Algorithmic Candlestick
Intelligence Platform

Background & Professional Experience

MarubozuScan is designed, built, and operated by a self-directed algorithmic systems developer with over a decade of hands-on experience in quantitative financial data analysis, full-stack web engineering, and automated market surveillance systems. The project emerged from a long-standing academic interest in Japanese candlestick charting methodology — specifically the mathematical geometry of the Marubozu formation — and the absence of a dedicated, open-access, real-time scanning tool for this pattern class.

My professional background spans financial technology (FinTech) backend development, statistical signal processing for time-series data, and the design of rule-based algorithmic engines that operate without discretionary human input. Prior to launching this platform, I spent several years working on proprietary market data aggregation pipelines and price-action alert systems for personal research purposes, building a rigorous understanding of OHLC (Open, High, Low, Close) data structures, exchange APIs, and the mathematical definitions that distinguish genuine pattern formations from noise.

MarubozuScan is the public-facing, educational expression of that research — a commitment to publishing transparent, fully documented, real-time candlestick scan results so that other market students and technical analysis enthusiasts can observe, study, and learn from pattern data in a live market context.

Algorithmic Goals & Research Mission

The primary mission of this platform is the systematic, real-time identification of Marubozu candlestick formations — candles characterized by minimal or absent upper and lower shadows, indicating decisive, one-sided price action across a given timeframe. The algorithmic goals of this project are strictly educational and analytical in nature:

• OBJ-01 Mathematical Pattern Fidelity. To apply precise, configurable mathematical thresholds to OHLC candlestick data that distinguish valid Marubozu formations (Full, Bullish, and Bearish variants) from partial or ambiguous patterns, using shadow-to-body ratio calculations as the primary discriminator.
• OBJ-02 Multi-Asset, Multi-Timeframe Coverage. To continuously scan a broad universe of instruments — including major Forex pairs, cryptocurrency spot markets, and global equity indices — across multiple candlestick timeframes (M15, H1, H4, D1) simultaneously, providing a comprehensive and timely snapshot of pattern emergence.
• OBJ-03 Data Transparency & Auditability. To present all detected pattern data with full attribution: the instrument, exchange, timeframe, OHLC values, detection timestamp (UTC), and the specific ratio values that triggered the classification, allowing any visitor to independently verify each result against their own charting platform.
• OBJ-04 Zero Predictive or Advisory Claims. To present all scan outputs strictly as historical and real-time descriptive data — never as forward-looking trade signals, investment advice, or performance forecasts. The platform documents what has occurred mathematically; it makes no claim about what will occur next.
• OBJ-05 Open Educational Resource. To serve as a free, accessible learning tool for students of technical analysis who wish to study the real-world frequency, distribution, and structural characteristics of Marubozu patterns across live market conditions — a resource the developer wished had existed during their own learning phase.

24/7 Self-Hosted Workstation Architecture

The platform operates on a purpose-built, self-hosted workstation located at a dedicated residential data node maintained by the developer. Unlike cloud-only deployments that introduce latency variance and cold-start delays, this infrastructure prioritizes low-latency, high-availability data ingestion through a persistent hardware setup running continuously without interruption.

Market data is ingested via authenticated WebSocket connections to multiple broker and exchange APIs, providing redundant OHLC tick data across covered instruments. Each incoming candle is immediately parsed, validated against data-quality checks (gap detection, timestamp continuity, volume sanity), and passed to the pattern detection engine. Detected patterns are written to a time-series database with full metadata and served to the front-end dashboard via a lightweight REST API with server-sent events (SSE) for live push updates.

Scheduled cron jobs perform daily data integrity reconciliation, comparing stored candle records against reference data from secondary sources to detect and flag any discrepancies. All system events, errors, and pattern detection logs are retained for a rolling 90-day window, available for review upon reasonable request.

Pure Candlestick Mathematical Tracking

The detection algorithm applies a strict, parameter-driven mathematical definition to each closed candlestick. This approach deliberately avoids interpretive or heuristic shorthand — every classification is the direct output of arithmetic operations on the four OHLC price values.

The core detection formula operates as follows: the candle body is defined as the absolute difference between the open and close prices. The upper shadow is the distance from the higher of open/close to the candle high. The lower shadow is the distance from the candle low to the lower of open/close. A full Marubozu is declared when both the upper shadow ratio (upper shadow ÷ body) and the lower shadow ratio (lower shadow ÷ body) fall below a configurable maximum threshold — defaulting to 1.5% for each — and the body itself represents at least a minimum percentage of the total candle range.

Bullish and Bearish half-Marubozu variants are detected independently by testing only the relevant shadow (the shadow on the side of the opening price for a Bullish, or the closing price for a Bearish). All threshold parameters are version-controlled and documented in the project's public-facing methodology page, ensuring that any observer can replicate results given the same raw OHLC data.

Our Commitment to Accuracy and Transparency

MarubozuScan is maintained as a personal research and educational publishing project. There is no investor, no sponsor, and no commercial entity directing the content or methodology of this platform. Every design decision — from the pattern detection thresholds to the data display format — is made solely by the developer with the goal of maximum transparency and educational utility.

We are committed to maintaining and publicly documenting the following standards: clearly labeled detection timestamps and data sources for every scan result; a public changelog for any modifications to detection parameters; explicit, repeated disclosure on all pages that this platform provides educational data only and does not constitute financial advice; and prompt correction and public acknowledgment of any errors in data processing or pattern classification.

This platform does not accept payment from brokers, signal services, or any third party for placement, promotion, or favorable presentation of any instrument, strategy, or service. Any third-party advertising displayed on this site (through Google AdSense) is served automatically by Google's algorithms and does not represent endorsement of any advertised product or service by the site operator.