Gold Spark

Extensive Developer Support, but Severe Technical Flaws and Suspected Over-Optimization I purchased Gold Spark on July 30, 2026. After nearly two months of rigorous backtesting, forward testing, and live testing, I am sharing my experience. While the developer is highly responsive and willing to communicate, the EA itself suffers from fundamental logic errors and what appears to be severe curve-fitting. The Positive: Developer Communication The only consistently positive aspect of this experience has been the developer's responsiveness. I reported numerous bugs, and he acknowledged them, releasing multiple updates (V1.60 to V1.70+) and providing his personal set files. He is polite and willing to engage. The Technical Issues (Based on real testing and developer communication): 1. Non-Functional Risk Controls (V1.60): Initially, changing drawdown limits (limit_dd_max_val, max_dd_stop_trade) had zero impact on backtest results. Two runs with vastly different risk parameters (20%/25% vs. 8%/10%) produced identical net profits, drawdowns, and trade counts. The developer later confirmed these had "potential bugs" and updated the logic. 2. Volume Cap Failure: In testing the developer's own provided set file, position sizes grew to an alarming 466-481 lots per trade. This occurred despite the vol_max_val being capped at 80 lots. The cap was simply not enforced. 3. Overlapping Trades: In V1.70, I discovered multiple overlapping positions (e.g., three trades open simultaneously on Jan 3, 2024, two in the same direction), which contradicts standard grid/martingale risk management. 4. SL Calculation Discrepancies: Using the exact same set file, my live account experienced Stop Loss hits on trades where the developer's MyFxBook account did not. The SL values were drastically different, indicating the EA calculates SL inconsistently depending on the broker/environment. 5. "Phantom" Trades & Negative Lot Sizes: The EA frequently logs LOT DEBUG BUY... Lot=-1.0000 without executing a trade. When it does execute, the lot size is positive. This indicates a critical calculation error at the signal stage, causing missed entries. Additionally, the EA logs a Drawdown error (Error:4014) and refuses to open new trades even when the account has only profitable positions open and zero drawdown. 6. Live vs. Backtest Disconnect: In recent versions, the EA trades in the Strategy Tester but fails to execute trades on a live account, providing no error logs as to why. 7. Lack of Strategy Entry Parameters (Critical Flaw): The EA provides a large number of input parameters, but none of them relate to the core strategy entry logic. Every available setting controls risk, money management, or basic trade management—not how or why the EA enters a trade. This means the EA is essentially a black box. Users cannot test different entry conditions, adjust the strategy to different market regimes, or verify if the entry logic has any edge. You are forced to accept the developer's fixed, undisclosed entry algorithm with no ability to adapt it. The Major Red Flag: Equity Curve Divergence The most concerning issue is the performance timeline. Pre-Release: The EA shows a smooth, highly profitable equity curve leading up to early May 2026. Post-Release: After early May 2026, the equity curve flattens out and drops significantly. The Reality: The EA was published on June 25, 2026. The transition from a beautiful backtest curve to poor live/forward performance happens exactly at the point where the data transitions from historical backtesting to post-release market conditions. No matter what settings I adjusted, I could not replicate the pre-May performance in the post-May market. This strongly suggests the EA is over-optimized to historical data (curve-fitted) and fails to adapt to current market dynamics. Conclusion: The developer is responsive and tries to fix bugs, but the core architecture of the EA appears flawed. With non-functional volume caps, overlapping trades, calculation errors resulting in missed trades, no access to entry logic for proper testing, and a highly suspicious equity curve that falls apart right after the release date, I cannot recommend this EA for live trading. It is currently not usable in its present state.

I consider Gold Spark a highly promising EA that offers exceptional flexibility in tuning and demonstrates "monster-class" potential depending on the settings used. While the default configuration operates as a single-shot EA, it also supports holding multiple simultaneous positions—including both long and short trades—which is truly the standout feature of this system. Unlike most multi-position grid EAs, which rely on "negative grid" strategies like averaging down (Nanpin) or Martingale, this EA takes a distinctly different approach. It functions as a "positive grid" EA that adds to positions when market movements are favorable; its robustness and profit-generating potential are in a league of its own compared to typical grid EAs on the market. Naturally, it supports features such as trailing stops, break-even settings, partial closes upon the start of a trailing stop, and compatibility with prop firm requirements, allowing for detailed customization to suit individual preferences. Offered at a very reasonable price, its performance potential is truly top-tier. I highly recommend it, particularly to intermediate and advanced traders who are comfortable verifying optimal settings through backtesting.