Soon we will release an improvement to the risk model for more accurate risk exposure by discounting the effect of return outliers on the calculation of beta. This will be out shortly, here's what's coming.

Quick recap: What is the risk model?
The risk model decomposes the risk of holding any stock (or portfolio of stocks) into a set of common risk factors and a residual risk. That residual risk is called asset/portfolio specific risk or specific risk for short. Colloquially, the return from specific risk is also referred to as alpha. It was released on the platform in November and you can read more about it in this thread.

Where can I use it?
You can use it in an algorithm, by loading pipeline factors around the risk factors and using the output data in constraints in the optimizer. This will allow you to select your universe based on a set of exposures and later limit the exposure in your target portfolio.

What's the latest improvement?
We detected outliers in the return streams and sector residuals, which in turn affected the beta calculation per-stock and per-sector. This means if your algo was constraining on beta, and picked up a stock with an extreme beta value, the optimizer would try to compensate to keep the portfolio beta neutral, and create a basket of stocks to counter this exposure. The outliers have now been clean up and removed, to closer reflect the true beta values.

For example, below you can see the return history of the security SAGE. The plot shows 3 outlier days, which when included in the original risk model release, tilted the linear regression and skewed the beta perception for the security. By removing these outliers, we have a more consistent view of the stock's return stream, and thus beta calculation.

To learn more about risk constraining your algorithm, check out this lecture. Enjoy!