Thursday, April 9, 2015

A Guide to Creating Your Own Smart Beta Fund

FT tries to define smart beta:

Smart beta is a rather elusive term in modern finance. It lacks a strict definition and is also sometimes known as advanced beta, alternative beta or strategy indices. It can be understood as an umbrella term for rules based investment strategies that do not use the conventional market capitalisation weights that have been criticised for delivering sub-optimal returns by overweighting overvalued stocks and, conversely, underweighting undervalued ones.
A pretty good definition, which is just broad enough to allow for anything and say it's a cure for almost anything. As a result, there has been a proliferation of smart beta ETFs and mutual funds that have, in many instances, gathered an absurd amount of assets from investors, often times with only backtested performance.

While no concept will work over every time frame, the more concerning aspect to me is that interesting concepts that backtested well in one category are being applied to other areas of the market irrespective of supporting analysis and/or whether there is proper liquidity. One highly regarded "fundamental" index provider now has a fundamental index not only within all major equity categories, but also across multiple areas of fixed income, including US high yield corporate bonds. While the 1.07% annualized underperformance of the "smart" index vs the Barclays US High Yield Index over the last 3 years is in itself an issue, the real concern is the incremental 1.10% of annualized underperformance the ETF has to its custom index due to fees + trading in an illiquid part of the credit market. This almost 2.2% annualized underperformance hasn't stopped this almost $700 million fund from taking in another $86mm in assets the first three months of the year!

So... lots of flows, based in large part on back-tested results, with actual results of only limited value. Let’s create our own…

Step 1: Find a Marketable Idea

Leveraging the work from my previous post Is there a Relationship Between the Economy and Stock Market?, let's create a systematic asset allocation index* and call it GDP Growth "Smart Beta". To sell a product, you need a concept that is unique, yet resonates with the masses:

I don't know of any ETFs that are systematically tied to short-term GDP growth as the main factor driving the stock / bond allocation decision

Despite the fact that economists (even good economists) have no idea what's going on, everyone thinks they are an economist (including me... case in point - I started a blog called EconomPic Data)

Step 2: Create a Pitch
Using an exclusive signal, the strategy shifts capital to assets capturing the equity risk premium when deemed attractive and de-risks when valuations are deemed poor, while a proprietary macro overlay identifies periods of relative economic weakness. The resulting portfolio provides equity-like returns during strong market environments, with bond-like characteristics during periods of heightened market uncertainty.

Step 3: Show Backtested Results

Show performance in the best format possible. If total returns are similar, but a time series highlights long periods of underperformance vs the traditional asset class... then only show the time series of the "smart beta" index (and show a total returns comparison vs. the traditional beta separately in a table).

Step 4: Add Leverage

Show how much better returns would have been at similar levels of risk as the stock market, while ignoring the fact that there is no way to know in advance how much leverage you would need to have matched that standard deviation.

Step 5: Profit

Actually, I missed a step. I need a good ETF ticker. Perhaps Goodrich Petroleum would be willing to sell theirs?


Despite what I outline above, I think there is a huge opportunity for systematic / rules based ETFs. That said, I have a hard time understanding why many of these products, sponsors, and providers get a free pass (or in many cases are put on a pedestal) from the same guys that are railing against active management. Also... just because something has worked well in the past and there is academic support that explains (or tries to explain) why it has worked, it doesn't mean it will work in all markets or across all asset classes.

*Each quarter, if the latest one year nominal GDP > the yield to worst on the Barclays US Aggregate Bond Index... allocate the next quarter to stocks (S&P 500); otherwise to bonds (Barclays Aggregate Index). 


  1. Bethlehem Steel got rid of their old ticker symbol. You can use that now if you can't get GDP.

  2. Is the GDP value for the backtest using point-in-time nominal GDP or the revised nominal GDP? If you have a link to point-in-time GDP data please share!


    1. I've actually been looking for that time series myself. Lagged it (so for example, using Q4 2014 GDP data for the 3/31/15 allocation), but obvious issues in the backtest when revisions occur.

    2. Though some of this issue is cushioned by the fact that one year (smoothed) GDP and the numbers are rarely close. Only 14 "switches" in 38 years.

  3. And finally... the results are bullshit. So very similar to most smart beta backtests.