3 ms·
I'm skeptical frontier LLMs can actually do well (e.g. alpha 5%+) without fine-tuning, especially on historical market data. Presumably you support fine-tuned m
by cromwellian 2mo ago
I'm skeptical frontier LLMs can actually do well (e.g. alpha 5%+) without fine-tuning, especially on historical market data. Presumably you support fine-tuned models?
- RuiWang0811 2mo agocofounder here - LLMs can do some model training, they train on ML competition data after all. But they do struggle with low signal to noise ratio of market data. But that’s exactly what our environments will teach.