Nparam Bull is a fast inference model that predicts the instantaneous price fluctuation of stocks given raw textual market data. Bull was trained on decades of financial and market news correlated with the daily percent changes of 100 top stocks; the whole model is ~7.5 billion parameters. This marks a new model paradigm called Large Stock Models (LSM), similar to the transformer architecture found in LLMs. Here, stock are represented with high dimensional contextualized embeddings, not just word tokens.