【Rising Golden Dragon】Backtesting collapsed, in live trading it was fine — the tale of the "economic indicator" that made the difference
Backtesting Falls Apart, Real Trading Goes Smooth
Typically, the results of backtests (historical verification) are better than those in real trading. Hence, the common warning is that “backtests can look good, but forward (real trading) may collapse.” However, with my Gold EA “Shokinryu,” the opposite phenomenoncan occur.Backtests may fail while real trading remains fine. Why do the results invert with the same EA and settings? Today, I’ll break down the reasons.
The biggest reason — Backtests cannot obtain “economic indicators”
Shokinryu includesa built-in feature to automatically stop new entries before and after economic indicators. At the moment of an indicator release, price gaps and spreads widen dramatically. This mechanism helps avoid that.
The problem is that,backtests cannot fetch this indicator calendar due to MT5 limitations. In other words, during past verifications, entries could be made in “volatile times” when they should have been avoided because the filter wasn’t active.
Shokinryu is a strategy that adds to long positions. If it blindly dives into sudden changes during indicator events, unrealized losses can deepen rapidly.Backtests fail but real trading does not mainly for this reason. In real trading, the indicator filter helps avoid these scenes, so the results can differ even in the same market.
※Illustrative image. Backtests tend to show tougher results since the indicator filter doesn’t operate.
Another factor — Spread and modeling differences
There is another factor widening the gap.Spread and history accuracyGold often widens spread around indicator releases and early morning. Depending on backtest settings, it may fail to reproduce the spread expansion accurately or the price movement within a single candlestick.
Whether you run with all-tick data and actual spreads or a rough 1-minute OHLC, the way the grid’s drawdown appears changes. Incomplete backtests can be either harsher or more lenient than real trading.
This is not a “soft” backtest but an “exceedingly strict” case
This is the key point. The usual caution is to beware of “backtests being too optimistic and failing in real trading.” But with Shokinryu, the opposite occurs:backtests run under the worst condition without an indicator filterand therefore show harsher numbers than real trading.
Thus, the backtest failure should be read as the worst case when you cannot avoid indicators, and in real trading, avoiding that moment changes the outcome.
However, it does not mean that real trading never fails
What I want to avoid is thinking this means“real trading is safe”. The indicator filter avoids releases but cannot prevent every abrupt move. Gaps at the next Monday or unanticipated one-direction moves can still lead to significant unrealized losses in real trading.
Shokinryu is designed to accumulate purchases. If you run with high leverage or oversized lots, there is a real risk of stop-out even in real trading.Freed-up capital and proper lot management are prerequisites— and that does not change.
Summary — Read the “why” behind the numbers
- The main reason Shokinryu fails in backtests isthat economic indicator filters do not operate in historical verifications.
- In real trading, the indicator release window halts new entries, so the same market can yield different results
- Spread and modeling differences (all-tick vs 1-minute) also matter
- — This is a case of backtests being overly strict. The failure number should be read as“the worst value when indicators could not be avoided”.
- Howeverreal trading is not automatically safe. Gaps, abrupt moves, and high leverage still carry risk. Sufficient capital is a prerequisite
When backtests and forward results diverge, what matters more than the numbers themselves is“why the divergence occurred”. If you understand the reason, you’ll know which numbers to trust. I also detail the difference between backtests and real trading inthis article.
※This article is for informational purposes only and is not an investment solicitation. The performance data shown reflects past results and does not guarantee future profits. FX/CFD trading involves risk. Please make investment decisions at your own risk.