[Episode 2] CHAOS_GOLD Why the "Anomaly Method"──Evidence hidden at 21:00, 18:00, and 04:00
“The average lifespan of commercial EAs is about six months.”
This is a commonly cited figure in the investment navigator industry. An EA that posted superb backtest results can have its profit curve crumble and stop being used once it starts running in real trading for even half a year.
In such a world, CHAOS_GOLD_LISKOFF_MT5has achieved net profit of 6.6 million yen, PF 2.96, and RF 11.49 over seven years of backtesting (2019–2026), and continues to operate on real accounts today. Why was this stability achievable? The key lies in the architecture called the “Anomaly Method,” which is rare in the industry.
CHAOS_GOLD does not use RSI or MACD; it operates only at three timeframes in JST: 9 PM, 6 PM, and 4 AM. This is a rare approach in the industry that does not rely on technical indicators.
In this second installment, we will explain in an accessible way why many EAs fail within half a year, while CHAOS_GOLD has survived seven years, revealing the core identity of its anomaly method.
An anomaly is a phenomenon where, inside markets that should move randomly,statistically biased tendencies are observed. In translation, it conveys ideas like “irregularity” or “bias.”
Representative gold market anomalies have been reported in academic papers and industry reports, such as:
- “Gold tends to rise on Tuesdays”
- “Early in the month, first few trading days tend to be strong”
- “In the hours before NY closes, directionality tends to emerge”
- “In the 30 minutes after releases of indicators, movement toward a specific direction is likely”
These are not absolute laws. They are probabilistic biases, and there are days when they fail. However, when observed over long periods, persistent biases remain that cannot be explained by randomness alone. Academically, these are studied as “by-products of market structure” arising from factors like investor psychology, participant turnover, and economic indicator cycles.
CHAOS_GOLD uses this concept of “time window × direction” anomaly.Seven years of gold market data revealed that movements in specific directions tend to dominate in particular time windows, leading to an entry design that only operates in those windows.
The three time windows CHAOS_GOLD employs—9 PM, 6 PM, 4 AM—are all moments when participants in the gold market transition. When overlapped with global market hours, the reason these three are used becomes visually understandable.
Japan time 18:00 is just after the London market opens. Until then, the market lacks direction due to Tokyo’s mid-day lull, but with European participants entering, liquidity surges. Gold becomes one of the most actively traded assets in the London market, and statistically, trends tend to start in this time window.
Many US economic indicators are released at 21:30 JST. The market begins to price these in just before, and volatility expands rapidly. The overlap between London and NY from 21:00 to 2:00 is called gold’s golden time, where most daily price movement concentrates.
Japan time 4:00 is just before NY closes at 5:00. This is when US traders adjust positions for the next day, with statistical bias toward certain directions. It’s also near the start of Asia time, a market “gap” zone, making it easier for entry signals to appear. Among the three windows, it’s the least accessible for Asian traders, but ideal for automated trading to capitalize on.
Technical-indicator-based EAs trade on price movement. While powerful, they have structural weaknesses that give rise to the reality that the average lifespan of commercial EAs is about six months.
The biggest weakness of technical-indicator-type EAs isproneness to over-optimization (curve fitting)— If you fix RSI period to a certain value and MACD settings to another, past data can look spectacular, but the model often fails as soon as the validation window shifts to live trading.
Anomaly type reasons have inputs that are fixed elements—“time” and “day of week”—that stay constant from past to future. The moment market participants swap in that time window will persist in the same places unless the market structure changes fundamentally.Low structural risk that indicator effectiveness changes over time. This is the structural reason CHAOS_GOLD has endured seven years.
Entry opportunities are limited to a maximum of three per day, but the philosophy is to improve the accuracy of each entry. CHAOS_GOLD focuses on precise entries rather than sheer volume.
We will organize three aspects of how anomaly methodology concretely benefits CHAOS_GOLD.
Whether in trending or range-bound markets, London at 18:00 is still London at 18:00. NY close is still NY close at 4:00. The physical act of market participant turnover occurs regardless of market conditions.From the low-volatility years of 2019, through the corona shock of 2020, the sudden shifts of 2022, and the recent gold all-time high in 2025, CHAOS_GOLD maintained positive performance across these completely different markets because the inputs were “time and day,” not price movement.
Three times a day at fixed times. The next entry time is fully predictable by the user.Unplanned mass entries or midnight averaging in EA would not happen with CHAOS_GOLD due to its structural design. This also directly improves ease of monitoring in live or demo, making it safe to run unattended.
Because it’s time-based, you can also incorporate the day of the week. CHAOS_GOLD’s Thursday closure is part of this expansion.By combining the three axes of time, day, and market participants, we build multi-layered entry conditions that cannot be realized by technical-indicator-based systems. Details will be explored in the next installment (Episode 3)..
Anomaly methodology is conceptually simple, but the industry rarity stems from the fact thatyou need seven years of high-quality historical data and thousands of statistical tests to reliably identify useful time windows.
- Securing long-term data: At least five years, ideally seven years of tick-level data. Short-term data cannot distinguish random biases from true anomalies.
- Accumulation of statistical verification: Cross-checks across day-of-week × time × price movement direction. There are thousands of combinations.
- Elimination of spurious correlations: Rules of thumb to determine whether detected biases come from random economic events or structural anomalies.
- Ongoing validation through operation: Verifying in live operation that the discovered time window retains its effectiveness.
CHAOS_GOLD’s 21:00, 18:00, and 4:00 are the results of these validations. Not products of short-term optimization,they are structurally sustainable time windows backed by seven years of gold market data. Therefore, they are expected to continue functioning as long as the market structure does not change significantly.
EAs that dramatize high win rates with technical indicators can be created within a week. However anomaly-type EAs require years of validation time. The reason most commercial EAs are technical-type lies here.
- Average lifespan of commercial EAs is about six months. The key to CHAOS_GOLD surviving seven years is the “anomaly method.”
- Anomaly refers to statistically observed biases within the market. A method that leverages probabilistic edge.
- CHAOS_GOLD’s 21:00, 18:00, and 4:00 are three hours selected at moments when market participants swap.
- Compared with technical-indicator-based approaches, a design with lower over-optimization risk. Therefore, it remains functional in the long term.
- Because inputs are “time and day,” it is less affected by changes in market environment, delivering positive results through the entire period from the corona era to rapid shifts and new highs.
- Limited to three entries per day, but designed so users can completely predict the next entry time with confidence.
- Finding anomalies requires seven years of data and thousands of statistical checks. A rare design philosophy seldom used in commercial EAs.
Next time (Episode 3), we will explain why CHAOS_GOLD closes on Thursdays—how a four-day operation stabilizes the revenue curve.
※This article is provided for information purposes and is not an investment solicitation. The operating results shown are historical performance and do not guarantee future profits. FX and CFD trading carry risks. Please make investment decisions at your own risk.
