Ann: Investment Portfolio Report at 30 September 2021, page-27

yeah the interesting thing about court cases is that though they are random, they are truly *independent* random events. That is, the outcome of one case has no correlation with the other. So the variance of a large basket of cases really does reduce in proportion to sqrt(N), unlike say, owning N stocks, or a mortgage backed security comprising N mortgages - in which the performance of the units is positively correlated with the rest.

I've not crunched the numbers myself, but assuming their win rate is constant (ie they're not forced to lower their standards for taking on cases with the larger funding level), you ought to be able to plug in the historical win rate and get a pretty reliable estimate of their earnings going forward.