2 Tactics to "Moneyball" Your Funding Company

In the past entries of this series on using data analytics to supplement decision making in legal funding, we've examined initial steps to using data analytics, including gathering data, cleaning data and structuring data.

Now, we'll talk about some actual analytical techniques that can be used once you have clean, relevant data in order to help you make decisions. After reading this post, you'll have armed yourself with the baseline knowledge and vocabulary of analyses you can run at your funding company.

There are two major types of data analytics that we'll take a deeper look at in this post -univariate analysis, sometimes called descriptive statistics, and multivariate analysis. 

Univariate Analysis

Univariate analysis examines one variable at a time, often looking at trends in that variable.

In my experience with legal funding, I've found there are a number of very useful univariate statistical tests, but 3 stand out as being most important and useful for legal funders when pitching or reporting to investors.

Each of these stats is closely tracked by institutional investors who assess portfolio health in litigation funding claims.

Finding the moving average of portfolio returns to investment - this statistic helps investors understand whether their portfolio performance is improving or deteriorating. Think of it as the equivalent of tracking quarterly gross margins in a retail setting or net interest margin for a bank.

Average time to resolution of a case - much like how retailers and manufacturers care about turning over inventory quickly, litigation financiers have the same goal with their "inventory" of investments.

If this is a metric you wish to optimize your business for, there are two primary ways to decrease the time to resolution.

First, you can identify which case types, jurisdictions, or attorneys/law firms produce the quickest returning cases.

Second, adopt a "brokering" strategy in which after cases hit a pre-determined age your team actively attempts to get them bought out by another funder by using either an in-house or external broker.

Median return on a case in portfolio - means or averages are easily skewed by unusually big or small data points, such as those belonging to big victories on rare cases. When it comes to assessing expected return of a matter one is considering in the future, however, the median or 50th percentile is often more useful.

Tracking these statistics will not only help your internal business functions, but also help you accurately report to current investors and build credibility when pitching potential investors.

Multivariate Analysis

In addition to univariate statistics, legal funders should consider how different variables in their data are related to one another. Any statistical technique used to analyze information arising from more than one variable is called multivariate analysis.

Most often in legal funding, this means looking at case characteristics as they relate to case duration or case returns. For example, one might want to assess the correlation between a case's duration and its locale, or the type of case (using dummy variables to represent region or case niche as discussed in the previous entry on structuring data).

It's beneficial to go beyond this process and examine regressions as well. A regression is a means to look at multiple different variables simultaneously and assess their impact on a variable of interest. For example, one could look at the impact on case duration on referring law firm, case type, geographic region, case size and plaintiff investment.

Using a regression gives funders the opportunity to understand how many different factors impact a single variable of interest.

A Note on Correlation vs. Causation

When using regressions, it is important to be cautious and have an explicit objective before actually running the test.For example, one might find that cases in which attorneys collect higher fees are, on average, more profitable after controlling for other factors.

This outcome has two possible explanations:

1. Higher-paid attorneys are more capable and generate better results for their clients and litigation finance firms investing in those matters. 
2. Higher-paid attorneys are more willing to accept promising cases with higher expected returns. 

Either explanation is plausible here, and so we must be careful not to assume that higher paid attorneys cause better case outcomes.

In other words, the regression alone does not give us a causal explanation; there are plausible explanations just based on correlation. Funders must complete additional tests and analysis so as to decide which explanation is correct  Such is the subject for the next post.