Scholarlink

In the current landscape, Venture capitalists are opting for more data-driven decision-making to achieve high returns while mitigating risks. All VCs seek a home run in their portfolio, for which it is essential to select the optimal exit. The process can be smoothed with the prediction of the type of exit- IPO, M&A, or the next funding round. Further,it is also to gauge which type of exit yields maximum profit for the particular climate. In this paper, through the utilisation of six machine learning models - Elastic Net, SVR, Kernel Ridge, Gradient Boost, XGBoost, and Random Forest, trained on factual data sourced from Crunchbase. The machine learning models are evaluated on the basis of 2 metrics- RMSLE and R-squared Score. We propose an ensemble learning model with Root Mean Square Logarithmic Error (RMSLE) score of 0.29 and R-squared score of 0.76. This weight-averaged ensemble has been validated to predict the optimal exit for a startup under six broad labels - IPO, Acquisition, Leveraged Buyout, Merger, Acquihire, and Management Buyout.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Predicting Startup Exit Strategies with Ensemble Machine Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • M. Deekshitha Reddy,
  • Geetika Vadali,
  • Garima Jaiswal,
  • Ritu Rani,
  • Amita Dev

摘要

In the current landscape, Venture capitalists are opting for more data-driven decision-making to achieve high returns while mitigating risks. All VCs seek a home run in their portfolio, for which it is essential to select the optimal exit. The process can be smoothed with the prediction of the type of exit- IPO, M&A, or the next funding round. Further,it is also to gauge which type of exit yields maximum profit for the particular climate. In this paper, through the utilisation of six machine learning models - Elastic Net, SVR, Kernel Ridge, Gradient Boost, XGBoost, and Random Forest, trained on factual data sourced from Crunchbase. The machine learning models are evaluated on the basis of 2 metrics- RMSLE and R-squared Score. We propose an ensemble learning model with Root Mean Square Logarithmic Error (RMSLE) score of 0.29 and R-squared score of 0.76. This weight-averaged ensemble has been validated to predict the optimal exit for a startup under six broad labels - IPO, Acquisition, Leveraged Buyout, Merger, Acquihire, and Management Buyout.