Skip to content
View h-a-forster's full-sized avatar

Block or report h-a-forster

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse

Popular repositories Loading

  1. Quantitative-Backtesting-Engine-for-Rule-Based-Trading-Systems-Python- Quantitative-Backtesting-Engine-for-Rule-Based-Trading-Systems-Python- Public

    Python backtesting engine. Evaluates systematic trading strategies. Includes support for trend-following, mean-reversion, time-series momentum, volatility-targeting. Provides equity curves, drawdow…

    Python 1

  2. A-Microstructure-Playground-LOB-Dynamics-and-a-Market-Maker A-Microstructure-Playground-LOB-Dynamics-and-a-Market-Maker Public

    A Python limit order book simulator with price–time priority and synthetic order flow, plus an inventory-aware market maker. Built as a learning sandbox to study spreads, depth, adverse selection, …

    Python 1

  3. Algorithmic-Trading-The-Low-Volatility-Momentum-Rotator-L-VMR- Algorithmic-Trading-The-Low-Volatility-Momentum-Rotator-L-VMR- Public

    This project implements a rule-based, low-turnover trading strategy for Swedish large-cap equities. The strategy rotates between a concentrated basket of OMXS30 constituents and a defensive risk-of…

    Python 1

  4. agent-ab agent-ab Public

    Controlled A/B experiments for coding-agent setups: instructions, skills, models, effort and prompts, measured on hidden tests with paired statistics.

    Python 1

  5. skinflint skinflint Public

    Local proxy for the Anthropic and OpenAI APIs: hard spend caps for coding agents, a context token profiler, and a prompt-cache doctor.

    Python 1

  6. shadowgate shadowgate Public

    Confidence-gated LLM cascades that measure how often the cheap tier is wrong on the cases it answers alone.

    Python 1