An In-Depth Look at Fidelity's Contrafund

Below is the performance for the last 10 calendar years, 2005 through 2014.

The results are again similar – the Contrafund earned significantly higher returns combined with slightly higher volatility and a much higher Sharpe ratio. It doesn’t appear that FCNTX has been burdened by the increased level of assets under management that tend to come with success.

The table below shows performance for the latest five-year period, 2010 through 2014.

There’s no longer anything special about the fund’s performance over this period. While FCNTX did produce lower returns than the S&P 500 Index, it did so with lower volatility, which resulted in a virtually identical Sharpe ratio. FCNTX’s Sharpe ratio was just slightly higher.

While it’s hard to draw a conclusion from this data – five years isn’t a very long period – it’s possible the growth in assets under management has increased the burden on the fund to such a degree that it will no longer be able to generate superior performance.

Only time will reveal a conclusive answer. All that is clear now is that there has been no superior performance over the past five calendar years.

What are the sources of the fund’s returns?

I’ll now dive even deeper into the fund’s performance to try to identify its sources of returns. In other words, was the fund’s great long-term track record a result of its ability to identify individual stocks that would outperform? Was it instead from identifying the key factors (characteristics) that became the source of its returns? Or was it a combination of both? I’ll use the factor analysis tools provided by Portfolio Visualizer to analyze the Contrafund’s performance.

I begin by performing a three-factor analysis (beta, size and value) of the fund’s performance. Data is available starting in February 1980. The table below shows the results for the period from February 1980 through December 2014.

The three-factor analysis shows that FCNTX had a beta a bit lower than that of the market, had basically no exposure to either the size or value factors and generated an annual alpha of 1.24%, although that wasn’t statistically significant at the 5% level. The r-squared was relatively low at 0.78. You can get a low r-squared when portfolios are highly concentrated. However, that isn’t the case here because the fund holds more than 300 stocks. Another possible explanation for the r-squared figure is that the fund, although domestic, does hold a significant percentage of foreign stocks. Its current foreign holdings are equal to about 9% of the portfolio. Portfolio Visualizer uses a domestic model. There could be other explanations, such as additional factors, that might explain the fund’s returns in a better way.

Asset pricing models are not perfect. If they were, they would be called laws. That doesn’t mean they don’t have value. The way to think about them is as engines that help us understand how markets work, though they aren’t pictures or perfect representations of the world.

With this in mind, the next step is to examine the FCNTX’s performance through the lens of a four-factor model, adding momentum as the fourth factor.

Adding momentum as a fourth factor presents a somewhat different picture. While the beta, size and value loadings are basically unchanged, the fund did load significantly on the momentum factor. And, importantly, loading accounted for more than 100% of the alpha from the three-factor model. However, this negative alpha wasn’t quite statistically significant at the 5% level (a t-stat of 2 is required).

The model is, in effect, revealing that the outperformance of FCNTX is likely not explained by the ability of its manager to identify which individual stocks to overweight. Instead, it derives from the manager’s ability to identify the type of stocks to buy – stocks with positive momentum. While this explains the fund’s returns, it doesn’t detract in any way from its performance. Finally, while adding the momentum factor to the model does improve its explanatory power (the r-squared increased to 0.81) a higher correlation would create more confidence in the analysis.

I’ll now add the quality factor and see if that provides additional explanatory power.

 
Adding the fifth factor, quality, once again changes our view of the fund. Now, the loading on market beta has increased to almost one. There is small but significant loading on the size factor, statistically significant loading on both momentum and quality and alpha has become even more negative at -1.94, although it’s still not yet quite statistically significant at the 5% level. And finally, adding the quality factor didn’t improve the explanatory power of the model because the r-squared figure remained at 0.81.

There’s one more factor we can add, low beta.

 
The addition of the low beta factor has changed the landscape once again. The fund loaded significantly on both momentum and low beta and benefited from those loadings. In addition, negative alpha increase to -2.24% and approaches statistical significance. The 95% confidence level for the fund’s monthly alpha was between -0.40 and +0.03. While the r-squared isn’t as high as we would ideally like to see (it’s still at 0.81), it does appear that the fund’s excellent performance was a result of identifying the type of stocks to buy: quality stocks with positive momentum and low betas.

Recent performance

I’ll do one last round of analysis. I will examine the fund’s performance for the most recent 15-, 10-, and five-year periods through the lens of our six-factor model to try to detect any trends. I’ll begin with the last 15 calendar years.

The analysis shows that, while the fund outperformed the S&P 500 Index by 2.7 percentage points a year (6.9% versus 4.2%), it did not achieve alpha once its exposure to our six factors was taken into account. And the explanatory power of the model became much stronger. The r-squared rose to 0.91. At the 95% confidence level, its monthly alpha was between -0.19% and +0.19%.

I’ll now look at the last 10 calendar years.

The analysis for the last 10 years presents an interesting picture. First, the explanatory power of the model has improved over time. The r-squared rose to 0.95. The fund truly is actively managed. It moves from having basically no exposure to the value factor to a statistically significant and large negative exposure to it. The fund also moved from having a small positive loading on the size factor to a small negative one. In addition, the quality loading was both negative and statistically significant, while there was now no loading on the low beta factor over this period. And finally, for this period, the fund generated a large, statistically significant alpha of 2.25%.

That’s a very impressive performance. One possible explanation for the large positive alpha is that the fund’s managers may have been very good about identifying which factors would do well and then selecting stocks with exposures to those factors. They may not have been doing this explicitly (that is, they may or may not have been targeting those factors), but they could have been doing it indirectly via the stocks they select.

The last chart is for the most recent, five-year period.

The loadings over the last five years are similar to the loadings for the last 10 years. However, the alpha became slightly negative (though with no statistical significance). And the explanatory power of the model was high at 0.95.

What does the future hold for Contrafund investors?

Fidelity’s Contrafund certainly has a long and distinguished record of superior performance. With that said, the most likely explanation for the fund’s superior performance isn’t found in individual stock picking. Instead, the fund’s managers identified key factors driving performance (loading on momentum, quality and low beta) before the academics did.

And finally, it’s possible that the large increase in assets under management, a result of the fund’s superior performance, might have created too high a hurdle for the fund to overcome. Its outperformance has basically disappeared over the most recent five-year period. Only time will tell if it can re-achieve its prior level of outperformance. 

Larry Swedroe is director of research for the BAM Alliance, a community of more than 150 independent registered investment advisors throughout the country.

Read more articles by Larry Swedroe