The MFWire
Manage Email Alerts | Sponsorships | About MFWire | Who We Are

Subscribe to MFWire.com's News Alerts [click]

Rating:Machine Learning Comes to Fund Ratings Not Rated 0.0 Email Routing List Email & Route  Print Print
Wednesday, March 07, 2018

Machine Learning Comes to Fund Ratings

News summary by MFWire's editors

Morningstar chief Kunal Kapoor has extended the investment research giant's forward-looking ratings to cover the vast majority of U.S. open-end mutual funds and ETFs, thanks to machine learning.

On Monday Jeffrey Ptak, global head of manager research, and Lee Davidson, head of quantitative research, unveiled M*'s new Morningstar Quantitative Rating system, the companion to the forward-looking Morningstar Analyst Rating system that M* launched in 2011. The two systems combined will cover all but four to five percent of the U.S. fund universe, Davidson tells MFWire.

M*'s Tim Strauts takes a deeper dive into the logic behind the new rating system, and here's more information on the methodology. InvestmentNews, CityWire, and Fund Action also covered the news.

The only funds uncovered by both the quantitative ratings and the analyst ratings, Davidson says, fall into one of three categories: M* has a conflict of interest, the fund is brand new, or the fund is in what M* classifies as an un-ratable category (like leveraged funds). Note that these restrictions are different from those for M*'s original, backward-looking star ratings, which require a minimum performance track record and leave more than 22 percent of funds unrated.

In simple terms, M*'s new quantitative ratings are created by drawing on the analyst ratings (and the rationale behind those ratings) to come up with similar ratings for funds not currently covered by any of M*'s human analysts. Like the analyst ratings, the quantitative ratings will be built out of sub-ratings in five pillars (parent, people, performance, price, and process), and the final ratings will come in five flavors (gold, silver, bronze, neutral, and negative).

"We've trained our machine-learning model to emulate how our analysts make decisions, greatly expanding fund coverage," Davidson states.

Like with the analyst ratings, the quantitative ratings' forward-looking nature makes them inherently more responsive to things like PM changes that may take longer to affect a backward-looking assessment like a star rating. Fundsters and PMs with smaller funds, be ready for more ratings to head your way. 

Edited by: Neil Anderson, Managing Editor

Stay ahead of the news ... Sign up for our email alerts now

 Do You Recommend This Story?

Return to Top
 News Archives
2019: Q4Q3Q2Q1
2018: Q4Q3Q2Q1
2017: Q4Q3Q2Q1
2016: Q4Q3Q2Q1
2015: Q4Q3Q2Q1
2014: Q4Q3Q2Q1
2013: Q4Q3Q2Q1
2012: Q4Q3Q2Q1
2011: Q4Q3Q2Q1
2010: Q4Q3Q2Q1
2009: Q4Q3Q2Q1
2008: Q4Q3Q2Q1
2007: Q4Q3Q2Q1
2006: Q4Q3Q2Q1
2005: Q4Q3Q2Q1
2004: Q4Q3Q2Q1
2003: Q4Q3Q2Q1
2002: Q4Q3Q2Q1
 Subscribe via RSS:
Add to My Yahoo!
follow us in feedly

©All rights reserved to InvestmentWires, Inc. 1997-2019
14 Wall Street | 20th Floor | New York, NY 10005 | P: 212-331-8968 | F: 212-331-8998
Privacy Policy :: Terms of Use