ETF Comparison, Rankings and Ratings

Source: CFRA Research; Ratings as of Dec 31, 2023; 2024 Returns as of Dec 31, 2024 

We see a similar trend for US sector ETFs, where the 5 STAR ETFs returned 20% on average in 2024, while 1 STAR funds returned 7% (see Figure 2).  

Figure 2: Average ETF Return in 2024 by CFRA Rating for U.S. Sector Equity ETFs 

Source: CFRA Research; Ratings as of Dec 31, 2023; 2024 Returns as of Dec 31, 2024 

It is important to note that no ratings system is perfect, and a higher STAR rating is not guaranteed to be indicative of higher future returns. However, as shown in Figures 1 and 2, ratings can be a very useful guide in the ETF selection and portfolio construction process to drive higher risk-adjusted returns.  

A Framework for Designing a Robust ETF Rating Methodology 

Using a ratings system that is designed specifically for ETFs and that uses recent technology is critical. Some legacy ratings systems were designed for traditional mutual funds and tend to rely solely on past performance. By contrast, CFRA uses a quantitative machine-learning based model that is designed specifically for ETFs. A robust ETF rating system must have the following features: 

• Multi-factor: Many legacy ratings systems primarily use historical returns to predict future performance. Ideally, a good ratings system should incorporate multiple factors such as cost, reward and risk parameters.
• Constituent Holdings Based: Most ETFs are indexed, so they replicate a benchmark rather than trying to beat it. Therefore, the ratings system should be based on evaluating current constituent holdings, rather than just measuring relative performance against a broad benchmark.
• Timely: Ideally an ETF should be rated within a few months of its launch, rather than needing to wait for 3 years to be rated.  
• Forward looking: A good ratings system should attempt to model the probability of future outperformance of an ETF relative to its peer group, so that investors can use the rating for portfolio construction.  
• Dynamic Adjustment of Factors: Some input factors may be more effective in predicting future outperformance, and these may change over time. A well-designed model must learn based on its track record and dynamically adjust its factor weightings over time.  
• Smoothing algorithm: Finally, it is important that ratings have stability. If a rating flips dramatically from 5 STARs to 1 STAR or vice versa, it makes the system difficult to use in practice. Ideally a good model will use algorithms that provide stability in ratings while still providing flexibility for ratings to change over time.  

Having a robust system that incorporates all the elements above will help advisors and investors make more informed investment decisions.  

CFRA’s ETF Rating Methodology 

CFRA’s ETF ratings are designed to integrate the key elements described above. Using a machine learning based technique, CFRA’s model interprets a wide array of investment signals related to the total cost of ownership, expected return, and downside risk. Eligible equity and bond ETFs receive ratings ranging from five-stars (highest) to one-star (lowest) to identify which funds have the highest and lowest probability of outperformance relative to a similar group of ETFs over the next nine to twelve months. In addition to an overall STAR Rating, the model assigns three sub-model ratings across Reward, Risk and Cost sub-categories to assess a fund’s position relative to similar ETFs based on relative percentile ranking 0-100. In each category, scores of 100 are most favorable and imply the fund is positioned favorably relative to a similar group of ETFs. 

Figure 3: An example of CFRA ETF Rating Category Scores.

A key input into the model is the use of CFRA’s proprietary research on the underlying stock holdings of every equity ETF. This includes STARS—CFRA’s fundamental research assessment of an equity’s 12-month investment outlook. It also includes CFRA Earnings Score—CFRA’s forensic accounting assessment of a company’s earnings quality. In addition, CFRA ratings incorporate fund-specific characteristics, including expense ratio and other parameters such as price-NAV premiums. For equity ETFs, CFRA incorporates metrics like investor sentiment and financial statement analysis. For fixed income ETFs, the model uses factors such as yield and duration, a measure of interest-rate sensitivity. 

The quality of the data underpinning the model is essential to having reliable results. CFRA uses its proprietary ETF database to source data on granular ETF classification and reference information, constituent holdings and weights, as well as daily return and flow statistics.  

Finally, every eligible ETF is typically rated within 2 months of its launch. The model scores and overall ratings are refreshed monthly with the parameter weightings shifting based on machine-learning techniques. A smoothing algorithm is also applied to prevent ratings from ‘flipping’ too frequently. This methodology is designed to ensure accuracy, stability and reliability in ratings. 

Benefits of CFRA’s ETF Ratings Approach  

CFRA’s unique approach to ratings offers significant benefits to advisors and investors as summarized in Table 1 below. 

Table 1: Benefits of CFRA’s ETF Ratings Methodology

Using CFRA’s ETF Tools in an Investment Workflow 

Users can access CFRA’s ETF tools via the MarketScope Advisor platform. The platform provides access to the FUNDynamix tab, a section designed specifically for ETF analysis. 

Users can start their ETF analysis by screening the ETF universe using a granular screener. This screener (shown in Figure 4 below) allows users to sort the ETF universe on multiple parameters including categories such factor, theme, geography, index, cost and other fund level parameters. CFRA ratings can also be used as a screening criterion.  

Figure 4: Granular Screener for ETF Sorting & Rankings 

Screened and selected ETFs can then be compared and ranked (see Figure 5) on multiple parameters including cost, return, risk, net inflows and ratings.  

Figure 5: Granular Screener for ETF Sorting & Rankings 

Finally, users can do a deep dive into individual ETFs to view their ratings and category scores as well as other data items such as changes in sector exposure or net flows over time.  

Figure 6: Granular Screener for ETF Sorting & Rankings 

CFRA’s research and tools are used by tens of thousands of financial advisors, as well as by a range of established institutional clients. Trial access for CFRA’s ETF data and tools can be requested here.