JA//SYSTEMS eagle mark JA//SYSTEMS
SYSTEM HEARTBEAT UTC --:--:--
X12 XAUUSD
LIVE FUNDED ACCT
V8MR EURCHF
DEMO FWD VALIDATION
ORB US500
VALIDATED PRE-DEPLOY

Judah Ahiaba · Systematic Trader & Quant Researcher

Three systems.
Zero correlation.
Deflated numbers only.

I build and run automated trading systems the way a desk would audit them: look-ahead-free backtests, selection-bias deflation, published negative results, and kill-switches that exist in code, not in a document. One system trades live capital. The other two wait until their code matches their research. That discipline is the product.

Inspect the systems Read the research

3,900+

Configurations tested

12 survived the acceptance gates. The rest were published as failures, not buried.

≈ −0.04

Cross-system correlation

Gold trend, FX mean reversion, index momentum: three mechanisms, three instruments, three sessions.

0.7 – 1.2

Planning Sharpe · deflated

The number after selection-bias deflation, not the 1.55–1.79 the raw backtests would let me claim.

Method

Why every number here is smaller than it could be

Every headline expectancy on this site is the best survivor of a large search, and I say so. A best-of-1,188 result is not the same evidence as a single pre-registered test, so each system carries two numbers: the validated backtest figure, and the deflated planning band I actually size risk from.

When a bias was found in my own backtest, I quantified it with a control run and republished: 70% of the claimed edge evaporated. The full story (and two more like it) is on the research page. If a trading site only shows you its wins, you're reading marketing. This page is a trade journal.

The book

Three uncorrelated sleeves

XAUUSD · Gold

X12 · Gold Trend Engine

Long-volatility trend following

LIVE

Multi-timeframe trend alignment gates a volatility-adjusted breakout entry. There is no profit target: losers are cut at a fixed, server-side stop and winners are ridden with an automated trailing exit. The system's own exit study showed the edge lives in the exit, not the entry. Capping winners destroys it.

Expectancy / trade

+0.363R

Profit factor

1.66

Win rate

45.6%

Trades

263 / 3y

Max drawdown

10.9R

Years positive

4 / 4

Deflated planning number · +0.20 to +0.30R

Headline is the best of a 20-variant exit study; after selection-bias deflation the honest planning band is +0.20–0.30R, with profits concentrated in the 2025–26 gold regime. Sized and monitored accordingly.

Equity curve

+95.6R

Cumulative R, 263 trades: corrected, look-ahead-free backtest (2023–2026, real spreads)

EURCHF · FX

V8MR · EURCHF Mean Reversion

Short-volatility statistical reversion

DEMO

Fades statistically extreme dislocations during the hours when European order flow is absent and the pair oscillates around its mean. Fixed-risk stops, defined targets, and a time-based abort. Screened from 207 instruments; the edge survives on exactly one pair, and is published as such.

Expectancy / trade

+0.232R

Profit factor

1.45

Trades

254 / 4y

Out-of-sample 2021

+0.226R

Max drawdown

9.1R

Bootstrap 95% CI

+0.08–0.38R

Deflated planning number · +0.10 to +0.15R

Best of a 288-config grid: deflated planning band is +0.10–0.15R in a normal volatility regime, and ≈ +0.03R in the current compressed regime. That is why it is on demo, not live.

Equity curve

+67.5R

Cumulative R, 318 trades: validation trade log incl. held-out 2021 (real spreads)

US500 · Index CFD

ORB · US500 Open-Range Continuation

Intraday momentum, no overnight risk

VALIDATED

Trades the highest-participation event of the trading day: the New York equity open. A clean break of the opening auction's range, confirmed rather than chased, is ridden intraday and everything is flat by the close. 1,188 configurations tested; 9 of 11 instruments failed. The survivors form one tight, coherent cluster.

Expectancy / trade

+0.278R

Profit factor

1.46

Trades

548 / 4y

t-statistic

3.2

Max drawdown

14.2R

Years positive

5 / 5

Deflated planning number · +0.15 to +0.20R

Best of 1,188 combinations: the deflated planning band is +0.15–0.20R. Only 0.4% of tested configurations passed, and they share one recipe: that clustering, not any single t-stat, is the evidence the edge is real.

Equity curve

+152.3R

Cumulative R by year, 548-trade validation run 2022–2026 (annual resolution; 2026 partial)

Portfolio framing

Average-day correlation ≈ 0 is the easy claim; the honest one is that correlations converge in the left tail, so the combined book is planned at 10 – 15% peak-to-trough drawdown and 0.7 – 1.2 Sharpe at 10–15% vol. Stress scenarios are written down, not discovered live.

Portfolio detail

Research, including the failures

The three studies that define how I work

01 · BIAS HUNT

The backtest that read tomorrow's newspaper

How a subtle look-ahead bias inflated a portfolio edge by ~70× its honest size: found, quantified with a control run, and published.

Read the case study →

02 · NEGATIVE RESULT

123 pairs, 116,199 setups, no edge

A manual-trading belief I'd used for years, tested properly across the full FX universe. Verdict: a coin flip. Strategy retired.

Read the case study →

03 · DISCOVERY

The edge was only ever on one side

Re-cutting a validated mean-reversion system by direction revealed all of the edge lives in one side of the book, and why that changes the tail risk.

Read the case study →

Need a system built, or a backtest you can actually trust?

I build MT5 expert advisors with validation pipelines, audit existing systems for the biases that quietly kill accounts, and mentor traders making the shift to systematic.

Services & pricing Get in touch