Hey everyone, I’m Joseph. I’m an electrician by trade, an active trader, and definitely not someone who came into this with a traditional software-development background. I started this project because I couldn’t find a trading platform that approached the market the way I wanted. Most of the “AI trading” products I found were either long-term investment tools, basic signal generators, or black boxes making decisions without enough explanation or control. I’m interested in something very different: an agentic system designed specifically to observe fast-moving market conditions, develop a trade thesis, manage risk, and explain what it sees—without treating an LLM’s opinion as permission to place a trade. The project has grown into a fairly substantial trading platform. It brings together market data, news and catalyst analysis, technical structure, risk controls, strategy selection, position management, journaling, replay, and decision evidence. The goal is for the different parts to work together as a supervised trading team rather than as one model trying to do everything. A few principles have shaped the build: - AI can interpret context and coordinate decisions, but hard risk limits and execution permissions should be deterministic. - Missing, stale, contradictory, or ambiguous information should cause the system to stop—not guess. - Every important decision should be traceable to the evidence that supported it. - A human approval button is only a real safeguard if the human receives enough information to make an independent decision. - Paper trading and replay are for collecting evidence and exposing weak logic, not just producing a pretty profit number. At the moment, the core platform and supervised orchestration system are built. I’m now working through controlled live-data validation, paper-trading evidence collection, edge cases, and the boundaries between analysis, recommendation, approval, and eventual execution. The long-term goal is greater autonomy, but only after each earlier stage has demonstrated that it can fail safely and explain its behavior.