With the expansion of financial markets and the surge in financial text data, accurately recognizing professional entities associated with corporate investment has become crucial for financial analysis and decision-making. However, existing named entity recognition (NER) methods still struggle with accuracy, especially when identifying complex and specialized entities related to enterprise investment. To solve this problem, we propose a novel model, FBMG (FinBERT-BiLSTM-MA-Global Pointer), for Chinese named entity recognition in the field of financial investment. Firstly, the pre-trained financial language model FinBERT was used to generate dynamic semantic vectors based on the context to obtain semantic information of corporate investment corpus. Secondly, the BiLSTM-Global Pointer model architecture is used to capture long-range textual information, nested and non-nested named entities are considered from a global perspective, and a multi-head attention mechanism is incorporated on this basis for information weight assignment to fully exploit the semantic features of the text. Furthermore, considering the characteristics of different entities, the model achieved the best F1 score of 84.69% using rule-based and lexicon data enhancement, compared with other models. The experimental results show that the proposed method effectively improve the identification of corporate investment named entities, which can be a reference for similar financial entity recognition.