Intelligent Cryptocurrency Trading System Using Machine Learning: A RandomForest-Based Adaptive Signal Generation Approach with Real-Time Risk Management

Abstract — Cryptocurrency markets present one of the most volatile and unpredictable environments for automated financial decision-making. Operating continuously across global time zones with no centralized authority, these markets are susceptible to sudden regime shifts driven by regulatory announcements, macroeconomic signals, and social media sentiment. This paper presents CryptoBazaar, a web-based intelligent cryptocurrency trading system that integrates a supervised machine learning pipeline with a full-stack deployment framework. The core inference engine employs a RandomForestClassifier trained on five thousand synthetic OHLCV records with thirteen engineered technical indicators including RSI, MACD, Bollinger Bands, SMA crossovers, and momentum returns. The model generates three-class BUY, SELL, or HOLD signals with 51.6% test accuracy, representing a statistically meaningful 55% improvement over the 33.3% random baseline for three-class classification problems. The HOLD class achieves 69% precision. Beyond signal generation, the system provides a complete trading infrastructure encompassing user authentication, portfolio tracking, real-time risk management with stop-loss automation, historical signal persistence in SQLite, and performance reporting. Backtesting over 365 simulated trading days on BTC/INR data demonstrates a 43.2% portfolio return from an initial capital of Rs.10,00,000 with a Sharpe ratio of 1.14, a win rate of 67.4%, and maximum drawdown constrained to 14.3%. The system directly resolves five critical limitations of manual and fixed-rule automated trading: slow reaction time, emotional bias in decision-making, poor entry and exit timing, static non-adaptive strategy logic, and absence of real-time protective mechanisms.

Authors

Institutions

Publication Details

Journal
HAL
Published
2026-11-16
DOI
https://doi.org/10.5281/zenodo.24271261
Primary Topic
Stock Market Forecasting Methods
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Intelligent Cryptocurrency Trading System Using Machine Learning: A RandomForest-Based Adaptive Signal Generation Approach with Real-Time Risk Management

Koganti Gayathri, SAI KARUN NANDIPATI, Mohammad Aman Shafiqua
2 links
HAL
Stock Market Forecasting Methods
preprint

Intelligent Cryptocurrency Trading System Using Machine Learning: A RandomForest-Based Adaptive Signal Generation Approach with Real-Time Risk Management

Koganti Gayathri, SAI KARUN NANDIPATI, Mohammad Aman Shafiqua
preprint en

Abstract

Abstract — Cryptocurrency markets present one of the most volatile and unpredictable environments for automated financial decision-making. Operating continuously across global time zones with no centralized authority, these markets are susceptible to sudden regime shifts driven by regulatory announcements, macroeconomic signals, and social media sentiment. This paper presents CryptoBazaar, a web-based intelligent cryptocurrency trading system that integrates a supervised machine learning pipeline with a full-stack deployment framework. The core inference engine employs a RandomForestClassifier trained on five thousand synthetic OHLCV records with thirteen engineered technical indicators including RSI, MACD, Bollinger Bands, SMA crossovers, and momentum returns. The model generates three-class BUY, SELL, or HOLD signals with 51.6% test accuracy, representing a statistically meaningful 55% improvement over the 33.3% random baseline for three-class classification problems. The HOLD class achieves 69% precision. Beyond signal generation, the system provides a complete trading infrastructure encompassing user authentication, portfolio tracking, real-time risk management with stop-loss automation, historical signal persistence in SQLite, and performance reporting. Backtesting over 365 simulated trading days on BTC/INR data demonstrates a 43.2% portfolio return from an initial capital of Rs.10,00,000 with a Sharpe ratio of 1.14, a win rate of 67.4%, and maximum drawdown constrained to 14.3%. The system directly resolves five critical limitations of manual and fixed-rule automated trading: slow reaction time, emotional bias in decision-making, poor entry and exit timing, static non-adaptive strategy logic, and absence of real-time protective mechanisms.

HAL
Siddhartha Medical College (IN), Siddhartha University (NP), Siddhartha Academy of Higher Education (IN)
Stock Market Forecasting Methods
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon