Both entry and exit prices are utilized in training the privateMetrics data. The privateMetrics asset pricing model is calibrated using a representative set of approximately 10,000 private market transactions, including both entry and exit transactions, since 2013. These transactions are collected from sources such as Pitchbook, Capital IQ, and fund manager annual reports.

Key aspects of how entry and exit prices are used:

• Comprehensive Transaction Data: The model incorporates all primary transactions in private companies, encompassing entry, exit, and take-private transactions. Fund-level transactions and fund-reported Net Asset Values (NAVs) are excluded.
• Calibration Process: The model is calibrated monthly with the latest market transactions to update its estimation of the price of risk for each factor. This dynamic estimation process uses a Kalman filter to retain the memory of past factor prices while allowing new transactions to influence the relationship.
• Factor Prices: By observing the relationship between various factors (e.g., size, revenue growth, EBITDA margin, leverage, maturity, and country risk) and valuation among reported transactions, the model infers how much premium or discount an investor is willing to pay for purchasing a private company.
• PECCS Classification: Transactions are classified by PECCS® segments, which cover aspects such as activity, lifecycle phase, revenue model, customer model, and value chain. These classifications are used as control variables in the model.
• Price-to-Sales Ratio: The model uses the Price-to-Sales (P/S) ratio of observable transactions as the modeled variable. The P/S ratio is favored due to its statistical properties, stability, and applicability to loss-making companies.
• Shadow Pricing: The calibrated model is used to shadow price a broad market universe for which factor exposure data is available. This involves estimating the prices of a large number of private companies based on their characteristics and the calibrated factor prices.
• Model Robustness: The model’s robustness is confirmed by the close alignment of average observed transaction prices and model-predicted values (shadow prices), even at the individual market segment level.
• Addressing Potential Biases: While the model is based on P/S ratios, it also ensures that predictions for EV/EBITDA are not biased by computing the EV based on the book value of debt and predicted equity valuation. These predictions are compared to transaction-implied ratios.