[Technical Report] The "Beautiful Lie" of FX Black Ship cTrader and the Steering Wheel Usurpation by Process Design

■ Introduction: Expectations for Modern Architecture and the Harsh Reality

In the early morning, the inorganic main monitor冷たい displays the backtest results from Python that have been running since midnight — the countless corpses of logic — in a cold, merciless manner.

We, Semura Lab, realized the limits of the 32-bit aging vessel MT4 (legacy architecture) we had depended on for years, and introduced the next-generation “black ship” cTrader. Modern UI, visualization of depth of market, and a refined development environment with C# (cAlgo). We were confident that we would be liberated from the cursed MQL and achieve more precise validation and execution.

However, that engineering expectation would, within just a few days of operation, transform into “despair and exposure of structural flaws.”

■ 1. The Elusive Holy Grail Generated by the Backtest Engine

The backtest engine of cTrader was indeed faster and more sophisticated than MT4. On the screen, a magnificent upward-sloping equity curve (the so-called holy grail) could be drawn with ease. However, you cannot deceive the eyes of an process design engineer during an audit. That beautiful curve was composed of multiple “engineering lies” forged by internal platform specifications.

① “Future Prediction (CAP)” under the guise of tick data interpolation

The internal behavior of the seemingly precise 1-minute bars. But it did not reflect real market noise; it was a convenient interpolation to reduce system computational load, a “smooth illusion.” In real markets, stop losses (SL) that should certainly be filled at the wicks would not exist in the test environment and would magically reach take profit (TP). This is the same as the “front-running bias (CAP)” we hate most in system development.

② Concealment of physical frictions (spreads and slippage)

During early morning or at times of economic releases, reality is violent with spreads widening by several pips. But in internal tests, this physical friction was smoothed or undervalued. For us who have bled at spreads of 0.6 pips and razor-thin profits of 5–8 pips (risk-reward 1:1), that is a sterile, worthless theory produced in a clean room.

③ “Distortion of logic” occurring during translation to C#

Our core logic is built in a pure Python environment using Pandas and NumPy. To run autonomously on cTrader, it must be translated into C#, and in this process, “rounding errors in data” and minor timing misalignments in references inevitably occur. In other words, what was validated inside the platform was not the pure logic emitted by Python, but a degraded, different thing produced by translation.

[■ Insertion: an image here contrasting the “overly smooth unrealistic upward-sloping equity curve drawn on cTrader’s backtest screen” with the “equity curve reflecting harsh reality in Python”]