AQR and Vanguard funds performance vs. stated benchmarks, updated 8/31/2020

[nisiprius]

On 11/9/2018 I reviewed all the AQR funds to see how they had performed relative to AQR's stated benchmarks. I've now updated this with data up to 8/31/2020.

Six of those funds no longer exist; I've retained the previous data from 2018. They are shown on the chart with light blue bars and asterisks next to their ticker symbols. AQR has changed the benchmarks used for a number of the funds since 2018. I am using their benchmark data, so I am using whatever benchmark they are using today, even if it's different from the one they were using in 2018.

38 out of 48 AQR funds underperformed their benchmark, or 79%.
The average underperformance was -1.46% per year.



Key
AQRIX — Risk Parity Fund (now renamed "AQR Multi-Asset Fund")
QRHIX — Risk Parity II HV Fund
QRMIX — Risk Parity II MV Fund
QCPIX — Core Plus Bond Fund
ARCIX — Risk-Balanced Commodities Strategy Fund
QICLX — International Multi-Style Fund
QIRIX — International Relaxed Constraint Equity Fund
AQIIX — International Equity Fund
ANDIX — International Defensive Style Fund
*QEMLX — Emerging Momentum Style Fund
*QERIX — Emerging Relaxed Constraint Equity Fund
AZEIX — Emerging Defensive Style Fund
QEELX — Emerging Multi-Style Fund
QTELX — TM Emerging Multi-Style Fund
*QEMLX — Emerging Momentum Style Fund
AZEIX — Emerging Defensive Style Fund
AQGIX — Global Equity Fund
QLEIX — Long/Short Equity Fund
AIMOX — International Momentum Style Fund
ATIMX — TM International Momentum Style Fund
AIMOX — International Momentum Style Fund
*ANDIX — International Defensive Style Fund
*QIMLX — TM International Multi-Style Fund
*QLRIX — Large Cap Relaxed Constraint Equity Fund
AMOMX — Large Cap Momentum Style Fund
AMOMX — Large Cap Momentum Style Fund
ATMOX — TM Large Cap Momentum Style Fund
ATMOX — TM Large Cap Momentum Style Fund
AUEIX — Large Cap Defensive Style Fund
AUEIX — Large Cap Defensive Style Fund
QCELX — Large Cap Multi-Style Fund
QTLLX — TM Large Cap Multi-Style Fund
*QSRIX — Small Cap Relaxed Constraint Equity Fund
ASMOX — Small Cap Momentum Style Fund
ATIMX — TM International Momentum Style Fund
ATSMX — TM Small Cap Momentum Style Fund
QSMLX — Small Cap Multi-Style Fund
QSSLX — TM Small Cap Multi-Style Fund
QLEIX — Long/Short Equity Fund
ADAIX — Diversified Arbitrage Fund
AQMIX — Managed Futures Strategy Fund
ASAIX — Multi-Strategy Alternative Fund
QGMIX — Global Macro Fund
QMHIX — Managed Futures Strategy HV Fund
QMNIX — Equity Market Neutral Fund
*QRPIX — Alternative Risk Premia Fund
QSLIX — Style Premia Alternative LV Fund
QSPIX — Style Premia Alternative Fund


Additional note: all performance data is from AQR's website; for example, for AQRIX,



Differences from the benchmark were calculated by the technically correct "geometric difference," (1 + actual return) / (1 + benchmark return) - 1. In this case, 1.0562 / 1.1016 - 1 = -4.12%. This is slightly different from the result of simply subtracting the benchmark return from the actual return.

[Random Musings]

Solid underperformance from the AQR shop. Better have some magical darts when picking those funds.

RM

[000]

Thanks, this is good stuff.

[physixfan]

This amount of stable underperformance is shocking... Wondering how these rich people who invested in these funds feel about this...

[Forester]

The exotic strategies have done worse than their stock funds. Lesson is, if you're going to tilt, buy a fund which is low cost, where one has some grasp of the methodology, avoid "black box" oddities.

[nisiprius]

And here is a similar chart for Vanguard's actively managed stock funds, somewhat marred by the fact that Vanguard is not showing benchmark performance since fund inception for six of the funds--possibly because the funds are probably older than the benchmark.

Of the 21 funds for which Vanguard provided benchmark-since-inception data, 12 of them, or 57.14% missed their benchmarks. However, the average difference from the benchmark was +0.18%, because the positive departures were bigger than the negative departures.

Because of data availability issues in both fund families, and weird stuff like AQR changing the name and strategy of their Risk Parity fund and Vanguard changing the name and strategy of their Precious Metals and Mining fund, I'm not ready to try to calculate significance, but certainly I get an overwhelming impression that whatever Vanguard has been doing in its non-indexed stock funds has been "mostly harmless," and that that has not been the case for AQR's fund family.

[Taylor Larimore]

Nisiprius:

You did a tremendous amount of work preparing and presenting the graphs showing the terrible underperformance of the AQR funds co-founded by Cliff Asness, one of the first designers of "smart-beta" and "factor" strategies. Thank you!

From the AQR website:

The ideas behind AQR were born in academia, and education has been in our DNA ever since. For us, learning never stops. We've published hundreds of journal articles, white papers, and thought leadership pieces in our pursuit of excellence, truth and clarity in investing.

Lesson learned: Academia cannot "beat the market."

Best wishes.
Taylor

Jack Bogle's Words of Wisdom: “Cliff Asness is a brilliant academic and the greatest marketer I’ve ever met.” -- "Absolutely no one knows what the stock market is going to do tomorrow, let alone next year. Nor which sector, style or region will lead and which will lag. Given this absolute uncertainty, the most logical strategy is to invest as broadly as possible. -- Total market indexing is the gold standard. Anything else, like sector investing, is a dilution of that standard."

[garlandwhizzer]

Great post, nisi, thanks for posting.

Those who believe that alpha can be generated on a consistent basis by academic backtesting derived strategies should take a hard look at these results. Cliff Asness is brilliant and has a vast wealth of academic credentials but what that bounty has produced the opposite of the desired results for investors. Backtesting results especially with complex expensive models often look a lot better in data mining the past than going forward.

Remember the sales pitch some years ago that certain "alternates" were expected to produce stock-like returns with bond-like risk? Actually at the time that's how it looked on backtesting. That was why those claims were made. The only problem was that the magic ceased shortly after their discovery and subsequent popularity. Plain old quality bonds turned out to be vastly better portfolio diversifiers. The financial industry is constantly seeking a better mousetrap and has promoted some rather expensive models to achieve it with very limited success. In some cases like RZV, the S&P 600 Small Cap Pure Value Fund which has maximal loadings to small and value factors, it has produced portfolio destroying tracking error since its inception more than 14 years ago. It theory it looked so appealing. In practice, a disaster. Results from Portfolio Visualizer:

https://www.portfoliovisualizer.com/bac ... ion2_2=100

I'm not saying that every investor is optimally served with TSM, although I do believe many are. It seems to me reasonable to me, however, based on real fund results to be a bit skeptical about claims of expected risk adjusted outperformance from new and improved investing products. The key word there is "expected." What's expected doesn't always happen especially if it comes with an expensive price tag like many of AQR's offerings.

Garland Whizzer

[cheezit]

https://www.portfoliovisualizer.com/bac ... ion2_2=100

Any particular reason you set the start date to 1993 when RZV was only created in 2006?

[Alpha4]

https://www.portfoliovisualizer.com/bac ... ion2_2=100

Any particular reason you set the start date to 1993 when RZV was only created in 2006?

Possibly because 1993 was the first full year VTSMX was in existence?

[nisiprius]

https://www.portfoliovisualizer.com/bac ... ion2_2=100

Any particular reason you set the start date to 1993 when RZV was only created in 2006?

It don't know why GarlandWhizzer didn't leave it set at the default--January, 1985, but it doesn't matter, since PortfolioVisualizer automatically constrains the date range to available data for the funds. In this case, PortfolioVisualizer tells us:

Portfolio Analysis Results (Apr 2006 - Aug 2020) [Link] [PDF] [Excel]
Note: The time period was constrained by the available data for Invesco S&P SmallCap 600 Pure Value ETF (RZV) [Apr 2006 - Aug 2020].

You can click on the link and try it for yourself. There the results are identical as long as the starting date is no later than April 2006.

[Alpha4]

Great post, nisi, thanks for posting.

Those who believe that alpha can be generated on a consistent basis by academic backtesting derived strategies should take a hard look at these results. Cliff Asness is brilliant and has a vast wealth of academic credentials but what that bounty has produced the opposite of the desired results for investors. Backtesting results especially with complex expensive models often look a lot better in data mining the past than going forward.

Remember the sales pitch some years ago that certain "alternates" were expected to produce stock-like returns with bond-like risk? Actually at the time that's how it looked on backtesting. That was why those claims were made. The only problem was that the magic ceased shortly after their discovery and subsequent popularity. Plain old quality bonds turned out to be vastly better portfolio diversifiers. The financial industry is constantly seeking a better mousetrap and has promoted some rather expensive models to achieve it with very limited success. In some cases like RZV, the S&P 600 Small Cap Pure Value Fund which has maximal loadings to small and value factors, it has produced portfolio destroying tracking error since its inception more than 14 years ago. It theory it looked so appealing. In practice, a disaster. Results from Portfolio Visualizer:

https://www.portfoliovisualizer.com/bac ... ion2_2=100

I'm not saying that every investor is optimally served with TSM, although I do believe many are. It seems to me reasonable to me, however, based on real fund results to be a bit skeptical about claims of expected risk adjusted outperformance from new and improved investing products. The key word there is "expected." What's expected doesn't always happen especially if it comes with an expensive price tag like many of AQR's offerings.

Garland Whizzer

Is it really fair, though, to compare RZV to a total stock market index fund? In other words, is a TSM fund (which is heavily dominated by large-cap growth stocks) really the proper benchmark for "tracking error" for a fund that was designed for maximal loading on the small cap and value factors? RZV appears to me to be an ETF with even high loadings to the smallcap and value factors than a "normal" SCV fund (like DFA Small Value or Vanguard Small Value Index); given that smallcap value has underperformed the market as a whole over the past several years--which has happened several times before....1929-31, 1937-39, 1946-49, 1951-53, 1969-73, 1986-87, 1989-90, mid 1994-1999 (with an especially yawning gap opening up between late 1997 and December 1999), spring of 2006 to early 2009, etc--wouldn't we expect to see some tracking error vs the TSM for any small value fund? From mid-2014 onwards (and especially from 2017 to the present) SCV has underperformed the market as a whole (which again, is quite to be expected at times) and it appears from the PV link you posted that RZV has tracked this; it was almost neck and neck with VTSMX at the end of 1Q 2014 and even at year-end 2016 it was pretty close to it; virtually all of the tracking error has come in the last 3.5 years or so.

Going by the following:

https://www.portfoliovisualizer.com/bac ... ion3_3=100

It appears that RZV is doing exactly what it is supposed to do; provide exposure to the small and value factors in addition to exposure to market beta overall; as the fund choices get higher loadings to small and value (i.e. from VISVX to DFSVX to RZV in that order) they appear to do well (or better) when SCV does well and poorly (or more poorly) when SCV doesn't do so well. Isn't that exactly what a small value fund would be expected to do? Isn't that exactly what it is supposed to do?

[MotoTrojan]

Great post, nisi, thanks for posting.

Those who believe that alpha can be generated on a consistent basis by academic backtesting derived strategies should take a hard look at these results. Cliff Asness is brilliant and has a vast wealth of academic credentials but what that bounty has produced the opposite of the desired results for investors. Backtesting results especially with complex expensive models often look a lot better in data mining the past than going forward.

Remember the sales pitch some years ago that certain "alternates" were expected to produce stock-like returns with bond-like risk? Actually at the time that's how it looked on backtesting. That was why those claims were made. The only problem was that the magic ceased shortly after their discovery and subsequent popularity. Plain old quality bonds turned out to be vastly better portfolio diversifiers. The financial industry is constantly seeking a better mousetrap and has promoted some rather expensive models to achieve it with very limited success. In some cases like RZV, the S&P 600 Small Cap Pure Value Fund which has maximal loadings to small and value factors, it has produced portfolio destroying tracking error since its inception more than 14 years ago. It theory it looked so appealing. In practice, a disaster. Results from Portfolio Visualizer:

https://www.portfoliovisualizer.com/bac ... ion2_2=100

I'm not saying that every investor is optimally served with TSM, although I do believe many are. It seems to me reasonable to me, however, based on real fund results to be a bit skeptical about claims of expected risk adjusted outperformance from new and improved investing products. The key word there is "expected." What's expected doesn't always happen especially if it comes with an expensive price tag like many of AQR's offerings.

Garland Whizzer

Is it really fair, though, to compare RZV to a total stock market index fund? In other words, is a TSM fund (which is heavily dominated by large-cap growth stocks) really the proper benchmark for "tracking error" for a fund that was designed for maximal loading on the small cap and value factors? RZV appears to me to be an ETF with even high loadings to the smallcap and value factors than a "normal" SCV fund (like DFA Small Value or Vanguard Small Value Index); given that smallcap value has underperformed the market as a whole over the past several years--which has happened several times before....1929-31, 1937-39, 1946-49, 1951-53, 1969-73, 1986-87, 1989-90, mid 1994-1999 (with an especially yawning gap opening up between late 1997 and December 1999), spring of 2006 to early 2009, etc--wouldn't we expect to see some tracking error vs the TSM for any small value fund? From mid-2014 onwards (and especially from 2017 to the present) SCV has underperformed the market as a whole (which again, is quite to be expected at times) and it appears from the PV link you posted that RZV has tracked this; it was almost neck and neck with VTSMX at the end of 1Q 2014 and even at year-end 2016 it was pretty close to it; virtually all of the tracking error has come in the last 3.5 years or so.

Going by the following:

https://www.portfoliovisualizer.com/bac ... ion3_3=100

It appears that RZV is doing exactly what it is supposed to do; provide exposure to the small and value factors in addition to exposure to market beta overall; as the fund choices get higher loadings to small and value (i.e. from VISVX to DFSVX to RZV in that order) they appear to do well (or better) when SCV does well and poorly (or more poorly) when SCV doesn't do so well. Isn't that exactly what a small value fund would be expected to do? Isn't that exactly what it is supposed to do?

Here is another way to think about RZV, by using Portfolio Visualizers Match Factor Exposure tool.

Let's say your desired asset allocation is 75% in the S&P500, and 25% in VBR (Vanguards CRSP Small-value Index fund). As your portfolio continues to grow, most of your contributions are going to a 401k that only has the S&P500 while you have an additional 15% in an IRA (not enough for that 25% VBR exposure; what are you to do?!?!).

As this shows, you can get similar factor exposure to 75/25 VFINX/VBR by using 88/12 VFINX/RZV; in this case, the RZV-based portfolio actually outperforms overall!

https://www.portfoliovisualizer.com/mat ... tion2_1=25

So yes, RZV did exactly what it was intended to do, it provided a ton of factor exposure in a small (and reasonably priced) package.

[nisiprius]

Just to be clear:

1) RZV is not an AQR fund.

2) If anyone is blaming the overall woes of AQR's whole fund family on recent underperformance by value, I would point out that not one AQR fund has the word "value" in its name, nor do I see anything on AQR's website suggesting that "value" is integral to most of their fund strategies. In fact they use names like "multi-style," "defensive style," and "momentum style" that suggest diversification and and an intention not to rely too heavily on the value factor.