Downloads · 30 days
3
10% of all-time downloads
lucas-duartedelima/qwen2.5-7b-earnings-volatility-lora
qwen2.5-7b-earnings-volatility-lora is a text generation model from lucas-duartedelima. Use it when you need the model to write or continue text. It is set up for peft.
This repository contains a LoRA fine tuned adapter for Qwen2.5 7B Instruct. The model was fine tuned for an MSc Finance thesis to predict annualised realised stock volatility over the 63 trading days following an earn…
Downloads · 30 days
3
10% of all-time downloads
All-time downloads
31
Public
Repo size
173 MB
Likes
0
Public
Click a slice to open those files.
.safetensors162 MB · 93%
From the Hugging Face model README
This repository contains a LoRA fine tuned adapter for Qwen2.5 7B Instruct. The model was fine tuned for an MSc Finance thesis to predict annualised realised stock volatility over the 63 trading days following an earnings call, using CEO responses from the Q&A section as input.
Qwen/Qwen2.5-7B-Instruct
Input: CEO earnings call Q&A text
Output: Numerical forecast of 63 trading day annualised realised volatility
The model was fine tuned using QLoRA with a supervised instruction format. The base model was loaded in 4 bit quantised format and LoRA adapters were trained for the volatility prediction task.
This model is intended for academic research on financial textual analysis and volatility forecasting. It should not be used for investment decisions.
The model was trained on a limited empirical dataset and may not generalise across all firms, sectors, or market regimes. Realised volatility is affected by many events outside the earnings call transcript.
Developed as part of an MSc Finance thesis on fine tuning large language models on earnings calls.