Snapshots of Exotic Callable Structured Products Valuation via Randomized Neural Networks

Key Points

• This research explores the effectiveness of randomized neural networks for valuing complex callable structured products.
• Utilized randomized neural networks to assess option premiums for callable structured products.
• Compared results with Longstaff–Schwartz valuation method.
• Evaluated computational efficiency against traditional fully trained neural networks.
• Randomized neural networks provided higher estimated option premiums than Longstaff–Schwartz in most scenarios.
• Significantly reduced computational requirements compared to fully trained neural networks.