FOUNDER & CIO
において
Founder & CIO of Flow Dynamics Capital. Quantitative researcher and
full-stack developer with 18+ years of systematic R&D in financial
markets ; manuscript-based mathematical research validated manually
for years before any algorithmic implementation.
I build production-grade systematic trading infrastructure. Not
indicators stitched together. Not curve-fitted EAs. Real edge,
proven through institutional-grade methodology.
— — —
WHAT I DO
▸ Expert Advisors and indicators in MQL5 (MT4/MT5) — 10.000 line
institutional codebases, fully audited, zero warnings, zero errors.
▸ Python pipelines for backtest analysis, Monte Carlo simulation,
Deflated Sharpe Ratio, PBO, walk-forward evaluation, Combinatorial
Purged Cross-Validation.
▸ C++ for performance-critical modules.
▸ Genetic optimization with custom OnTester() criteria — engineered
to penalize overfitting, not reward it. If your strategy passes
my walk-forward, it has actual structural edge.
▸ Risk management modules: dynamic position sizing, exposure caps,
drawdown-aware regime filters.
▸ Full audit reports: data integrity, OOS/IS partitioning, capacity
analysis, transaction cost stress-testing, sensitivity plateaus.
— — —
WHY MOST EAs FAIL — AND WHY MINE DON'T
Most freelance MQL5 work is overfitted to the IS window and dies
in production. I treat backtesting as falsification, not
confirmation. Every strategy I build or audit goes through:
• Out-of-sample partitioning (true OOS, not pseudo-OOS)
• Walk-forward validation with retention metrics
• Deflated Sharpe Ratio + Probability of Backtest Overfitting
• Slippage and commission stress-testing across realistic
execution latency bands
• Parameter sensitivity analysis — if your edge collapses at
±10% perturbation, it was never edge
This is the methodology behind every line of code I deliver.
— — —
BACKGROUND
I am a private quantitative researcher. I run my own systematic
strategies and I have built proprietary algorithms over the years
that remain under NDA — they are not for sale and never will be.
I am on MQL5 Freelance not to liquidate research, but because I
genuinely enjoy the craft of writing institutional-grade MQL5
code for clients who want it done right the first time.
I spent 20+ years in financial markets, including extensive
discretionary trading before transitioning fully to systematic
research. That dual perspective matters: I understand what your
strategy is trying to capture before I write a single line.
— — —
HOW I WORK
▸ Specifications first. I will not start coding before we agree on
exactly what success looks like.
▸ Production-grade code. Documented, modular, maintainable. No
spaghetti, no copy-paste from forums.
▸ Honest delivery. If your idea has a structural flaw, I will tell
you before you spend money on it. If it has merit, I will build
it cleanly.
▸ Fast turnaround. I do not juggle 15 jobs at once. When I take
your project, it gets full focus.
— — —
Languages: English (fluent), French (native).
Stack: MQL5, MQL4, Python (NumPy, pandas, matplotlib, ReportLab),
C++, MT5 Strategy Tester (custom OnTester / CustomMax), genetic
optimization, walk-forward frameworks.
Alpha through flow intelligence.
full-stack developer with 18+ years of systematic R&D in financial
markets ; manuscript-based mathematical research validated manually
for years before any algorithmic implementation.
I build production-grade systematic trading infrastructure. Not
indicators stitched together. Not curve-fitted EAs. Real edge,
proven through institutional-grade methodology.
— — —
WHAT I DO
▸ Expert Advisors and indicators in MQL5 (MT4/MT5) — 10.000 line
institutional codebases, fully audited, zero warnings, zero errors.
▸ Python pipelines for backtest analysis, Monte Carlo simulation,
Deflated Sharpe Ratio, PBO, walk-forward evaluation, Combinatorial
Purged Cross-Validation.
▸ C++ for performance-critical modules.
▸ Genetic optimization with custom OnTester() criteria — engineered
to penalize overfitting, not reward it. If your strategy passes
my walk-forward, it has actual structural edge.
▸ Risk management modules: dynamic position sizing, exposure caps,
drawdown-aware regime filters.
▸ Full audit reports: data integrity, OOS/IS partitioning, capacity
analysis, transaction cost stress-testing, sensitivity plateaus.
— — —
WHY MOST EAs FAIL — AND WHY MINE DON'T
Most freelance MQL5 work is overfitted to the IS window and dies
in production. I treat backtesting as falsification, not
confirmation. Every strategy I build or audit goes through:
• Out-of-sample partitioning (true OOS, not pseudo-OOS)
• Walk-forward validation with retention metrics
• Deflated Sharpe Ratio + Probability of Backtest Overfitting
• Slippage and commission stress-testing across realistic
execution latency bands
• Parameter sensitivity analysis — if your edge collapses at
±10% perturbation, it was never edge
This is the methodology behind every line of code I deliver.
— — —
BACKGROUND
I am a private quantitative researcher. I run my own systematic
strategies and I have built proprietary algorithms over the years
that remain under NDA — they are not for sale and never will be.
I am on MQL5 Freelance not to liquidate research, but because I
genuinely enjoy the craft of writing institutional-grade MQL5
code for clients who want it done right the first time.
I spent 20+ years in financial markets, including extensive
discretionary trading before transitioning fully to systematic
research. That dual perspective matters: I understand what your
strategy is trying to capture before I write a single line.
— — —
HOW I WORK
▸ Specifications first. I will not start coding before we agree on
exactly what success looks like.
▸ Production-grade code. Documented, modular, maintainable. No
spaghetti, no copy-paste from forums.
▸ Honest delivery. If your idea has a structural flaw, I will tell
you before you spend money on it. If it has merit, I will build
it cleanly.
▸ Fast turnaround. I do not juggle 15 jobs at once. When I take
your project, it gets full focus.
— — —
Languages: English (fluent), French (native).
Stack: MQL5, MQL4, Python (NumPy, pandas, matplotlib, ReportLab),
C++, MT5 Strategy Tester (custom OnTester / CustomMax), genetic
optimization, walk-forward frameworks.
Alpha through flow intelligence.
