AQR Style Premia Update; Certainly Low Correlations

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

The AQR Style Premia Fund is sure zigging when the equity market is zagging. Thought I'd update its performance. Long-short across 4 styles and 4 asset classes.

Returns in %

VTSAX DFMPX QSPRX

1 month 0.1 0 5.5
3 months 3.9 -0.8 18.1
YTD 10.2 -0.8 25.5
6 months 17.5 -0.6 25.7
1 year 45.5 0.5 14.3
2 years 45.5 2.2 -8.7
5 years 105 1.6 -20.8

VTSAX =VG Total Stock Market Admiral
DFMPX=DFA Int Term Muni
QSPRX=AQR Style Premia
000
Posts: 8211
Joined: Thu Jul 23, 2020 12:04 am

Re: AQR Style Premia Update; Certainly Low Correlations

Post by 000 »

"zigging when the equity market is zagging" isn't the objective of the fund though.

From the prospectus (emphasis mine):
Investment Objective

The AQR Style Premia Alternative Fund (the “Fund”) seeks positive absolute returns. As further described under “Details About the AQR Style Premia Alternative Fund” in the Fund’s prospectus, a “positive absolute return” seeks to earn a positive total return over a reasonable period of time regardless of market conditions or general market direction.
Is five years a reasonable period of time?
MishkaWorries
Posts: 1363
Joined: Wed Aug 14, 2019 4:39 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by MishkaWorries »

1.87 ER!?!

The fund has returned 1.73 percent over the past year, -9.23 percent over the past three years, and -3.31 percent over the past five years.
We plan. G-d laughs.
Jebediah
Posts: 805
Joined: Tue Aug 28, 2012 9:19 pm
Location: Austin TX

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Jebediah »

Random Walker-

The uncorrelation is the easy part. It would be interesting to know how you personally feel about QSPIX's 1% return, 8% stdev, and -40% drawdown since inception. Is it fulfilling its promise of equity-like returns with bond-like volatility? In particular, what do you make of the fact that the management fees are greater than the annual return? Is that, in your estimation, a good investment?
gtwhitegold
Posts: 673
Joined: Fri Sep 21, 2012 1:55 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by gtwhitegold »

I'm good with it. I actually invested more in it back in December, so I'm in the black overall. I plan on sticking with it until I have a very good reason not to. Explainable poor performance isn't a very good reason for me to stop investing in it.
Alchemist
Posts: 638
Joined: Sat Aug 30, 2014 6:35 am

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Alchemist »

You could have just held cash in a HYSA and done far better. Even in a brokerage account a STT fund would have stomped QSPRX.

Also don't forget that in addition to the huge ER of the fund, you need to pay an advisor for the privilege of having access to it :oops:

So the middlemen are getting paid at least.
Angst
Posts: 2968
Joined: Sat Jun 09, 2007 11:31 am

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Angst »

Here you go RW. Note the 5yr VTSAX of 105. Is that correct?

Code: Select all

                VTSAX		DFMPX		QSPRX
					
1 month		  0.1		 0		  5.5
3 months	  3.9		-0.8		 18.1
YTD	         10.2		-0.8		 25.5
6 months	 17.5		-0.6		 25.7
1 year		 45.5		 0.5		 14.3
2 years		 45.5		 2.2		 -8.7
5 years		105		 1.6		-20.8
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

Angst wrote: Thu May 13, 2021 5:55 am Here you go RW. Note the 5yr VTSAX of 105. Is that correct?

Code: Select all

                VTSAX		DFMPX		QSPRX
					
1 month		  0.1		 0		  5.5
3 months	  3.9		-0.8		 18.1
YTD	         10.2		-0.8		 25.5
6 months	 17.5		-0.6		 25.7
1 year		 45.5		 0.5		 14.3
2 years		 45.5		 2.2		 -8.7
5 years		105		 1.6		-20.8
Yes same numbers I got. You were able to format better than me
Dave
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

Alchemist wrote: Thu May 13, 2021 4:48 am You could have just held cash in a HYSA and done far better. Even in a brokerage account a STT fund would have stomped QSPRX.

Also don't forget that in addition to the huge ER of the fund, you need to pay an advisor for the privilege of having access to it :oops:

So the middlemen are getting paid at least.
Cant judge a strategy by the outcome. Alternative histories could play out. Each of the styles does have an expected return above cash. Certainly didn’t work out over last several years, but maybe this year we are starting to see some benefit?

Dave
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

Jebediah wrote: Wed May 12, 2021 10:22 pm Random Walker-

The uncorrelation is the easy part. It would be interesting to know how you personally feel about QSPIX's 1% return, 8% stdev, and -40% drawdown since inception. Is it fulfilling its promise of equity-like returns with bond-like volatility? In particular, what do you make of the fact that the management fees are greater than the annual return? Is that, in your estimation, a good investment?
So far, no not good investment. Looking for after tax returns above bonds. But maybe the benefit is starting to appear now. 5 years feels like forever, but it’s really too short a period to evaluate.

Dave
Alchemist
Posts: 638
Joined: Sat Aug 30, 2014 6:35 am

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Alchemist »

Random Walker wrote: Thu May 13, 2021 7:48 am Cant judge a strategy by the outcome.
Sounds like something an advisor would say to excuse bad advice.

The *only* way to judge a strategy is by its outcome.
User avatar
typical.investor
Posts: 5263
Joined: Mon Jun 11, 2018 3:17 am

Re: AQR Style Premia Update; Certainly Low Correlations

Post by typical.investor »

Random Walker wrote: Thu May 13, 2021 7:48 am
Alchemist wrote: Thu May 13, 2021 4:48 am You could have just held cash in a HYSA and done far better. Even in a brokerage account a STT fund would have stomped QSPRX.

Also don't forget that in addition to the huge ER of the fund, you need to pay an advisor for the privilege of having access to it :oops:

So the middlemen are getting paid at least.
Cant judge a strategy by the outcome. Alternative histories could play out. Each of the styles does have an expected return above cash. Certainly didn’t work out over last several years, but maybe this year we are starting to see some benefit?

Dave
I agree with Dave. My high ER leveraged muni CEF fund has walloped QSPRX over it’s life but as we now go into a little more inflation ... things will likely change.

Anyway, selling value type investments after underperformance just isn’t good strategy. Not sure why momentum hasn’t helped it more... but maybe shorting growth has just been that bad.
User avatar
typical.investor
Posts: 5263
Joined: Mon Jun 11, 2018 3:17 am

Re: AQR Style Premia Update; Certainly Low Correlations

Post by typical.investor »

Random Walker wrote: Thu May 13, 2021 7:48 am
Alchemist wrote: Thu May 13, 2021 4:48 am You could have just held cash in a HYSA and done far better. Even in a brokerage account a STT fund would have stomped QSPRX.

Also don't forget that in addition to the huge ER of the fund, you need to pay an advisor for the privilege of having access to it :oops:

So the middlemen are getting paid at least.
Cant judge a strategy by the outcome. Alternative histories could play out. Each of the styles does have an expected return above cash. Certainly didn’t work out over last several years, but maybe this year we are starting to see some benefit?

Dave
I agree with Dave. My high ER leveraged muni CEF fund has walloped QSPRX over it’s life but as we now go into a little more inflation ... things will likely change.

Anyway, selling value type investments after underperformance just isn’t good strategy. Not sure why momentum hasn’t helped it more... but maybe shorting growth has just been that bad.
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

Alchemist wrote: Thu May 13, 2021 8:00 am
Random Walker wrote: Thu May 13, 2021 7:48 am Cant judge a strategy by the outcome.
Sounds like something an advisor would say to excuse bad advice.

The *only* way to judge a strategy is by its outcome.
Are you sure? What if you had entered the stock market in 2000 and evaluated your results in 2012? Would have given up on the market factor forever? The S&P500 underperformed TBills for three long stretches: 1929-43, 1966-82, 2000-12. It’s really really hard to know when to give up on a strategy. I’ll agree with you that the threshold for throwing in the towel is probably way lower when expenses are high!

Dave
bgf
Posts: 2085
Joined: Fri Nov 10, 2017 8:35 am

Re: AQR Style Premia Update; Certainly Low Correlations

Post by bgf »

Alchemist wrote: Thu May 13, 2021 8:00 am
Random Walker wrote: Thu May 13, 2021 7:48 am Cant judge a strategy by the outcome.
Sounds like something an advisor would say to excuse bad advice.

The *only* way to judge a strategy is by its outcome.
You are 100% wrong, and I will die on this hill.
“TE OCCIDERE POSSUNT SED TE EDERE NON POSSUNT NEFAS EST"
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

bgf wrote: Thu May 13, 2021 8:53 am
Alchemist wrote: Thu May 13, 2021 8:00 am
Random Walker wrote: Thu May 13, 2021 7:48 am Cant judge a strategy by the outcome.
Sounds like something an advisor would say to excuse bad advice.

The *only* way to judge a strategy is by its outcome.
You are 100% wrong, and I will die on this hill.
And conversely, is a strategy necessarily good just because the outcome is good? Would it have been an excellent strategy to put all my money in Bitcoin or single stock Tesla a decade ago? In retrospect, sure wish I did. Sometimes it’s better to be lucky than good. But I would still argue that those highly successful strategies were still poor strategies.

Dave
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

I was looking for Cliff Asness’ Gut Punch article where I think he explains an initial whipsaw effect where momentum gets hurt while value makes a turn around. On my way to Gut Punch I found this article.

https://www.institutionalinvestor.com/a ... g-Comeback

Dave
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

typical.investor wrote: Thu May 13, 2021 8:06 am Anyway, selling value type investments after underperformance just isn’t good strategy. Not sure why momentum hasn’t helped it more... but maybe shorting growth has just been that bad.
Here’s Cliff’s Gut Punch essay. He describes how the bad streak is due first to value. During this period momentum helped but not enough. Second, when value started to do well, it wasn’t enough to overcome a poor stretch for momentum. I think he’s saying the rapidity of the value turn around is what hurt momentum. He hopes that if value starts to make an extended positive run, then momentum will likewise go in a positive direction. Strongly recommend this essay. Actually, I recommend all of Cliff Asness’ essays. His style is great. Educational, clear, and funny. The footnotes are must read material.

https://www.aqr.com/Insights/Perspectives/A-Gut-Punch

Dave
marcopolo
Posts: 8446
Joined: Sat Dec 03, 2016 9:22 am

Re: AQR Style Premia Update; Certainly Low Correlations

Post by marcopolo »

Random Walker wrote: Thu May 13, 2021 7:48 am
Alchemist wrote: Thu May 13, 2021 4:48 am You could have just held cash in a HYSA and done far better. Even in a brokerage account a STT fund would have stomped QSPRX.

Also don't forget that in addition to the huge ER of the fund, you need to pay an advisor for the privilege of having access to it :oops:

So the middlemen are getting paid at least.
Cant judge a strategy by the outcome. Alternative histories could play out. Each of the styles does have an expected return above cash. Certainly didn’t work out over last several years, but maybe this year we are starting to see some benefit?

Dave
So, what is the point of this post?
I have noticed several times when you have posted how well the strategy is "working", by showing the outcome of some very short period of time, then when people point out how poor it has worked over a longer period, the refrain is "outcome is not as important as the strategy".

So, if I understand it, outcome does not matter over five years when it contradicts your narrative, but it is worth celebrating over a couple months, if it supports your narrative?
Once in a while you get shown the light, in the strangest of places if you look at it right.
gtwhitegold
Posts: 673
Joined: Fri Sep 21, 2012 1:55 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by gtwhitegold »

I don't pay an advisor since I invested through Fidelity when it was still available through them. My returns are positive overall, partly because I made a considerable investment near the bottom in December.

I still feel that the argument towards a fund like this is valid and strong. If it wasn't, then I wouldn't have invested in it in the first place. Since some of the more vocal participants on this forum are adverse to factor investing in general, this fund is definitely not for them.

If you don't have confidence that a strategy will work then don't invest in that fund. A lot of posts about this fund look more like attacks on the investors than a reasonable discussion.
gtwhitegold
Posts: 673
Joined: Fri Sep 21, 2012 1:55 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by gtwhitegold »

marcopolo wrote: Thu May 13, 2021 11:46 am
Random Walker wrote: Thu May 13, 2021 7:48 am
Alchemist wrote: Thu May 13, 2021 4:48 am You could have just held cash in a HYSA and done far better. Even in a brokerage account a STT fund would have stomped QSPRX.

Also don't forget that in addition to the huge ER of the fund, you need to pay an advisor for the privilege of having access to it :oops:

So the middlemen are getting paid at least.
Cant judge a strategy by the outcome. Alternative histories could play out. Each of the styles does have an expected return above cash. Certainly didn’t work out over last several years, but maybe this year we are starting to see some benefit?

Dave
So, what is the point of this post?
I have noticed several times when you have posted how well the strategy is "working", by showing the outcome of some very short period of time, then when people point out how poor it has worked over a longer period, the refrain is "outcome is not as important as the strategy".

So, if I understand it, outcome does not matter over five years when it contradicts your narrative, but it is worth celebrating over a couple months, if it supports your narrative?
You're digging too far into this. Saying that the fund may have turned the corner is not the same as saying that the trend will continue.

You also need to understand what is going on and if it makes sense even if you are just investing in Total Stock Market, Total International Stock Market, and Total Bond Market. Quite frequently we get questions on this forum about why a fund lost value after a dividend distribution or someone giving up on International or bonds because of the recent performance of US stocks.

If you don't understand the Style Premia fund, then don't invest in it.
marcopolo
Posts: 8446
Joined: Sat Dec 03, 2016 9:22 am

Re: AQR Style Premia Update; Certainly Low Correlations

Post by marcopolo »

gtwhitegold wrote: Thu May 13, 2021 12:22 pm
marcopolo wrote: Thu May 13, 2021 11:46 am
Random Walker wrote: Thu May 13, 2021 7:48 am
Alchemist wrote: Thu May 13, 2021 4:48 am You could have just held cash in a HYSA and done far better. Even in a brokerage account a STT fund would have stomped QSPRX.

Also don't forget that in addition to the huge ER of the fund, you need to pay an advisor for the privilege of having access to it :oops:

So the middlemen are getting paid at least.
Cant judge a strategy by the outcome. Alternative histories could play out. Each of the styles does have an expected return above cash. Certainly didn’t work out over last several years, but maybe this year we are starting to see some benefit?

Dave
So, what is the point of this post?
I have noticed several times when you have posted how well the strategy is "working", by showing the outcome of some very short period of time, then when people point out how poor it has worked over a longer period, the refrain is "outcome is not as important as the strategy".

So, if I understand it, outcome does not matter over five years when it contradicts your narrative, but it is worth celebrating over a couple months, if it supports your narrative?
You're digging too far into this. Saying that the fund may have turned the corner is not the same as saying that the trend will continue.

You also need to understand what is going on and if it makes sense even if you are just investing in Total Stock Market, Total International Stock Market, and Total Bond Market. Quite frequently we get questions on this forum about why a fund lost value after a dividend distribution or someone giving up on International or bonds because of the recent performance of US stocks.

If you don't understand the Style Premia fund, then don't invest in it.
I don't think i asserted that there was any claim of the trend continuing.

Not sure what i need "understand what is going on" . Perhaps you could enlighten us as to what you think is "going on" if you understand it.

Not sure what any of this has to do with people's confusion about dividend distributions?!?

So, do outcomes matter or not? If they do, over what period of time do they matter? If they don't matter at all, then what other criteria does one use to evaluate whether a strategy continues to be a good one?

It seems odd to simultaneously claim that outcomes don't matter, but then post whenever the outcome starts to look favorable.
Once in a while you get shown the light, in the strangest of places if you look at it right.
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

marcopolo wrote: Thu May 13, 2021 11:46 am
Random Walker wrote: Thu May 13, 2021 7:48 am
Alchemist wrote: Thu May 13, 2021 4:48 am You could have just held cash in a HYSA and done far better. Even in a brokerage account a STT fund would have stomped QSPRX.

Also don't forget that in addition to the huge ER of the fund, you need to pay an advisor for the privilege of having access to it :oops:

So the middlemen are getting paid at least.
Cant judge a strategy by the outcome. Alternative histories could play out. Each of the styles does have an expected return above cash. Certainly didn’t work out over last several years, but maybe this year we are starting to see some benefit?

Dave
So, what is the point of this post?
I have noticed several times when you have posted how well the strategy is "working", by showing the outcome of some very short period of time, then when people point out how poor it has worked over a longer period, the refrain is "outcome is not as important as the strategy".

So, if I understand it, outcome does not matter over five years when it contradicts your narrative, but it is worth celebrating over a couple months, if it supports your narrative?
Outcome matters a lot. But you don’t know outcomes when you choose a strategy ex ante. I haven’t posted that the strategy is working well. Just pointing out what the strategy is doing. Let people draw their own conclusions from the returns over different time frames relative to equities and bonds. Discussion about QSPIX started here if I remember a bit more than halfway through a huge bull market for equities in general and large growth in particular. Since then we had continued large Groth equity run, the big Covid hit, increasing valuation changes between growth and value, and possible start of value comeback. Thought it would be interesting to take another look at QSPIX performance over different timeframes now that the market has taken some twists, turns, tumbles, and one of many potential alternative histories has played out.

Dave
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

marcopolo wrote: Thu May 13, 2021 12:56 pm It seems odd to simultaneously claim that outcomes don't matter, but then post whenever the outcome starts to look favorable.
Well, yes I would be guilty at least part of that :-). Obviously outcomes do matter a lot. And what really matters is the outcome of the portfolio as a whole in meeting the investor’ personal objectives. Since QSPIX has overall underperformed expectations and since it has been beaten up around here pretty substantially, I’ve made the effort to show results at times where it has had a positive effect on a portfolio dominated by equities and bonds. It is true that we shouldn’t confuse strategy with outcome. And outcome depends on different start and end dates for all of us.

In a recent video conference including David Booth of DFA (I think there is a thread about it here on this site) someone asks him how long is long enough to make a judgement on a strategy and decide to throw in the towel. His response was something to the effect “1 year longer than you’re willing to give me” :-)

Dave
User avatar
nisiprius
Advisory Board
Posts: 52217
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: AQR Style Premia Update; Certainly Low Correlations

Post by nisiprius »

Low correlation in itself isn't automatically valuable.

Low correlation + similar volatility + similar return = big win
Low correlation + similar volatility + low return = ?????

To riff on Bogle's phrase, another application of the "relentless rules of humble arithmetic" is that the return of a portfolio is the weighted average of the returns of its parts. It's as simple as that. Nothing about correlations or volatility can change it. It follows that if you have a portfolio P and a proposed diversifier D, if D has lower return than P, then adding D to the portfolio will reduce its return.

So that's bedrock. As your own numbers show, QSPRX has lost money over the past 2, 3, 5 years and since inception. In order to give the strongest effect from the recent gains, I'll use the last two years as a time period.

If a diversifier has lost money, can it still improve a portfolio through low correlation? The answer is "yes-but." You have to define "improve" as "risk-adjusted return." The diversifier is going to cut return, but it is theoretically possible that it might cut volatility through negative correlation, and if the stars align it might cut volatility so much that it would improve the risk-adjusted return of the portfolio. It might even be possible to take advantage of the lowered risk to modify the portfolio by increasing the stock allocation and take the improvement in the form of same risk, higher return.

But QSPRX didn't have a negative correlation with stocks over the last two years, it had a low positive correlation, ρ = +0.13. Low. Not negative.

Consider a portfolio of 60% Total Stock, 40% Total Bond (blue), and suppose we reduce the risk by adding a 20% allocation to QSPRX, i.e. 48% Total Stock, 32% Total Bond, 20% QSPRX.

Source

Image

So, yes, QSPRX would have reduced risk as measured by StDev. But not by a lot. And of course (having lost money) it also would have reduced return. And it would have lowered the Sharpe and Sortino ratios, meaning that risk was not lowered enough to justify the lower return.''

The appropriate way to evaluate exotic additions to a basic portfolio is not compare them to doing nothing (i.e. to the original portfolio). The appropriate way is to compare them to the effect of reducing risk in the simple way: increasing the bond allocation of the original portfolio.

So let's compare the effect of adding a 20% allocation to QSPRX the simple way to reduce risk: changing the allocation of the original portfolio to 40/60.

Source

Image

As you see, a 40/60 portfolio combined higher return with lower standard deviation... compared to the results of adding a sophisticated long-short liquid alt confection with a 1.71% expense ratio.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
afan
Posts: 8195
Joined: Sun Jul 25, 2010 4:01 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by afan »

To say nothing of the fees paid to an adviser who put one into this high price, low performance, actively managed fund.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama
antman50
Posts: 53
Joined: Wed Aug 12, 2020 11:54 am

Re: AQR Style Premia Update; Certainly Low Correlations

Post by antman50 »

nisiprius wrote: Thu May 13, 2021 2:52 pm Low correlation in itself isn't automatically valuable.

Low correlation + similar volatility + similar return = big win
Low correlation + similar volatility + low return = ?????

To riff on Bogle's phrase, another application of the "relentless rules of humble arithmetic" is that the return of a portfolio is the weighted average of the returns of its parts. It's as simple as that. Nothing about correlations or volatility can change it. It follows that if you have a portfolio P and a proposed diversifier D, if D has lower return than P, then adding D to the portfolio will reduce its return.

So that's bedrock. As your own numbers show, QSPRX has lost money over the past 2, 3, 5 years and since inception. In order to give the strongest effect from the recent gains, I'll use the last two years as a time period.

If a diversifier has lost money, can it still improve a portfolio through low correlation? The answer is "yes-but." You have to define "improve" as "risk-adjusted return." The diversifier is going to cut return, but it is theoretically possible that it might cut volatility through negative correlation, and if the stars align it might cut volatility so much that it would improve the risk-adjusted return of the portfolio. It might even be possible to take advantage of the lowered risk to modify the portfolio by increasing the stock allocation and take the improvement in the form of same risk, higher return.

But QSPRX didn't have a negative correlation with stocks over the last two years, it had a low positive correlation, ρ = +0.13. Low. Not negative.

Consider a portfolio of 60% Total Stock, 40% Total Bond (blue), and suppose we reduce the risk by adding a 20% allocation to QSPRX, i.e. 48% Total Stock, 32% Total Bond, 20% QSPRX.

Source

Image

So, yes, QSPRX would have reduced risk as measured by StDev. But not by a lot. And of course (having lost money) it also would have reduced return. And it would have lowered the Sharpe and Sortino ratios, meaning that risk was not lowered enough to justify the lower return.''

The appropriate way to evaluate exotic additions to a basic portfolio is not compare them to doing nothing (i.e. to the original portfolio). The appropriate way is to compare them to the effect of reducing risk in the simple way: increasing the bond allocation of the original portfolio.

So let's compare the effect of adding a 20% allocation to QSPRX the simple way to reduce risk: changing the allocation of the original portfolio to 40/60.

Source

Image

As you see, a 40/60 portfolio combined higher return with lower standard deviation... compared to the results of adding a sophisticated long-short liquid alt confection with a 1.71% expense ratio.
not so sure either of those are fair comps. Of course, if you pull from stocks, you're going to get a lower return considering you are pulling from the highest returning thing over that time period. If you pull from bonds (which have had a great stretch) and switch that allocation with qspix (which has been horrific as a standalone), you'll find that the portfolios look near identical. ex.... 60/40 vs. 60/20/20 is much more fair.
afan
Posts: 8195
Joined: Sun Jul 25, 2010 4:01 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by afan »

Many bad strategies have bad outcomes.

Active management is a bad strategy, marked by poor outcomes.

If one is paying 1% of assets, or whatever, for the advice to get into these funds, this just makes a bad strategy worse.

Paying attention to marketing claims for funds is another bad strategy.

The markets are remarkably efficient. There is abundant evidence for this statement, familiar to bogleheads. Claims that one has found a loophole to market efficiency are easy to come by. Convincing evidence that these claims are true is rare

For a fund, such as this one, that changes allocations based on the decisions of an active manager, it is impossible to specify what to expect in composition, since the approach will change.

It is a simple pitch for the investing skill of an active manager, dressed up with some claims about the unique approach.

Plus nosebleed fees.

Plus nosebleed fees for an adviser.

Plus, if you swing that way, fees for an adviser to help you pick an adviser.

A sad way to see people waste their money. But a free country, do what you please.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama
CuriousTacos
Posts: 690
Joined: Thu Apr 12, 2018 3:31 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by CuriousTacos »

nisiprius wrote: Thu May 13, 2021 2:52 pm the return of a portfolio is the weighted average of the returns of its parts. It's as simple as that. Nothing about correlations or volatility can change it. It follows that if you have a portfolio P and a proposed diversifier D, if D has lower return than P, then adding D to the portfolio will reduce its return.
I don't necessarily disagree with your conclusions, but in the above, I think you are referring to arithmetic means rather than geometric means (aka CAGR). It is possible to increase the CAGR of a portfolio by adding an asset with a lower CAGR, and I think many of us typically care more about CAGR than the arithmetic mean of returns. Historically (i.e. since the early 70s), adding gold had such an effect. Whether gold or QSPRX will in the future is certainly up for debate.
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

CuriousTacos wrote: Thu May 13, 2021 4:55 pm
nisiprius wrote: Thu May 13, 2021 2:52 pm the return of a portfolio is the weighted average of the returns of its parts. It's as simple as that. Nothing about correlations or volatility can change it. It follows that if you have a portfolio P and a proposed diversifier D, if D has lower return than P, then adding D to the portfolio will reduce its return.
I don't necessarily disagree with your conclusions, but in the above, I think you are referring to arithmetic means rather than geometric means (aka CAGR). It is possible to increase the CAGR of a portfolio by adding an asset with a lower CAGR, and I think many of us typically care more about CAGR than the arithmetic mean of returns. Historically (i.e. since the early 70s), adding gold had such an effect. Whether gold or QSPRX will in the future is certainly up for debate.
Yes. Two portfolios with the same weighted simple average return will have different compounded returns if their volatilities are different. The portfolio with lower volatility will have a higher compounded return and accumulate more money. It’s compounded return will be closer to its simple average return. The volatility of a portfolio is always less than the weighted simple average of component volatilities because of correlations less than 1 between components.

Dave
afan
Posts: 8195
Joined: Sun Jul 25, 2010 4:01 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by afan »

Random Walker wrote: Thu May 13, 2021 5:29 pm
CuriousTacos wrote: Thu May 13, 2021 4:55 pm
nisiprius wrote: Thu May 13, 2021 2:52 pm the return of a portfolio is the weighted average of the returns of its parts. It's as simple as that. Nothing about correlations or volatility can change it. It follows that if you have a portfolio P and a proposed diversifier D, if D has lower return than P, then adding D to the portfolio will reduce its return.
I don't necessarily disagree with your conclusions, but in the above, I think you are referring to arithmetic means rather than geometric means (aka CAGR). It is possible to increase the CAGR of a portfolio by adding an asset with a lower CAGR, and I think many of us typically care more about CAGR than the arithmetic mean of returns. Historically (i.e. since the early 70s), adding gold had such an effect. Whether gold or QSPRX will in the future is certainly up for debate.
Yes. Two portfolios with the same weighted simple average return will have different compounded returns if their volatilities are different. The portfolio with lower volatility will have a higher compounded return and accumulate more money. It’s compounded return will be closer to its simple average return. The volatility of a portfolio is always less than the weighted simple average of component volatilities because of correlations less than 1 between components.

Dave
Reading nisiprius's posts, I am pretty sure they know this.

I believe the comparison remains useful when one looks at relative risk adjusted return, which captures the volatility differences.

This fund has outstanding marketing. Always has. They got a lot of paid advisers on board, who helped with the sales job.

I suppose those who are willing to pay 1% AUM for this kind of advice are also willing to pay 1.81% for the fund itself. Given that one can compose a portfolio based on solid evidence for an all-in cost of less than 10 basis points, the appeal of paying 28x as much is completely lost on me.

I would need overwhelming evidence in favor of this approach to even consider it. As it is, there is nothing close. Just marketing claims.

And huge costs.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama
Jebediah
Posts: 805
Joined: Tue Aug 28, 2012 9:19 pm
Location: Austin TX

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Jebediah »

Random Walker wrote: Thu May 13, 2021 9:39 am
typical.investor wrote: Thu May 13, 2021 8:06 am Anyway, selling value type investments after underperformance just isn’t good strategy. Not sure why momentum hasn’t helped it more... but maybe shorting growth has just been that bad.
Here’s Cliff’s Gut Punch essay. He describes how the bad streak is due first to value. During this period momentum helped but not enough. Second, when value started to do well, it wasn’t enough to overcome a poor stretch for momentum. I think he’s saying the rapidity of the value turn around is what hurt momentum. He hopes that if value starts to make an extended positive run, then momentum will likewise go in a positive direction. Strongly recommend this essay. Actually, I recommend all of Cliff Asness’ essays. His style is great. Educational, clear, and funny. The footnotes are must read material.

https://www.aqr.com/Insights/Perspectives/A-Gut-Punch

Dave
In other words, the strategy is such that this fund will most often find a way to lose. It's not about outcomes, the strategy is the problem. When they told you "equity-like expected returns" they were not being honest, and it was easier to trust the advisor than to verify that claim for yourself. In the vernacular, this is known as a hustle.
User avatar
nisiprius
Advisory Board
Posts: 52217
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: AQR Style Premia Update; Certainly Low Correlations

Post by nisiprius »

CuriousTacos wrote: Thu May 13, 2021 4:55 pm
nisiprius wrote: Thu May 13, 2021 2:52 pm the return of a portfolio is the weighted average of the returns of its parts. It's as simple as that. Nothing about correlations or volatility can change it. It follows that if you have a portfolio P and a proposed diversifier D, if D has lower return than P, then adding D to the portfolio will reduce its return.
I don't necessarily disagree with your conclusions, but in the above, I think you are referring to arithmetic means rather than geometric means (aka CAGR).
Yes, because geometric means are appropriate when we are dealing with a series of consecutive time period, each building on the final value of the previous one... a parlay, if you like. But when we are dealing with several assets growing simultaneously over the same time period, arithmetic means are appropriate.

Over any period of time whatsoever, the return of a portfolio is the plain old linear arithmetic weighted average of the returns of its components. Compounding doesn't enter into it, or, rather, it has already entered into each of the individual returns before they are averaged together.

If a portfolio is $10,000 each of asset A and B, and over some period of time
asset A grows from $10,000 to $20,000, a return of +100%, and
asset B grows from $10,000 to $30,000, a return of +200%,
the portfolio as a whole grows from $20,000 to $50,000, which is a return of +150%...
which is the exact plain old arithmetic mean +100% and +200%... of 50% of 100% + 50% of 200% = 150%.

To do that calculation I didn't need to know how long that period of time was, what path A and B each followed to get to their final value, how much they fluctuated or how much the portfolio fluctuated or what their correlation was. None of those entered into that calculation.

If portfolio P would have earned 10% over some specified time period, and portfolio D would have lost -10% over that same time period, then a portfolio of 90% P, 10% D would have earned 90% of 10% + 10% of (-10%) = 9% - 1% = 8%. In general, a portfolio will earn less money if you add a money-losing asset to it.

We can talk about beneficial effects on volatility that might (or might not) make it worthwhile to add D anyway, but the basic arithmetic of return is what it is. You have to start from that bedrock--adding a losing asset lowers return. Then, you can discuss whether lowering return is always bad--no, it isn't if you lower volatility or maximum drawdown or something else you care about. And then you can discuss whether the particular combination of particular assets meet the rather stringent criteria under which a money-losing asset can theoretically improve a portfolio by metrics other than simple return.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
okwriter
Posts: 242
Joined: Mon Apr 12, 2021 2:00 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by okwriter »

nisiprius wrote: Thu May 13, 2021 2:52 pm To riff on Bogle's phrase, another application of the "relentless rules of humble arithmetic" is that the return of a portfolio is the weighted average of the returns of its parts. It's as simple as that. Nothing about correlations or volatility can change it.
Maybe I'm misinterpreting this statement, but this example seems to contradict it. 90% TSM + 10% ext. duration treasury has better returns than either of its components.
CuriousTacos
Posts: 690
Joined: Thu Apr 12, 2018 3:31 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by CuriousTacos »

nisiprius wrote: Thu May 13, 2021 6:13 pm
CuriousTacos wrote: Thu May 13, 2021 4:55 pm
nisiprius wrote: Thu May 13, 2021 2:52 pm the return of a portfolio is the weighted average of the returns of its parts. It's as simple as that. Nothing about correlations or volatility can change it. It follows that if you have a portfolio P and a proposed diversifier D, if D has lower return than P, then adding D to the portfolio will reduce its return.
I don't necessarily disagree with your conclusions, but in the above, I think you are referring to arithmetic means rather than geometric means (aka CAGR).
Yes, because geometric means are appropriate when we are dealing with a series of consecutive time period, each building on the final value of the previous one... a parlay, if you like. But when we are dealing with several assets growing simultaneously over the same time period, arithmetic means are appropriate.

Over any period of time whatsoever, the return of a portfolio is the plain old linear arithmetic weighted average of the returns of its components. Compounding doesn't enter into it, or, rather, it has already entered into each of the individual returns before they are averaged together.

If a portfolio is $10,000 each of asset A and B, and over some period of time
asset A grows from $10,000 to $20,000, a return of +100%, and
asset B grows from $10,000 to $30,000, a return of +200%,
the portfolio as a whole grows from $20,000 to $50,000, which is a return of +150%...
which is the exact plain old arithmetic mean +100% and +200%... of 50% of 100% + 50% of 200% = 150%.

To do that calculation I didn't need to know how long that period of time was, what path A and B each followed to get to their final value, how much they fluctuated or how much the portfolio fluctuated or what their correlation was. None of those entered into that calculation.

If portfolio P would have earned 10% over some specified time period, and portfolio D would have lost -10% over that same time period, then a portfolio of 90% P, 10% D would have earned 90% of 10% + 10% of (-10%) = 9% - 1% = 8%. In general, a portfolio will earn less money if you add a money-losing asset to it.

We can talk about beneficial effects on volatility that might (or might not) make it worthwhile to add D anyway, but the basic arithmetic of return is what it is. You have to start from that bedrock--adding a losing asset lowers return. Then, you can discuss whether lowering return is always bad--no, it isn't if you lower volatility or maximum drawdown or something else you care about. And then you can discuss whether the particular combination of particular assets meet the rather stringent criteria under which a money-losing asset can theoretically improve a portfolio by metrics other than simple return.
I agree that weighted arithmetic means are correct for a single time period of any length without rebalancing, but over longer periods of time where a typical investor would be rebalancing, CAGR is more applicable.
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

nisiprius wrote: Thu May 13, 2021 6:13 pm
CuriousTacos wrote: Thu May 13, 2021 4:55 pm
nisiprius wrote: Thu May 13, 2021 2:52 pm the return of a portfolio is the weighted average of the returns of its parts. It's as simple as that. Nothing about correlations or volatility can change it. It follows that if you have a portfolio P and a proposed diversifier D, if D has lower return than P, then adding D to the portfolio will reduce its return.
I don't necessarily disagree with your conclusions, but in the above, I think you are referring to arithmetic means rather than geometric means (aka CAGR).
Yes, because geometric means are appropriate when we are dealing with a series of consecutive time period, each building on the final value of the previous one... a parlay, if you like. But when we are dealing with several assets growing simultaneously over the same time period, arithmetic means are appropriate.

Over any period of time whatsoever, the return of a portfolio is the plain old linear arithmetic weighted average of the returns of its components. Compounding doesn't enter into it, or, rather, it has already entered into each of the individual returns before they are averaged together.

If a portfolio is $10,000 each of asset A and B, and over some period of time
asset A grows from $10,000 to $20,000, a return of +100%, and
asset B grows from $10,000 to $30,000, a return of +200%,
the portfolio as a whole grows from $20,000 to $50,000, which is a return of +150%...
which is the exact plain old arithmetic mean +100% and +200%... of 50% of 100% + 50% of 200% = 150%.

To do that calculation I didn't need to know how long that period of time was, what path A and B each followed to get to their final value, how much they fluctuated or how much the portfolio fluctuated or what their correlation was. None of those entered into that calculation.

If portfolio P would have earned 10% over some specified time period, and portfolio D would have lost -10% over that same time period, then a portfolio of 90% P, 10% D would have earned 90% of 10% + 10% of (-10%) = 9% - 1% = 8%. In general, a portfolio will earn less money if you add a money-losing asset to it.

We can talk about beneficial effects on volatility that might (or might not) make it worthwhile to add D anyway, but the basic arithmetic of return is what it is. You have to start from that bedrock--adding a losing asset lowers return. Then, you can discuss whether lowering return is always bad--no, it isn't if you lower volatility or maximum drawdown or something else you care about. And then you can discuss whether the particular combination of particular assets meet the rather stringent criteria under which a money-losing asset can theoretically improve a portfolio by metrics other than simple return.
I am not a mathematician or statistician but I believe your example is a bit off. Your example is looking backwards with returns that are already known. Yes when you know the starting value and terminal value, the compounded return is implicit. But we only invest looking forward with unknown returns looking forward. We deal with forward looking expected returns, which are simple mean average returns. We do know that volatility will be a drag on the portfolio. The geometric return will be less than the portfolio’s simple average return. So if two portfolios have the same mean expected return but different expected volatilities, we can expect the lower volatility portfolio to have a greater compounded return going forward; it’s a more efficient portfolio. Of course these are only expected returns, expected volatilities, expected correlations. And necessarily this is why we diversify.

Dave
Last edited by Random Walker on Thu May 13, 2021 7:23 pm, edited 1 time in total.
JamesDean44
Posts: 241
Joined: Sat Jan 13, 2018 6:36 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by JamesDean44 »

okwriter wrote: Thu May 13, 2021 6:27 pm
nisiprius wrote: Thu May 13, 2021 2:52 pm To riff on Bogle's phrase, another application of the "relentless rules of humble arithmetic" is that the return of a portfolio is the weighted average of the returns of its parts. It's as simple as that. Nothing about correlations or volatility can change it.
Maybe I'm misinterpreting this statement, but this example seems to contradict it. 90% TSM + 10% ext. duration treasury has better returns than either of its components.
The statement is false because it doesn't account for rebalancing.
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

JamesDean44 wrote: Thu May 13, 2021 7:14 pm
okwriter wrote: Thu May 13, 2021 6:27 pm
nisiprius wrote: Thu May 13, 2021 2:52 pm To riff on Bogle's phrase, another application of the "relentless rules of humble arithmetic" is that the return of a portfolio is the weighted average of the returns of its parts. It's as simple as that. Nothing about correlations or volatility can change it.
Maybe I'm misinterpreting this statement, but this example seems to contradict it. 90% TSM + 10% ext. duration treasury has better returns than either of its components.
The statement is false because it doesn't account for rebalancing.
Important to also appreciate that we don’t even need rebalancing to see the benefit of reduced volatility. After a big decline, the less volatile portfolio will have seen a smaller drawdown and thus have more money to reinvest in the subsequent gain.

Portfolio 1
Lose 50%, gain 100% to break even: simple average return 25%, CAGR 0%
Portfolio 2
Lose 25%, gain 33% to break even: simple average return 4%, CAGR 0%

Dave
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: AQR Style Premia Update; Certainly Low Correlations

Post by willthrill81 »

Random Walker wrote: Thu May 13, 2021 7:48 am
Alchemist wrote: Thu May 13, 2021 4:48 am You could have just held cash in a HYSA and done far better. Even in a brokerage account a STT fund would have stomped QSPRX.

Also don't forget that in addition to the huge ER of the fund, you need to pay an advisor for the privilege of having access to it :oops:

So the middlemen are getting paid at least.
Cant judge a strategy by the outcome.
Well....maybe. We should be careful to avoid 'resulting' (i.e., basing the quality of a decision on the basis of an unknown outcome pertaining to that decision). But at the same time, if an investment strategy pretty consistently loses money, it may be evidence that there are aspects, variables, etc. pertaining to the decision that were misjudged, not taken into account, etc., and thereby made the decision a poor one due to incomplete and/or incorrect information being used.

If an investment strategy lost money for 50 years straight, it doesn't really matter how good the theory underlying the decision is; virtually everyone would agree that it's a poor strategy. But in this context, 5 years may be too short of a time in which to judge the quality of the decision. After all, we already know that factors such as market beta, the small premiums, etc., can be negative for 20 years or longer.
The Sensible Steward
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

willthrill81 wrote: Thu May 13, 2021 7:35 pm
Random Walker wrote: Thu May 13, 2021 7:48 am
Alchemist wrote: Thu May 13, 2021 4:48 am You could have just held cash in a HYSA and done far better. Even in a brokerage account a STT fund would have stomped QSPRX.

Also don't forget that in addition to the huge ER of the fund, you need to pay an advisor for the privilege of having access to it :oops:

So the middlemen are getting paid at least.
Cant judge a strategy by the outcome.
Well....maybe. We should be careful to avoid 'resulting' (i.e., basing the quality of a decision on the basis of an unknown outcome pertaining to that decision). But at the same time, if an investment strategy pretty consistently loses money, it may be evidence that there are aspects, variables, etc. pertaining to the decision that were misjudged, not taken into account, etc., and thereby made the decision a poor one due to incomplete and/or incorrect information being used.

If an investment strategy lost money for 50 years straight, it doesn't really matter how good the theory underlying the decision is; virtually everyone would agree that it's a poor strategy. But in this context, 5 years may be too short of a time in which to judge the quality of the decision. After all, we already know that factors such as market beta, the small premiums, etc., can be negative for 20 years or longer.
Good balanced discussion of the importance of ultimate ex post result and significance of ex ante strategy. We can only invest looking forward. I think a good way to evaluate potential future investments is the criteria laid out in Larry’s factor book: persistence, pervasive, robust, intuitive, investable. The all important costs fall into the investable category.

Dave
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: AQR Style Premia Update; Certainly Low Correlations

Post by willthrill81 »

Random Walker wrote: Thu May 13, 2021 7:44 pm
willthrill81 wrote: Thu May 13, 2021 7:35 pm
Random Walker wrote: Thu May 13, 2021 7:48 am
Alchemist wrote: Thu May 13, 2021 4:48 am You could have just held cash in a HYSA and done far better. Even in a brokerage account a STT fund would have stomped QSPRX.

Also don't forget that in addition to the huge ER of the fund, you need to pay an advisor for the privilege of having access to it :oops:

So the middlemen are getting paid at least.
Cant judge a strategy by the outcome.
Well....maybe. We should be careful to avoid 'resulting' (i.e., basing the quality of a decision on the basis of an unknown outcome pertaining to that decision). But at the same time, if an investment strategy pretty consistently loses money, it may be evidence that there are aspects, variables, etc. pertaining to the decision that were misjudged, not taken into account, etc., and thereby made the decision a poor one due to incomplete and/or incorrect information being used.

If an investment strategy lost money for 50 years straight, it doesn't really matter how good the theory underlying the decision is; virtually everyone would agree that it's a poor strategy. But in this context, 5 years may be too short of a time in which to judge the quality of the decision. After all, we already know that factors such as market beta, the small premiums, etc., can be negative for 20 years or longer.
Good balanced discussion of the importance of ultimate ex post result and significance of ex ante strategy. We can only invest looking forward. I think a good way to evaluate potential future investments is the criteria laid out in Larry’s factor book: persistence, pervasive, robust, intuitive, investable. The all important costs fall into the investable category.

Dave
I agree that Larry's criteria are sound. What I am unsure of is whether the costs of a fund like QSPRX are offset by its diversification benefit.

Also, I think that Larry has played a little fast and loose with the persistence and pervasive aspects of these decisions. For instance, he was (and still may be) recommending LENDX pretty strongly when investors had not even had access to the consumer credit asset class for more than about a decade. Granted, banks have had access to this asset class for decades, but I don't know if that means that we can equate what banks were doing with what LENDX had done at the time.
The Sensible Steward
Alchemist
Posts: 638
Joined: Sat Aug 30, 2014 6:35 am

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Alchemist »

Random Walker wrote: Thu May 13, 2021 8:38 am Are you sure? What if you had entered the stock market in 2000 and evaluated your results in 2012? Would have given up on the market factor forever? The S&P500 underperformed TBills for three long stretches: 1929-43, 1966-82, 2000-12. It’s really really hard to know when to give up on a strategy. I’ll agree with you that the threshold for throwing in the towel is probably way lower when expenses are high!

Dave
My previous comment was short on details so I will attempt to expand on my point below.

A strategy must be evaluated on its outcome compared to its goals. For instance if you want to lower the risk of your portfolio by adding more bonds then your potential returns will be expected to decrease. If your addition of bonds increased the risk of your portfolio then something is wrong. Maybe you added too many munis/corporates and would have been better off with more treasuries for instance. But the outcome is how you know if it worked or not...judged against the objective.

I think most bogleheads owning equities expect there to be periods of low performance, particularly when caused by market crashes. So if your equity strategy includes the assumptions of high risk of crashes/disappointing returns and those crashes/disappointing returns show up then the outcome is still expected for that strategy. This is the reason many bogleheads have high quality bonds and/or cash emergency funds as part of their strategy.

In the case of QSPRX and other Alts, the strategy's objective was 'bond like returns with equity like returns'. Except over the five years these funds existed they have done the complete opposite. Stock like volatility with sub-bond (in fact, negative) returns. The strategy advertised by AQR to provide nominal positive returns in all market environments failed. The strategy is thus shown to have been a bad one as the outcome is far from the objective.
Jebediah
Posts: 805
Joined: Tue Aug 28, 2012 9:19 pm
Location: Austin TX

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Jebediah »

I posit that the outcome (8% stdev, 1% return) is exactly what is expected from the strategy. The marketing was simply dishonest about the expected returns.
User avatar
typical.investor
Posts: 5263
Joined: Mon Jun 11, 2018 3:17 am

Re: AQR Style Premia Update; Certainly Low Correlations

Post by typical.investor »

Jebediah wrote: Sat May 15, 2021 10:38 am I posit that the outcome (8% stdev, 1% return) is exactly what is expected from the strategy. The marketing was simply dishonest about the expected returns.
The index tracks the BofA Merrill Lynch US 3-Month Treasury Bill.

Oct 2014 - Apr 2021
QSPRX
CAGR -0.42%
Stdev 8.91%

BIL (Barclays 1-3 Mth T-Bill ETF)
CAGR 0.72%
Stdev 0.27%

I think it's fair to say value has gone under a harsh drawdown in the last few years, and has suffered for longer than that. That's typically the excuse for the performance and so may be it. But given that performance seems so tied to value, why add this to a value tiled portfolio?
afan
Posts: 8195
Joined: Sun Jul 25, 2010 4:01 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by afan »

if an investment strategy pretty consistently loses money, it may be evidence that there are aspects, variables, etc. pertaining to the decision that were misjudged, not taken into account, etc.,
Costs matter.

Remember that many investors got in through advisers who charged huge fees themselves. Most of these investors can only wish their losses matched those of the fund.

Paying 1.81% of assets is always a losing strategy.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

typical.investor wrote: Sat May 15, 2021 12:39 pm
Jebediah wrote: Sat May 15, 2021 10:38 am I posit that the outcome (8% stdev, 1% return) is exactly what is expected from the strategy. The marketing was simply dishonest about the expected returns.
The index tracks the BofA Merrill Lynch US 3-Month Treasury Bill.

Oct 2014 - Apr 2021
QSPRX
CAGR -0.42%
Stdev 8.91%

BIL (Barclays 1-3 Mth T-Bill ETF)
CAGR 0.72%
Stdev 0.27%

I think it's fair to say value has gone under a harsh drawdown in the last few years, and has suffered for longer than that. That's typically the excuse for the performance and so may be it. But given that performance seems so tied to value, why add this to a value tiled portfolio?
I don’t think TBills are a good comparison. Should expect much more from the fund.
The reason to choose this fund potentially as a complement to a long only value fund is that a long only value fund is still dominated by market beta. Value and market have very low correlation. So the goal is to invest in unique and independent factors, a move towards risk parity. This fund provides access to value without taking on more market factor. Also accesses value across asset classes other than equity.

Dave
afan
Posts: 8195
Joined: Sun Jul 25, 2010 4:01 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by afan »

Random Walker wrote: Sat May 15, 2021 1:34 pm
I don’t think TBills are a good comparison. Should expect much more from the fund.
T bills are the benchmark the fund proposes. This implies low volatility. It is hopelessly off the mark on that measure.

T bills also have low correlation with stocks and bonds. One could get low correlation with those markets, at vastly lower cost, with dramatically lower volatility by just picking up some bills. The superior performance is just a bonus.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama
User avatar
nedsaid
Posts: 19275
Joined: Fri Nov 23, 2012 11:33 am

Re: AQR Style Premia Update; Certainly Low Correlations

Post by nedsaid »

I want to say that I am rooting for Larry Swedroe, Cliff Asness, and AQR to be proven right about Alternative Investments that use hedge fund like strategies. I am concerned this is the triumph of hope over experience. My very limited experience with such funds is disappointing and while I am with Swedroe and Asness regarding Academic Research and factors, I depart from them when it comes to these complex trading strategies. While I like long-only factor strategies; the use of shorts, levarage, and complex derivatives makes me nervous. It is like the movie where paratroopers were sent too far behind enemy lines with disastrous battlefield results, A Bridge Too Far. It is sort of like trying too hard or being too clever. No one likes to parachute into a Panzer Division.

I am applauding them for giving this the good old college try but I have a sense of dread that all will not work out as expected. I would like nothing better to be proven wrong as I cheer Swedroe, Asness, and AQR from the sidelines. I just hope it all works eventually.
A fool and his money are good for business.
Angst
Posts: 2968
Joined: Sat Jun 09, 2007 11:31 am

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Angst »

Random Walker wrote: Sat May 15, 2021 1:34 pm I don’t think TBills are a good comparison. Should expect much more from the fund.
The reason to choose this fund potentially as a complement to a long only value fund is that a long only value fund is still dominated by market beta. Value and market have very low correlation. So the goal is to invest in unique and independent factors, a move towards risk parity. This fund provides access to value without taking on more market factor. Also accesses value across asset classes other than equity.

Dave
But Dave, since when was QSPIX supposed to be considered a kind of "value" fund? It seems to me, this has only been since value's poor performance became the explanation for the fund's poor performance. My recollection though of when QSPIX first came out and was talked about here in the forum was that it's "value" (pun intended) was in it's diversification of exposure to all the different factors; they all were to contribute to what at the time seemed like a smart, sophisticated, fairly inscrutable and exclusive way of getting what was going to be "equity like" returns without correlation to equity! I belatedly came to see that slogan as explicitly oxymoronic. As such, the low bar of TBills might actually be fairly appropriate. :shock:
Topic Author
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: AQR Style Premia Update; Certainly Low Correlations

Post by Random Walker »

Angst wrote: Sun May 16, 2021 7:11 am
Random Walker wrote: Sat May 15, 2021 1:34 pm I don’t think TBills are a good comparison. Should expect much more from the fund.
The reason to choose this fund potentially as a complement to a long only value fund is that a long only value fund is still dominated by market beta. Value and market have very low correlation. So the goal is to invest in unique and independent factors, a move towards risk parity. This fund provides access to value without taking on more market factor. Also accesses value across asset classes other than equity.

Dave
But Dave, since when was QSPIX supposed to be considered a kind of "value" fund? It seems to me, this has only been since value's poor performance became the explanation for the fund's poor performance. My recollection though of when QSPIX first came out and was talked about here in the forum was that it's "value" (pun intended) was in it's diversification of exposure to all the different factors; they all were to contribute to what at the time seemed like a smart, sophisticated, fairly inscrutable and exclusive way of getting what was going to be "equity like" returns without correlation to equity! I belatedly came to see that slogan as explicitly oxymoronic. As such, the low bar of TBills might actually be fairly appropriate. :shock:
I agree. I was only responding to the focus on value, equity value in particular. When the portfolio invests in 4 styles across 4 asset classes, it concerns me too that 1 style in one asset class would have such a dominant effect.

Dave
Post Reply