Eugen Soloviov

Open source, customized, extendable trading platform that supports 100+ crypto exchanges.

The Telegram Client for Deep Focus

A CLI tool that generates concise and descriptive git commit messages using LLM

A curated list of awesome CCXT assets, resources, and more.

No description.

Automate Google AI — Gemini, Flow (Nano Banana, Veo 3, Omni), NotebookLM — in Python: generate text, images, video, music, slides & podcasts with no API keys, just your Firefox session. Stack: Camoufox + Playwright.

No description.

No description.

No description.

Desktop app for parallel Claude Code sessions with project management, drag-and-drop prompts, task boards, and remote access

No description.

Polyglot neural models that read the market across modalities — candlestick tokens, technical-indicator tokens, and L2 order-book microstructure. Research monorepo (PyTorch).

Claude Code skill set for preparing algotrading investor documents (models, proof, build, localize, skeptical-investor audit)

Chapter 6: Information Theory and Kelly Criterion for Trading

Chapter 7: Game Theory and Auctions for Trading

Chapter 1: Stochastic Calculus for Algorithmic Trading

Open Agent Manager — workspace monorepo with app and website submodules

No description.

No description.

How much Sharpe does a look-ahead leak manufacture? A controlled, reproducible study of three surgical leaks (same-bar fill, whole-series normalization, one-bar indicator peek) against known ground truth, with a lagged-feature control isolating the channel and a ~1/sqrt(L) magnitude law. Code + paper.

Does IPC kill the cross-language backtest engine? A controlled measurement of the process-boundary tax: in-process numba vs a real Rust engine over pipe/socket (equivalence-gated). Raw batched boundary ~0.1%; the tax is JSON (~1348x), chatty, spawn, granularity. Plus a measured ZigBolt shared-memory floor (~39ns vs 14us socket). Code + paper.

IPC latency benchmark: TCP, UDS, ZeroMQ, WebSocket, Redis, Shared Memory, Named Pipe — round-trip latency & throughput for algo trading

Local song generation with ACE-Step 1.5 via ComfyUI — markdown draft in, mp3 out. Built for scripts and AI agents. Text→audio, audio2audio remix, timbre transfer.

Suno automation toolkit for AI agents — programmatically generate songs and pull your liked tracks (audio + lyrics) into a versioned per-song library. Camoufox-based.

Chrome extension that blocks distracting sites — side panel UI, 21 languages, custom shortcut, and a guardian mascot

No description.

Open-source React component library for the Profitmaker.cc trading terminal

Chapter 3: Portfolio Optimization

Chapter 2: Market Microstructure and Order Book Modeling

Landing page for Open Agent Manager

The Algotrading Book: Machine Learning for Crypto Markets — 365 chapters

No description.

No description.

No description.

Generated imagery for Marketmaker.cc (strategies page and beyond)

No description.

No description.

No description.

No description.

No description.

No description.

No description.

Does GARCH(1,1) actually forecast volatility? A controlled, reproducible study (accompanies marketmaker.cc).

Asymmetry, fat tails, and the tail-risk cost of the wrong GARCH innovation - a controlled study.

Volatility targeting with GARCH forecasts: a controlled evaluation of risk stabilization and forecaster choice.

Recovering time-varying correlation with DCC-GARCH: controlled study of tracking + portfolio-risk misstatement.

Copy-paste React components for marketmaker-cc — AI trading platform. Next.js 15 + Tailwind CSS.

Automated video generation from audio narration + PDF slides: sync, encode, subtitle, metadata, thumbnail

Automated YouTube metadata pipeline: transcription, SRT subtitles, tags, timestamps from video + blog article

Roofline-разбор железа под локальный инференс больших MoE-моделей (GLM-5.2): память, KV-кэш, шины, экономика владения против аренды

Fur coat for Claude Code: a thin launcher that chains token-saving proxies (pxpipe, headroom, link-assistant/router) so they layer instead of fighting over ANTHROPIC_BASE_URL.

Ready-made token-saving setup for Claude Code: rtk + caveman + graphify (semantic extraction on cheap OpenRouter/deepseek, not your Claude tokens).

AI coding assistant skill (Claude Code, Codex, OpenCode, Cursor, Gemini CLI, and more). Turn any folder of code, SQL schemas, R scripts, shell scripts, docs, papers, images, or videos into a queryable knowledge graph. App code + database schema + infrastructure in one graph.

No description.

Part of claude-usage-stats workspace

Part of claude-usage-stats workspace

Part of claude-usage-stats workspace

How many backtest 'winners' survive deflation? A controlled study of the Deflated Sharpe Ratio, Harvey-Liu haircuts, and White's Reality Check on parameter searches with known ground truth. Naive best-of-N flags noise 100% of the time; principled deflation controls it; the effective-trial-count trap on correlated grids. Code + paper.

Do you need a GPU to speed up a backtest parameter sweep? A controlled speed ladder for a path-dependent HMA-cross kernel (pandas->numpy->numba->prange->mp+numba, identical PnL): 298x on a laptop CPU, and why a GPU is not the missing piece (finite width + 0.23s budget, not the event loop). Code + paper.

How overfit is your search? A controlled study of the Probability of Backtest Overfitting via CSCV (Bailey-Borwein-Lopez de Prado-Zhu 2017). The null is 0.5, not 1: a no-edge search is a coin flip out of sample; a real edge drives PBO to 0. Known ground truth; complements the Deflated Sharpe Ratio. Code + paper.

How much leakage does purged cross-validation actually remove? Known-truth quantification: phantom skill is an overlap x persistence interaction; purging fixes selection; embargo scope. Code + paper.

Do bootstrap confidence intervals for backtest statistics actually cover? Controlled study: iid vs trade-level vs block bootstrap for Sharpe and max drawdown under serial dependence. Code + paper.

Where does conformal prediction's coverage survive on financial returns and where does it break? Controlled accounting vs exact oracles: GARCH terciles, regime breaks, ACI repair, width costs. Code + paper.

Does walk-forward validation predict out-of-sample performance? Controlled validation of the walk-forward efficiency ratio (WFER), its folk thresholds, and window design. Code + paper.

When does Hierarchical Risk Parity beat Markowitz? Controlled comparison under known covariance: structure families x T/N, Ledoit-Wolf shrinkage, linkage conventions, estimation-error isolation. Code + paper.

Does N_eff = N/(1+(N-1)rho) actually predict how much correlated trading signals diversify? A controlled validation vs Meucci ENB and a correlated-Bernoulli capacity model. Code + paper.

Does 'prefer a plateau over a peak' actually prevent backtest overfitting? A controlled validation against PBO and the Deflated Sharpe Ratio. Code + paper.

No description.

The objective function chooses the strategy: a controlled study of how naive backtest metrics (per-trade Sharpe) select degenerate rare-trade optima, and how an exposure floor + trade-count shrinkage repair them -- though measuring on the full timeline is the real cure. Known ground truth; Gaussian control; code + paper.

Agent skills for the Trender Warehouse crypto market-data service (MCP + REST)

Source articles for the Marketmaker.cc blog + translation task (EN → 11 languages)

Toggle all Claude Code plugins + user MCP servers off/on without restart — global /cctoggle slash command

Go API + scheduler for w_popularity

Track audience growth across social platforms. Next.js + Go + Postgres.

Next.js dashboard for w_popularity

Social-metrics parser (discord) for the socials-auto family

Camoufox-based stealth-scraping HTTP wrapper for w_popularity

Habr parser for w_popularity

Reddit parser for w_popularity

MarketMaker auth users-count parser for w_popularity

GitHub parser for w_popularity (REST API)

T-Bank Pulse parser for w_popularity

Instagram parser for w_popularity

Stack Overflow parser for w_popularity

Facebook parser for w_popularity

LinkedIn parser for w_popularity

YouTube parser for w_popularity

X (Twitter) parser for w_popularity

Telegram parser for w_popularity

Social-metrics parser (github-repo) for the socials-auto family

Smart-Lab parser for w_popularity

Shared Go types and Parser interface for w_popularity

No description.

ZigBolt Documentation — Ultra-Low Latency Messaging System for HFT

ZigBolt C bindings — header and examples for the ultra-low-latency HFT messaging library

ZigBolt Rust bindings — safe FFI wrappers for the ultra-low-latency HFT messaging library

ZigBolt TypeScript/Node.js bindings — FFI wrappers for the ultra-low-latency HFT messaging library

ZigBolt Go bindings — cgo wrappers for the ultra-low-latency HFT messaging library

ZigBolt Python bindings — ctypes wrappers for the ultra-low-latency HFT messaging library

No description.

Portfolio optimization workspace — 11 algorithm crates as git submodules + Rust backend

HRP (Hierarchical Risk Parity) cross-language benchmark — Node.js

HRP (Hierarchical Risk Parity) cross-language benchmark — C++17

HRP (Hierarchical Risk Parity) cross-language benchmark — C reference implementation

HRP (Hierarchical Risk Parity) cross-language benchmark — Bun

HRP (Hierarchical Risk Parity) cross-language benchmark — Rust

Next.js frontend that visualizes Karabiner-Elements rule JSON as interactive SVG keyboard layouts

Codeforces Round 2229 problems

HRP (Hierarchical Risk Parity) cross-language benchmark — Zig

HRP (Hierarchical Risk Parity) cross-language benchmark — Python (numpy + scipy)

No description.

Professional crypto scalping terminal built with Rust & Iced

Workspace: MarketMaker Scalper — Rust/Iced terminal + Go backend + Astro landing

Portfolio optimizer — Next.js front-end for the portfolio-workspace algorithms

No description.

The Algotrading Book — Astro Starlight website with EN/RU support

Workspace umbrella: karabiner rules + visualizer as submodules

Composite portfolio optimization pipeline — HRP + long/short split + CVaR/Hull-White overlay (Rust)

Theory-Implied Correlation portfolio optimization (Rust)

Robust Bayesian Allocation portfolio optimization (Rust)

Online Portfolio Selection (Rust)

Entropy Pooling portfolio optimization (Rust)

Nested Clustered Optimization portfolio optimization (Rust)

Black-Litterman portfolio optimization (Rust)

Mean-Variance (Markowitz) portfolio optimization (Rust)

Generalized Hierarchical Risk Parity portfolio optimization (Rust)

Modified Hierarchical Risk Parity portfolio optimization (Rust)

Hierarchical Equal Risk Contribution portfolio optimization (Rust)

Hierarchical Risk Parity portfolio optimization (Rust)

my config for karabiner

Prediction market arbitrage scanner & executor bot. Polls Premarket.me API for cross-platform opportunities (Polymarket, Kalshi, Predict.fun, Opinion, Probable, Limitless).

Chapter 10: Regulatory и Risk Management — mathematics for trading

Chapter 9: Crypto Specifics - DeFi and MEV

Chapter 8: Options and Greeks for Trading

Chapter 5: Low-Latency Trading Systems

Chapter 4: ML for Financial Time Series

Warp is an agentic development environment, born out of the terminal.

Workflows and skills to help people and agents collaborate on open-source software with the power of Oz!

Huashu Design · HTML-native design skill for Claude Code · Claude Code 里 HTML 原生的设计 skill · 高保真原型 / 幻灯片 / 动画 + 20 设计哲学 + 5 维评审 + MP4 导出 · Agent-agnostic

Real-time global intelligence dashboard. AI-powered news aggregation, geopolitical monitoring, and infrastructure tracking in a unified situational awareness interface

Use multiple Claude Code–compatible LLM providers from one CLI, switching profiles instantly with simple clother-* commands.

No description.

No description.

Part of claude-usage-stats workspace

No description.

No description.

No description.

No description.

No description.

No description.

MacChill — macOS menu bar app to monitor thermal pressure & auto-toggle Low Power Mode on Apple Silicon

MacChill macOS app (SwiftUI, Apple Silicon M1-M4)

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.