Private Markets in 2031

Preqin forecasts provide a complete, data-driven, and granular overview of private markets. The forecasts produced by our model are intended to provide a base case for further analysis. The results are published annually and made available to Private Markets Research subscribers in October of each year. In 2026, we made predictions for 87 private markets strategy groups in four regions. 

We model the private market as a system consisting of sources of funding, aggregate cash flows, and capital accumulation. This system replicates the flow of capital through private capital markets and allows us to forecast multiple market characteristics (e.g., growth rate, performance, capital raised, and the ratio of dry powder to unrealized value) concurrently. This approach is designed to ensure that our forecasts are internally consistent and that the model will generate plausible scenarios for the future. 

The principal starting points of the model are fundraising and performance, for which we generate future values using a statistical method called ensemble forecasting. This year, our forecasts are supported by risk factors from BlackRock’s investment factor model, which are leveraged as explanatory variables. In private markets, the data is noisy and the number of available observations is limited. Our approach collects partial information from over 39,000 financial series. Next, we combine the output from a population of models, giving more prominence to those that were consistently predictive on data not used to train them. This allows us to deliver robust forecasts that are burdened with less risk. 

Once we have generated expected future values of capital raised and performance, we build forecasts of capital called and capital distributed. We combine these with historical values for dry powder and unrealized value, allowing us to make projections for both metrics.¹ Their sum gives us assets under management (AUM) – a measure of total market value. We repeat this process for each of the 87 strategy groups. Next, we aggregate all data series to generate values for the market, regions, asset classes, region–asset class pairs, and strategy groups at the global level. 

Every forecasting method assumes that certain data characteristics remain constant over time. The reliability of predictions generated by our approach relies on two such characteristics: the first is the patterns or long-term trends in each forecast data series; the second is the relationship between the predicted values and the observed variables. We test these jointly to ensure they are sufficiently reliable to consistently deliver useful predictions. 

Our model disentangles the effects of capital raised and performance on AUM. In this sense, it is causal. A limitation is that it does not fully identify the indirect impact of changes in macro variables – such as shifts in the sovereign yield curve. As a result, our model may not measure the total impact of any individual factor (e.g., monetary tightening). It also cannot answer questions that are counterfactual. 

Answers to FAQs 

• Whenever available, we use BlackRock’s Capital Market Assumptions (CMAs) as baseline performance numbers 

• All projections are net of fees 

• Where performance is provided as a percentage, it always represents internal rate of return 

• Our performance figures are nominal (that is, they are unadjusted for inflation) 

• In the case of private credit evergreen structures, we only report a top-down estimate of AUM. We do not account for evergreen structures when we forecast aggregate capital raised or performance 

• Dry powder, unrealized value, AUM, capital raised, and performance forecasts all exclude funds denominated in yuan renminbi. These funds are deemed less investible for many of our clients 

• We do not make assumptions about any macro variables (such as interest rates). When we forecast performance and fundraising, we take historical values and third-party forecasts as inputs 

• Since funds do not report the breakdown of their investments by country, we cannot produce predictions at a country level