Validation record that annihilated the standalone indicator 91 logic with EURUSD — The "mathematical ceiling of a single EA" seen in 47,000 configurations

“Does RSI actually work?” “Is the MACD Golden Cross still usable?” ── An indicator that has long been treated as a given, how far does it actually perform in real markets? I thoroughly tested this question using EURUSD1 as the instrument.

7 timeframes × 91 indicator logics × parameters × exit designs × stop widths, cumulativeover 47,000 configurationsin backtests.

In short,the only single-indicator EA that was statistically positive was RSI(28); all other 90 logics were negative.Even RSI(28) fell short of the target performance (monthly +5% / DD 5%) in EURUSD alone by 6.7x. “Stable side income from a standalone EA” is mathematically impossible in EURUSD. This article records the full details.

In this article,

① distribution of 47,000 configurations (what percentage were positive), ② the fact that out of the 91 standalone indicator logics, only one was positive, ③ the reason RSI(28) was an exception, ④ why the MTF environment recognition gate 3,552 configurations couldn’t save them, ⑤ why achieving target with a standalone EA is mathematically impossiblewill be written with actual numbers only.

---

Distribution of 47,000 configurations ─ what percentage were positive

First, look at the overall distribution of the 47,000 configurations that were tested.

In EURUSD1, the combinations involve seven timeframes from M1 to D1 × 91 logics × 8 exit types × 4 stop types × other factors.

Performance distribution for the full period (all 47,000 configurations)

• Monthly return +5% or more: about 19 configurations (0.04% of the total)
• Monthly return +3% to +5%: about 108 configurations (0.23%)
• Monthly return +1% to +3%: about 1,401 configurations (2.98%)
• Monthly return 0% to +1%: about 6,590 configurations (14.02%)
• Monthly return -1% to 0%: about 13,028 configurations (27.72%)
• Monthly return -1% or less: about 25,864 configurations (55.03%)

Ratio of positives to negatives

• Positive configurations: overallabout 17%(8,118 configurations)
• Negative configurations: overallabout 83%(38,892 configurations)

There are actually 17% configurations that look positive in backtests.These numbers up to this point are at a level where many people are happy thinking “this can win” after backtesting.

But from here, walk-forward testing (reproducing in an unseen period) collapses over 99% of the positive configurations.

Walk-Forward test (unknown period) kept positives

When selected in the training period and tested in the unknown period, the configurations that persisted areabout 9 configurations only(0.02% of the total).

Of the 17% positive configurations seen in backtests, less than 1% remained usable in real trading.

“Winning in backtests” for 99% is an illusion— this is the numerical conclusion from 47,000 configurations.

---

Record of the 91 standalone indicator logics

Among the 47,000 configurations, narrowing to the “standalone indicator (EA using a single indicator)” yielded even more striking results.

Out-of-Sample mean expectation for the 91 standalone indicator logics

For each major indicator family, the average expectation in Out-of-Sample (unknown period) is listed. The numbers are averages over multiple parameters.

• RSI family (average across all parameters): Expectation -0.023R
• MACD family: Expectation -0.041R
• Stochastic family: Expectation -0.038R
• ADX family: Expectation -0.029R
• Bollinger Bands family: Expectation -0.045R
• Donchian family: Expectation -0.058R
• MA Cross family: Expectation -0.042R
• Keltner family: Expectation -0.040R
• Ichimoku family: Expectation -0.056R
• PSAR family: Expectation -0.033R

All indicator families are negative. Not a single family had a positive average.

The sole exception within the 91 logics: RSI(28)

The overall average is negative, but when looking at parameters individually, there was exactly one exception.

• RSI(28) Daily chart, contrarian: Out-of-Sample expectation+0.101R

Among the 91 logics across all parameters, only this one produced a statistically significant positive result.The sole remaining standalone indicator edge with a 0.02% probability.

Why RSI(28) Daily only remained (discussion)

Reason RSI(28) performed exceptionally:

• The RSI’s textbook default value (14) is opposite and negative (-0.069R); ironically, the “settings everyone uses” are the worst
• RSI(28) captures longer-term momentum reversals. On EURUSD daily, it reverses after absorbing short-term RSI false signals
• The daily timeframe itself minimizes the impact of costs (spread + slippage)

There is an exceptional edge that goes against textbook conventional wisdom, and that was an important takeaway from standalone indicator testing.

---

From here on,the reasons for the best/worst values by system, why the MTF environment recognition gate 3,552 configurations couldn’t save them, the meaning of mo/DD ceiling 0.149 for a standalone EA, the 6.7x distance to target 1.0, and the numerical basis for the conclusion that diversification is the remaining pathwill be written.

---