Menhir: Quants for distressed debt
• Distressed credit investors often approach book buying with little data about the incoming portfolio
• This creates huge bid-ask spreads as sellers are not willing to share more information, wile buyers are not confident making a purchase decision with little information
• The client, an investment fund, approached Menhir to understand the potential future cashflows that a certain portfolio could generate
• The task was to create a book pricing algorithm with no prior experience in distressed loan book pricing
• Frequency & recency clustering: Menhir identified four clusters, related to the payment probability of those segments in the book. With that information, Menhir could anticipate loan cashflows at a more granular level by harnessing the power of unsupervised clustering models
• Book pricing model: Based on the output from the cash flow model, Menhir was able to price the book by the traditional DCF models. Using the output from the ML algorithms as input for valuation models
• Due Diligence time reduced from weeks to 48h
• Model provided the same price as a team of seasoned advisors