BEATING THE #1 FUNDS

Fund Selection Using the "Moneyball" Approach

BEYOND FUND RANKINGS: CREATING A WINNING TEAM

We created a platform of funds that lets us turn any of our asset allocations plans into a portfolio of investments. Our portfolio construction methodology uses traditional diversification principles in a non-traditional way: we apply these risk-management ideas to the excess returns our funds generate. To our delight, we see that the degree of risk reduction we achieve with active returns is much greater than what we observe in the asset allocation process. While we can typically eliminate nearly a quarter of market volatility through diversification, we have observed a reduction of as much as three quarters of active volatility (i.e., “tracking error”) by applying the same risk management ideas to active returns. The result is a greater degree of confidence in the active returns we expect from our investment portfolios.

The key to this success is our “team approach,” which moves beyond the traditional method of focusing on an individual fund’s performance relative to its asset benchmark. That approach focuses on the pieces, rather than on the portfolio. We focus on the performance of each fund within the portfolio. This is the “Moneyball” approach of not “paying up” for Tier-1 “Star” assets in isolation, but rather selecting complementary assets which optimize performance through their interactions. As noted in our prior article (“Assembling a Team of Funds”) we examine the effects of the pairings of all funds on our platform, and we use these correlations of active returns as a key factor in our fund selection process. We also optimize our selection process to create an overall alpha stream that meets our client’s return target while keeping both total return volatility and active volatility within guidelines.

This raises the question: “Can this optimized portfolio outperform the group of funds that ranked highest in each category?” After all, those funds that demonstrated the highest ratio of excess return vs tracking error (i.e., “Information Ratio” or “IR”) are considered the “best” in each category. Can we really beat the “All Star” team by adding or even swapping some of them for “lesser” players? Won’t these second-tier or third-tier (or even-lower-tier) players drag the team down?

The answer is clearly and unequivocally “NO!”

Why? Because the “All Stars” are NOT a team; they are simply a naïve grouping of statistical winners in isolation. How they integrate as parts of a portfolio is completely unknown when they are selected. They represent individual performance, rather than team performance. Our goal is to create a portfolio, and this requires understanding how the pieces fit together to deliver the required market exposures along with an efficient active return. In this installment, we will demonstrate the “Moneyball” team approach, using the platform and asset allocation we established in our introductory articles on portfolio construction.

As to the timing of these events, we note that Harry Markowitz wrote his doctoral dissertation on portfolio theory in 1952, while the events around the Oakland Athletics baseball team took place during the 2002 season. We also note that Markowitz was awarded the Nobel prize for his contributions in 1990, while research into this approach to fund construction began in 2005, within the active portfolio management process

SELECTING FUNDS FOR A SPECIFIC ASSET ALLOCATION

We customize our fund team to the client’s asset allocation strategy. Traditional approaches to fund selection use a “one size fits all approach” that simply pro-rates a set of fund exposures to every asset allocation. Our approach selects both the funds and their weightings to optimize results for every asset allocation. After all, the fund weightings (from zero to maximum exposure) are the key decision variable in the portfolio construction process.

#1 FUNDS: A VERY HIGH HURDLE

We start by evaluating the performance of those funds that ranked #1 in their respective categories, relative to the remaining funds on the fund platform. We see clearly that the #1 funds outperformed the other funds on the platform by over 115 bps in for our asset allocation. About 1/3^rd of the platform’s funds delivered negative excess return, and they display a negative trend in the relationship between excess return and tracking error.

THE TOP THREE INFORMATION RATIO GROUPS

We created three “Information Ratio” portfolios, using our asset allocation and the funds in the top three information ratio tiers within each asset segment. The underlying segment groups had as many as eight funds (large growth) and as few as three (emerging markets equity.) Each portfolio employed a single fund in each asset segment.

Our initial view of the results focused on total return and volatility. The active process within the funds is the sole driver of these differences from the benchmark, with respect to return and risk. We see that the top two tiers of funds increased overall volatility risk very sightly, while adding between 110 and 175 bps of excess return. The third tier of funds increased return by only 35 bps while increasing volatility by almost 50 bps. We uncover a useful insight into the relationship between the active process and total portfolio volatility: active risk added to total portfolio risk.

Examining the active results tells a slightly different story, with the three portfolios producing increasing improvements in their results with relatively similar levels of active risk. The information ratios for these three portfolios were 0.24, 0.66 and 1.13 for portfolio rankings #3, #2 and #1 respectively. Clearly, the portfolio of #1-ranked IR funds is in a league of its own.

BEATING THE BEST: THREE GOALS

Our portfolios are designed to outperform the portfolio of top-rated funds. This requires us to establish the goals that determine what “winning” means. We selected these three goals, relative to the “All Star” portfolio:

• Highest alpha for the same tracking error
• Lowest tracking error for the same alpha
• Highest alpha at a 99% confidence level

We shifted our key return metric from excess return to alpha. As noted in our prior paper, we take a “risk aware” approach to investment management. This is rooted in the concept that “risk matters” when it comes to evaluating success. This implies that we never discuss return without including the risk that was taken to earn that return.

The industry’s use of excess return ignores risk; it is simply the difference in return between the portfolio and its benchmark. This is a substantial error that leads to inefficiency. A fund that underperforms its benchmark slightly while taking much lower risk is an outperformer, yet in terms of excess return (the traditional approach) it is considered an underperformer. The same is true for a fund that takes substantially greater risk than its benchmark while delivering only a small excess return; this fund is considered a “winner” and yet it lies below the capital market line that is fundamental to investment theory.

All returns used in this paper have been adjusted for volatility risk, so that additional return is truly “risk-adjusted.”

TWO VIEWS OF OUR RESULTS

We have seen the superiority of the #1-ranked IR funds, and this may create some skepticism regarding our ability to improve on their results by including funds that do not meet their individual performance standard. Yet in terms of both total return and active return measures, our “team of funds” portfolio outperformed the “All Star” portfolio. What caused this extraordinary result?

Our success demonstrates the value of alpha diversification and the focus on risk-adjusted performance measures. These are critical aspects of our team approach, along with our focus on the portfolio, rather than its pieces.

Our first chart shows that the team portfolios outperformed on a risk-adjusted, total return basis, delivering slightly lower total return with substantially lower volatility. The alpha diversification process lowered volatility, whereas the naïve process of picking funds by their individual characteristics added to portfolio volatility. The proportion of “non-#1 IR” funds in these portfolios ranged from 28% to 37 percent.

Our second chart focused on our main goal of producing superior active results. We see a higher “efficient frontier” of active returns, where the set of solutions was determined by the client’s active goals. All three solutions outperformed the All-Star portfolio on an absolute and on a risk-adjusted basis. Clearly, our “team dynamics” are the key to active efficiency!

DIVING DEEPER INTO SUCCESS: ESTABLISHING CONFIDENCE

The results from this performance evaluation demonstrate with certainty what happened over this 12-year period. But if we were to extrapolate our expectations into a forecast of future results, we would need to introduce a level of confidence into our projections. This is typically done with a statistical procedure that “haircuts” our forecast return, based on our performance timeframe and the level of return volatility. This is true for both total return and active returns. The typical test of confidence is at 95% but we conducted our tests at a 99% confidence level.

We see that the alpha of the portfolios of second-tier and third-tier ranked funds was completely wiped out in our high-confidence test. We would expect these funds to underperform the benchmark, when making any statement with a high level of conviction. However, the #1 IR funds delivered a respectable 57 bps of high-confidence alpha.

Here again, we confirm the value of our team approach. By adding the lower-tier funds to our mix, and guided by alpha diversification, we increased our high-confidence alpha expectation from 57 bps to as high as 91 bps. This “downside” view of expected active return is critical in managing client expectations. We can now state a reasonable alpha estimate of between 175-200 bps, while at a high level of confidence we expect between 80-90 bps of alpha. Essentially, our team approach, downside scenario retains half our expected alpha, where the #1-IR funds retain less than a third.

RELATIONSHIP BETWEEN ALPHA AND TOTAL VOLATILITY

Shouldn’t your portfolio manager know whether alpha contributes to portfolio volatility – and if so, how much? And, was the tradeoff between return and higher volatility worth it?

Most managers do not explore this characteristic, and the performance measurement process remains silent in terms of risk attribution. Fortunately, we address this critical issue, because we believe that clients deserve to know the sources of both return and risk.

When we reviewed the traditional portfolios (with their emphasis on funds in isolation) we found that their excess returns contributed to portfolio volatility. Added return came with a cost of higher volatility. But in all three of our team strategies, the active return subtracted between 30bps to almost 60 bps of volatility. We demonstrated in our prior articles that this result is driven by a negative correlation between the excess returns and the portfolio’s total return. Any tracking error in these strategies is subtracted from total return volatility, as the product of the tracking error multiplied by the correlation.

The team approach produced the most efficient active portfolio, where return was increased and volatility was decreased.

COMBINING RISK-ADJUSTED PERFORMANCE MEASURES

We take an innovative approach to reviewing the traditional performance measures: we combine them to provide additional insights into the active process and its results. Our first example evaluates the relationship between active efficiency and total return volatility. Our chart of information ratio relative to volatility demonstrates that the team approach accomplished two beneficial goals:

• subtracted total return volatility
• simultaneously provided a higher active reward.

Our most innovative illustration brings together two risk-adjusted return measures to demonstrate the simultaneous success across both views of performance: reward for active risk vs reward for market risk. While the “in isolation” portfolios (based on IR rankings) do increase active efficiency as they are rewarded for market risk, the “team” portfolios dominate them.

SUMMARY: OUR HOLISTIC APPROACH TO PORTFOLIO CONSTRUCTION

Our portfolio construction process began by introducing a due diligence program that created a platform of approved investments which receive ongoing monitoring, allowing portfolio managers to assemble portfolios of investments that deliver the firm’s recommended asset allocation, along with the hope of higher returns (after adjusting for risk.) We confirmed the value of the fund platform by evaluating its results when applied to a typical investment strategy. Simply “showing up” to the platform provided a small-but-meaningful enhancement to return.

Portfolio construction turns an asset allocation plan into a set of investments that hope to deliver total returns that allow clients to meet their financial goals. We introduced the innovation of a “team approach” to selecting funds, where the complementary nature of these active funds produced superior results. Our “acid test” of the effectiveness of our approach was to see whether our alpha diversification could produce higher risk-adjusted active benefits than the “All Star” team of the highest-ranked funds on the platform.

To our delight, we found strong confirmation of the wisdom of our approach, as our fund team outperformed the naïve “All Star” team across every performance measure, delivering high-confidence results to meet our client’s return objectives. We applied traditional, time-tested diversification concepts where they work best: idiosyncratic, with active returns. This sets a new standard for portfolio construction.

OUR NEXT STEP IN PORTFOLIO CONSTRUCTION

Our next article introduces the concept of tailoring the fund team to the client’s alpha target.

The traditional approach to fund selection focuses on the desirability of each fund in isolation. This implies that the overall active result is unspecified – we simply take the active return these funds provide. All clients receive the same set of funds, and their alpha and tracking error are fixed, regardless of their individual goals and tolerance for risk.

We believe that delivering the appropriate active return and risk is a critical decision that deserves as much attention and rigor as the other investment decisions. Bringing this aspect of portfolio construction into our holistic approach ensures that our clients receive cohesive and consistent guidance that remains focused on achieving their financial goals within a framework of their individual perspectives.

.

.

.

Written in partnership with Stephen Campisi