size after a loss.
• No position without a stop loss.

That is the reason a 5-trade losing streak cost this account $31.78, and the reason maximum drawdown over the period was 13.48%.

Risk controls in V9

• Percentage-based position sizing, with an optional fixed lot cap set by the user.
• ATR-based volatility adaptation, so sizing and thresholds respond to current conditions.
• Volatility-relative entry thresholds — the main change introduced in V9.
• Automatic break-even management.
• Spread and broker-condition checks before entry.
• Margin and volume-limit validation.
• Broker stop-level verification, so orders are not sent that the broker will reject.
• A daily entry cutoff.

Guardian edition, included in the same package

For traders running funded or prop-firm accounts:

• Daily-loss halt with a configurable buffer.
• Maximum-drawdown halt.
• Target lock, which stops trading once an objective is reached.
• Optional Friday flat-close.
• Automatic initial-balance tracking.

These are mechanical limits that stop the EA at thresholds you set. They cannot prevent losses caused by gaps, slippage, execution or connectivity failures, and they are not a guarantee of any challenge outcome.

Specification

Included: Standard edition, Guardian edition, installation and settings guide, dashboard guide, FAQ, set files, updates, and support through MQL5 messaging.

The account documented above ran on $500, which is the minimum I suggest for live use. It was not funded with a large balance to make the drawdown percentages look smaller.

On my other published accounts

If you look through my published results you will find more than one account, so let me describe them rather than leave you to guess.

The account in this post is V9: +53.06% with 13.48% drawdown over 42 days.

An earlier live version ran from December 2025 to January 2026 on its own separate account, under the earlier naming. It gained more in percentage terms but did so with a 37.41% drawdown and an average trade length of 26 minutes. That version was considerably more aggressive. I stopped it in January and rebuilt the risk layer, and that rebuild is what became V9. V9 draws down less because I looked at that account and concluded the return did not justify the exposure.

I have also published an account for a different EA of mine, unrelated to this product, which lost almost its entire balance. It is a different strategy and a different codebase. I have left it published. Removing failed accounts would make every remaining account meaningless — results are only worth reading when the same person publishes the ones that went badly. That EA is also the reason V9 was built risk-layer first and entry-logic second.

How to check this yourself

The demo is free. I would rather you verify this than take my description of it.

1. Download the free demo from the product page.
2. Run it in the MetaTrader 5 Strategy Tester using the supplied XAUUSD set file, on real ticks, H1, over date ranges you choose. Build your own test rather than relying on mine.
3. Run it on a demo account for two weeks. You will see it skip days and take single trades. The inactivity is the filter working, not a fault.
4. Compare what your own testing produces against the figures in this post.
5. Message me through MQL5 with whatever question the above raises. Ask about EURJPY. Ask why August produced most of the result. Ask about the account that failed. I answer these personally.

If it does not convince you after that, do not buy it. An EA the user does not understand or trust gets switched off at the worst point in its cycle, which produces a worse outcome than never running it.

What I will publish next

• Updates on this same account at regular intervals, including periods where the result is negative, presented the same way as this one.
• Further work on the EURJPY logic based on live results rather than optimised history.
• Updates to the product remain available to existing owners at no additional cost.

Links

Live monitoring of account #37376132: MQL5 Signal

Product page and free demo: Apex Drawdown Zero V9 on the MQL5 Market

Questions: message me through MQL5.

Tshivhidzo Moss Mbedzi, MPC Labs, South Africa

Risk disclosure

Trading foreign exchange and commodities on margin carries a high level of risk and is not suitable for every investor. Past performance does not guarantee future results and is not indicative of future performance. The account described covers 42 days and 43 trades. That is a short sample, it is presented as a short sample, and nothing in this post should be read as a projection, forecast, promise or guarantee of any future result. Any EA can lose money, including this one. Risk-control features cannot eliminate losses arising from market gaps, slippage, broker execution or connectivity failure. Do not trade money you cannot afford to lose. Nothing in this post is investment advice.