What is VALDX?

VALDX is Valspresso’s patented Sentiment and Fundamental Data Feed, distributed by FactSet. VALDX is designed to generate persistent Alpha under various market conditions by reshaping risk.

What is the most important feature of VALDX?

Predictive Power. VALDX indicators and classification features uniquely predict future risk and return of companies every day. In our whitepaper, we demonstrate how Valspresso’s indicators and classifications are used to predict, with statistical precision, the risk and excess returns of companies over the next 30 days.  This predictive power gives quants and asset managers an edge to cost effectively build and deploy their own proprietary active equity strategies designed to deliver higher Alpha while lowering risk under various market conditions.

How is VALDX produced?

Valspresso performs AI-driven automated analysis of companies’ financial statements filed with the SEC. Price sentiment and fundamental indicators are calculated using Valspresso’s proprietary rules engine, coupled with pricing, financial statement data, and other data points from multiple sources. As these sources are analyzed daily, Valspresso’s algorithms perform a series of repair, linking and calculation processes that identify and calculate missing financial statement values, reconstruct history to represent point-in-time records, and generate VALDX’s indicator values. Scores and indicators are only provided for companies with at least three years of historical financial statement data.

Which companies do you cover?

We cover public companies traded on U.S. exchanges.  As of 6/28/2019, we cover 6,294 companies, 4,760 of those are on the major U.S. exchanges. For the S&P 500, as of 6/28/2019, we cover 480 companies (i.e. 96% coverage).  Some companies are not covered due to requirements that our algorithms match SEC filings for companies to pricing history for securities. Our mapping engine may not reconcile some securities to financial statements because of complicated corporate actions, multiple securities assignments to a single company, or non standardized rendering of ticker symbols.
Note that for indexes such as the S&P 500, a company may be represented more than once by different securities.  This is because a single company can issue multiple share classes. VALDX only represents a company once using the company’s primary security.  Therefore, even though the S&P 500 may contain 505 securities, we will at most contain 500 companies.

How do missing companies affect analysis and backtesting?

Missing companies are randomly distributed and therefore do not materially impact analysis or backtesting.

How do I get access to evaluate VALDX?

I have a subscription to VALDX, how do I evaluate the data?

Below are suggested steps for evaluating VALDX:

1. Review our short-term outperformance probabilities for each of our indicators (or classifications). In our technical whitepaper, we demonstrate how, with statistical significance, the risk and returns of companies can be predicted over the next 30 days. Compare those probabilities to your own indicators to see how we compare.

2. Try using our indicators or classifications in a simple backtested strategy. We have provided a simple use-case in our whitepaper called “Sleeper” as a good place to start. While classifications do not address all the real-world concerns of trading strategies, it does give you a baseline for evaluating Alpha.

3. Try incorporating our indicators or classifications into your own existing strategies (or build new stock selection Active Equity Strategies)  and see how it may amplify performance. 

When did Valspresso start producing the data feed?

Valspresso started producing live VALDX data feed in 2011 while the system was in beta. 2012 was our first full year of live production of data. Prior to 2011, Valspresso’s algorithms backfilled to 2003 (the beginning of Sarbanes-Oxley Act financial statement data) to represent the values our system would have calculated had we been in operation.

What is “reshaping risk”?

Risk, for purposes described here, is the distribution of returns relative to a benchmark.  For example, if you were to randomly choose a component of the S&P 500, there is a 50% chance you would outperform the S&P 500 Equal Weighted index.  There is also a 50% chance you would underperform the index. If you ignore the magnitude of component returns and simply rank components (e.g. from 1 to 500), this distribution would be uniform (flat).  This uniform distribution would be “market risk,” i.e. the risk you assume by buying components of the index and thus yield the 50% chance of under or over-performing the index. But if we could reshape this distribution so it was no longer uniform (flat), then we could have a greater than 50% probability of outperforming the index.  The probability is simply the area under the density curve of the distribution. Those density curves with larger areas on the right side indicate a higher probability of outperforming the market. This altering of the density curve of returns is what we mean by “reshaping risk”.

What is the purpose of the sentiment classification column?

The sentiment classification column represents ranges within the sentiment index (SI) indicator.  Those ranges are documented in the whitepaper and represent different interpretations for SI depending on which range it is located.  For example, a SI less than -1 has a different meaning from an SI that is between -1 and 0. SI also has a different meaning when it is a positive value.  Because the interpretation of SI changes depending on its value, the sentiment classification column provides a convenient way of qualifying the SI value. This qualification should be incorporated into your trading strategies.