// CHETRA — PHNOM PENH · SYSTEMATIC DAY TRADER
Discipline,
engineered.
I build the machines that keep me disciplined. An autonomous strategy factory — thousands of trials, anti-overfit gates, locked holdout data — and a risk ladder that never negotiates.
Performance metrics
SHARPE — REAL SPREAD
1.53
MT5 validation, 512 trades
HOLDOUT SHARPE
3.75
locked data, 2026 H1
DEFLATED SHARPE
0.91
gate ≥ 0.90 — passed
MAX DRAWDOWN
11.1%
real-spread backtest
TRIALS → SURVIVORS
3,911 → 1
the factory rejects 99.97%
Why discipline
// WHY DISCIPLINE
Most traders lose to themselves before they lose to the market.
I treat that as an engineering problem. My risk ladder is hard-coded: 1% risk drops to 0.5% after a loss. Two losses and the day is over. A kill-switch is always armed. The rules live in code because code doesn't renegotiate at 2 a.m.
Self-taught in Phnom Penh. The market doesn't care where you sit — only whether your process holds.
// CASE STUDY — THE STRATEGY FACTORY
A factory that fires strategies.
An autonomous pipeline that researches, tests, and rejects trading strategies without me in the loop. MT5 runs in Docker behind a Wine bridge; a vectorbt engine backtests against real spread costs; every candidate faces anti-overfit gates before it earns a forward test. 19 strategy modules. 8 symbols. One survivor so far — and that's the point.
- Research
idea → parameter space
- Screen
vectorbt, 3,911 trials
- Validate
MT5 real spreads
- Audit
DSR gate ≥ 0.90 + locked holdout
- Compare
3 profiles: firm / real / crypto
- Live
paper first. always.
THE FUNNEL
3,911 parameter trials
2,192 combos ranked
20 audited for overfit
1 passed every gate
THE SURVIVOR — ORB_SESSION · XAUUSD H1
- Screen Sharpe
- 2.50
- Real-spread Sharpe
- 1.53
- Retained after costs
- 61%
- Holdout Sharpe (locked)
- 3.75
- Win rate
- 52.5%
- Max drawdown
- 11.1%
- Trades
- 512
Screen numbers flatter. The factory publishes what survives real spreads — 2.50 became 1.53, and 1.53 is the number that matters.
Nineteen of twenty audited candidates failed the deflated-Sharpe gate. The one that passed was retested on five months of locked holdout data it had never seen.
STATUS: preparing HolaPrime 2-step challenge under the validated strategy and risk ladder. Live results will replace this line when they exist — not before.
THE COCKPIT
A real-time dashboard watches every stage — equity, open risk, pipeline state — backed by MongoDB Atlas.
- OPEN RISK
- 0.50%
- DAY P&L
- +0.82%
- SHARPE
- 1.53
- WIN RATE
- 52.5%
14:32:11 audit DSR 0.909 ≥ 0.90 ✓ pass
Projects
// PROJECTS
Strategy Factory
An autonomous pipeline that researches, screens, and rejects trading strategies before any of them see real money.
Python · Docker · vectorbt · pandas · MongoDB
Trading Cockpit
A real-time command view of the Strategy Factory — equity, open risk, and pipeline state in one dashboard.
TypeScript · FastAPI · MongoDB · ECharts
HolaPrime EA
A prop-firm Expert Advisor enforcing the risk ladder for a HolaPrime 2-step challenge account.
Python · MQL5
LLM-Trading
A leak-checked feature and bracket-trading research pipeline for XAUUSD, built around walk-forward validation instead of a single train/test split.
Python · Jupyter · Gymnasium · PPO · Next.js · TypeScript
Stack
Python vectorbt pandas FastAPI JavaScript ECharts D3 Docker MongoDB MQL5 MCP / multi-agent
// CONTACT
The numbers are validated.
The discipline is built.
If you fund traders — or want trading systems built with the same rigor — get in touch.