ALPHA
[SYS_ID: QUANT_DEV]
// Failed ML Prediction. Pivoted. Shipped Live.

Mechanism Over
Optimism.

Qwark is a systematic quantitative research practice building factor-based equity strategies with live out-of-sample validation and institutional-style risk management.

> Six months of failed ML predictions taught me the only thing that matters: mechanism.
> Now building systematic strategies with economic reasoning, kill conditions, and live validation.
> Not just backtested. Live validated.
Live OOS CAGR +14.5%
Live Since Jul 2026
Researching Since Nov 2025
[PROJECTS_MODULE]
folder_open

Live Projects

[PROJ_01] [STATUS: LIVE]
show_chart

Systematic Equity Strategy

FACTOR-BASED LONG EQUITY // US MID-CAP

Periodically-rebalanced systematic strategy in the US mid-cap space. Single fundamental signal with persistent cross-sectional predictive power. 23-year validated backtest with full factor attribution. Live out-of-sample.

23yr CAGR +14.0%
Full Sharpe 0.535
OOS CAGR +14.5%
Factor-Based US Mid-Cap Python Factor Attributed
[PROJ_02] [STATUS: RESEARCH]
trending_up

Strategy II

QUANTITATIVE RESEARCH // IN DEVELOPMENT

Second systematic strategy currently in research phase. Same discipline — mechanism-validated signal, pre-specified kill conditions, full factor attribution. Details under embargo until validation complete.

Domain
Signal
Phase RESEARCH
Quantitative In Research Mechanism-Validated
[RESEARCH_PIPELINE]
account_tree

Research Process

> Every strategy follows the same pipeline. No shortcuts, no skipping steps. The process is the risk management.

[STEP_01] lightbulb

Mechanism

State the economic reason this signal should work. If you can't, stop.

[STEP_02] science

Backtest

Validate signal over full sample. Document every assumption. No data snooping.

[STEP_03] pie_chart

Attribution

Multi-factor attribution. Know what drives your returns.

[STEP_04] rule

Kill Conditions

Define yellow/red flags before deployment. Breach = stop, then investigate.

[STEP_05] rocket_launch

Deploy + Monitor

Go live. Quarterly OOS review. Document underperformance — never hide it.

[RESEARCH_LOG]
biotech

Research Philosophy

> These aren't abstract principles. Each one was learned the hard way — by building something that worked on paper and failed in practice. The philosophy IS the risk management.

[PRINCIPLE_01]
science

Mechanism Before Alpha

Every signal must have a stated economic mechanism — a reason it should work that's grounded in how markets actually operate, not just in what historically correlated. Cross-sectional patterns without return-side implications are measurement artifacts. I've tested this the hard way.

[PRINCIPLE_02]
lock

Lock-Before-Search

No factor research until a specific, identified deficiency exists. "Increasing CAGR" is not a valid problem when the strategy already works. The correct question: what risk does this new factor reduce? If you can't answer that, the search is p-hacking with extra steps.

[PRINCIPLE_03]
rule

Pre-Specified Kill Conditions

Every strategy ships with yellow and red flags defined before deployment. Not after. If it breaches, you stop — then investigate. Not the other way around. This is the single hardest discipline in systematic trading and the one that saves you from yourself.

[OPERATOR_LOG]
person

Operator Profile

[IDENTITY_MODULE]

I build systematic strategies that have a mechanism — not just a backtest.

Quantitative researcher and developer. I started in November 2025 building ML price prediction models — they all failed. In May 2026, I stepped back and realized the problem: prediction without mechanism is just curve-fitting. I pivoted to systematic factor research, and within two months shipped a live strategy with a mechanism-validated signal, pre-specified kill conditions, and full factor attribution.

Every strategy I ship has a stated economic mechanism, pre-specified kill conditions, and honest factor attribution. I don't overclaim, I don't post-hoc rationalize drawdowns, and I don't add factors to a strategy that already works. The research process is the edge — not the signal.

[EDGE] psychology

Research Discipline

Mechanism-first. Lock-before-search. Kill conditions defined before deployment. No post-hoc rationalization.

[DOMAIN] domain

Expanding Research

Systematic equity factors with live OOS track record. Second strategy in research phase.

[PIVOT] speed

Pivot Over Ego

Six months of failed ML predictions. Pivoted instead of doubling down. Two months to live strategy after course correction. Kill your darlings, ship what works.

[TECH_STACK]

Technical Profile

Research
Python Factor Models Backtesting Engines NumPy / SciPy
Execution
REST APIs Order Execution Risk Systems Async Architecture
Data
Market Data API Integration Structured Logs JSONL Audit
[STANDING]

Standing

High school student. Self-taught quant. No finance degree, no prop desk, no mentor handoff. Started November 2025 with ML price prediction — failed. Pivoted May 2026 to mechanism-first factor research. Shipped live strategy within two months of the pivot.

I compete on output — not credentials. The track record either validates or it doesn't. No prestige shield, no network gate.

Active // Live Since Jul 2026
[TRAJECTORY]

Trajectory

Jun–Jul 2026

Second Strategy + Live Deployment

Strategy I goes live July 1. Research begins on Strategy II — same discipline, different mechanism.

Jun 2026

Strategy I Research Complete

Backtesting, factor attribution, kill conditions all validated. Signal proves persistent across multiple holding horizons.

May 2026

The Pivot

Abandoned ML price prediction. Realized directional forecasting is the wrong problem. Stepped back, rethought from first principles — mechanism, not prediction.

Nov 2025 – Apr 2026

ML Price Prediction (Failed)

Built machine learning models for directional price prediction. Every iteration failed. The lesson: prediction without mechanism is curve-fitting.

Pre-Nov 2025

Self-Directed Learning

Programming, probability, market microstructure. Consumed everything from scratch. No shortcuts — just the work before the work.

[COMMS_MODULE]
signal_cellular_alt

Signal

> Open to conversations about systematic strategies, factor research methodology, and quantitative infrastructure. Not looking for pitch decks — looking for people who think in mechanisms.