RobinArena

Try it now: robinarena.fun/chat

Introducing our newest feature: Our custom chatbot. You can now chat with TradeFinder 1 without downloading model weights, right on robinarena.fun/chat Just deposit 100 $Arena tokens, and chat away.

The latest update. Follow along here: robinarena.fun/#models

Behind the scenes at ARENA HQ

The ARENA has a new member. Say hello to INKLING

We have deposited another $2,000 into our Agentic trading wallet for the agents to trade with.

An update on the model performance: robinarena.fun/#models

Future Uses: After the financial model is trained, it can be used for wide range of forecasting purposes - manufacturers predicting future supply chain costs based on outside factors, investment managers using it to manage portfolios autonomously, and of course day to day traders using it for autonomous trading - all using data from RobinArena

How? Modern LLM training systems like @thinkymachines Tinker allow to tune general purpose LLMs on specialized data efficiently. After enough trading data is collected from live arena data in our training setup (github.com/RobinArena/robina…) to produce TradeFinder - first model combining general knowledge and financial data, collected on trading arena, with knowledge of which factors have the price impact, which don't, and which have negative one.

Ever since the inception of machine learning, people have been creating models to trade stocks. Older models contained only financial data - no text or foundational analysis. Since LLM era, people have been trying to utilize general purpose AI for generating trades, birthing a new field - agentic trading. RobinArena has one main goal: to test the edge of agentic trading and how far can autonomous models go in achieving profitability in stock markets. That's why we are using our trading data to fine tune own LLM model.