Small Cap Value heads Rejoice !!!

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
countmein
Posts: 564
Joined: Fri Dec 06, 2013 9:10 pm

Re: Small Cap Value heads Rejoice !!!

Post by countmein »

absolute zero wrote: Mon Jul 19, 2021 11:18 am
countmein wrote: Mon Jul 19, 2021 10:01 am Even if we round 0.35% up to 1%, that's not good. Doesn't look any different than noise.
Where are you getting 0.35%?
AQR model, HmL since 5/1/86
countmein
Posts: 564
Joined: Fri Dec 06, 2013 9:10 pm

Re: Small Cap Value heads Rejoice !!!

Post by countmein »

KyleAAA wrote: Mon Jul 19, 2021 11:05 am
countmein wrote: Mon Jul 19, 2021 10:01 am 1- Even if we round 0.35% up to 1%, that's not good. Doesn't look any different than noise.
FF shows a 1.13% value premium over the same period and even AQR shows an HML-Dev of 0.74% over that same period. Looks like a significant premium to me. 0.35% over 35 years makes a significant difference.
Ok, let's go with 1.13%. Your actual premium is 1.13 * .5 (long only) * .66 (DFSVX exposure) = 0.37% - 0.22 (AVUV ER minus VTI ER) = 0.15%

If we use AQR's 0.35%, the same calculation gives a premium of -0.10%

If you think that's worth double the volatility then you might like these things called BaB (8.15% over the same period) and QmJ (4.76%).
Nathan Drake
Posts: 1536
Joined: Mon Apr 11, 2011 12:28 am

Re: Small Cap Value heads Rejoice !!!

Post by Nathan Drake »

countmein wrote: Mon Jul 19, 2021 1:02 pm
KyleAAA wrote: Mon Jul 19, 2021 11:05 am
countmein wrote: Mon Jul 19, 2021 10:01 am 1- Even if we round 0.35% up to 1%, that's not good. Doesn't look any different than noise.
FF shows a 1.13% value premium over the same period and even AQR shows an HML-Dev of 0.74% over that same period. Looks like a significant premium to me. 0.35% over 35 years makes a significant difference.
Ok, let's go with 1.13%. Your actual premium is 1.13 * .5 (long only) * .66 (DFSVX exposure) = 0.37% - 0.22 (AVUV ER minus VTI ER) = 0.15%

If we use AQR's 0.35%, the same calculation gives a premium of -0.10%

If you think that's worth double the volatility then you might like these things called BaB (8.15% over the same period) and QmJ (4.76%).
Why are you multiplying by .5 and .66?

Portfoliovisualizer shows us the difference and it’s not as low as you suggest even in an incredibly ideal back test
countmein
Posts: 564
Joined: Fri Dec 06, 2013 9:10 pm

Re: Small Cap Value heads Rejoice !!!

Post by countmein »

Nathan Drake wrote: Mon Jul 19, 2021 2:33 pm
countmein wrote: Mon Jul 19, 2021 1:02 pm
KyleAAA wrote: Mon Jul 19, 2021 11:05 am
countmein wrote: Mon Jul 19, 2021 10:01 am 1- Even if we round 0.35% up to 1%, that's not good. Doesn't look any different than noise.
FF shows a 1.13% value premium over the same period and even AQR shows an HML-Dev of 0.74% over that same period. Looks like a significant premium to me. 0.35% over 35 years makes a significant difference.
Ok, let's go with 1.13%. Your actual premium is 1.13 * .5 (long only) * .66 (DFSVX exposure) = 0.37% - 0.22 (AVUV ER minus VTI ER) = 0.15%

If we use AQR's 0.35%, the same calculation gives a premium of -0.10%

If you think that's worth double the volatility then you might like these things called BaB (8.15% over the same period) and QmJ (4.76%).
Why are you multiplying by .5 and .66?

Portfoliovisualizer shows us the difference and it’s not as low as you suggest even in an incredibly ideal back test
You're right, we should only multiply by 0.66 (the HmL weight of DFSVX, intended to exemplify real world factor exposure).

So, for the past 35 years, on the high side (FF) we've got an HmL premium of (1.13 * 0.66) - 0.22 = 0.66%

On the low side (AQR), we've got an HmL premium of (0.35 * 0.66) - 0.22 = 0.01%

Seems fair to average them out: (.66 + .01)/2 = 0.34% is our value premium for the past 35 years.
Nathan Drake
Posts: 1536
Joined: Mon Apr 11, 2011 12:28 am

Re: Small Cap Value heads Rejoice !!!

Post by Nathan Drake »

countmein wrote: Mon Jul 19, 2021 3:25 pm
Nathan Drake wrote: Mon Jul 19, 2021 2:33 pm
countmein wrote: Mon Jul 19, 2021 1:02 pm
KyleAAA wrote: Mon Jul 19, 2021 11:05 am
countmein wrote: Mon Jul 19, 2021 10:01 am 1- Even if we round 0.35% up to 1%, that's not good. Doesn't look any different than noise.
FF shows a 1.13% value premium over the same period and even AQR shows an HML-Dev of 0.74% over that same period. Looks like a significant premium to me. 0.35% over 35 years makes a significant difference.
Ok, let's go with 1.13%. Your actual premium is 1.13 * .5 (long only) * .66 (DFSVX exposure) = 0.37% - 0.22 (AVUV ER minus VTI ER) = 0.15%

If we use AQR's 0.35%, the same calculation gives a premium of -0.10%

If you think that's worth double the volatility then you might like these things called BaB (8.15% over the same period) and QmJ (4.76%).
Why are you multiplying by .5 and .66?

Portfoliovisualizer shows us the difference and it’s not as low as you suggest even in an incredibly ideal back test
You're right, we should only multiply by 0.66 (the HmL weight of DFSVX, intended to exemplify real world factor exposure).

So, for the past 35 years, on the high side (FF) we've got an HmL premium of (1.13 * 0.66) - 0.22 = 0.66%

On the low side (AQR), we've got an HmL premium of (0.35 * 0.66) - 0.22 = 0.01%

Seems fair to average them out: (.66 + .01)/2 = 0.34% is our value premium for the past 35 years.
The backtest of 1.13% premium shows us the real world returns. I am missing why we need to multiply the real world premium by the factor loading?
KyleAAA
Posts: 8853
Joined: Wed Jul 01, 2009 5:35 pm
Contact:

Re: Small Cap Value heads Rejoice !!!

Post by KyleAAA »

countmein wrote: Mon Jul 19, 2021 3:25 pm
Nathan Drake wrote: Mon Jul 19, 2021 2:33 pm
countmein wrote: Mon Jul 19, 2021 1:02 pm
KyleAAA wrote: Mon Jul 19, 2021 11:05 am
countmein wrote: Mon Jul 19, 2021 10:01 am 1- Even if we round 0.35% up to 1%, that's not good. Doesn't look any different than noise.
FF shows a 1.13% value premium over the same period and even AQR shows an HML-Dev of 0.74% over that same period. Looks like a significant premium to me. 0.35% over 35 years makes a significant difference.
Ok, let's go with 1.13%. Your actual premium is 1.13 * .5 (long only) * .66 (DFSVX exposure) = 0.37% - 0.22 (AVUV ER minus VTI ER) = 0.15%

If we use AQR's 0.35%, the same calculation gives a premium of -0.10%

If you think that's worth double the volatility then you might like these things called BaB (8.15% over the same period) and QmJ (4.76%).
Why are you multiplying by .5 and .66?

Portfoliovisualizer shows us the difference and it’s not as low as you suggest even in an incredibly ideal back test
You're right, we should only multiply by 0.66 (the HmL weight of DFSVX, intended to exemplify real world factor exposure).

So, for the past 35 years, on the high side (FF) we've got an HmL premium of (1.13 * 0.66) - 0.22 = 0.66%

On the low side (AQR), we've got an HmL premium of (0.35 * 0.66) - 0.22 = 0.01%

Seems fair to average them out: (.66 + .01)/2 = 0.34% is our value premium for the past 35 years.
That's quite a large premium over 35 years.
countmein
Posts: 564
Joined: Fri Dec 06, 2013 9:10 pm

Re: Small Cap Value heads Rejoice !!!

Post by countmein »

Nathan Drake wrote: Mon Jul 19, 2021 4:09 pm
countmein wrote: Mon Jul 19, 2021 3:25 pm
Nathan Drake wrote: Mon Jul 19, 2021 2:33 pm
countmein wrote: Mon Jul 19, 2021 1:02 pm
KyleAAA wrote: Mon Jul 19, 2021 11:05 am

FF shows a 1.13% value premium over the same period and even AQR shows an HML-Dev of 0.74% over that same period. Looks like a significant premium to me. 0.35% over 35 years makes a significant difference.
Ok, let's go with 1.13%. Your actual premium is 1.13 * .5 (long only) * .66 (DFSVX exposure) = 0.37% - 0.22 (AVUV ER minus VTI ER) = 0.15%

If we use AQR's 0.35%, the same calculation gives a premium of -0.10%

If you think that's worth double the volatility then you might like these things called BaB (8.15% over the same period) and QmJ (4.76%).
Why are you multiplying by .5 and .66?

Portfoliovisualizer shows us the difference and it’s not as low as you suggest even in an incredibly ideal back test
You're right, we should only multiply by 0.66 (the HmL weight of DFSVX, intended to exemplify real world factor exposure).

So, for the past 35 years, on the high side (FF) we've got an HmL premium of (1.13 * 0.66) - 0.22 = 0.66%

On the low side (AQR), we've got an HmL premium of (0.35 * 0.66) - 0.22 = 0.01%

Seems fair to average them out: (.66 + .01)/2 = 0.34% is our value premium for the past 35 years.
The backtest of 1.13% premium shows us the real world returns. I am missing why we need to multiply the real world premium by the factor loading?
0.35% and 1.13% are the pure (long-short) factor returns. If your fund's loading were 1.0, you'd get these returns. Best real-world example of a deep small value fund with a long history is DFSVX. It captures about 66% of the HmL factor. I used AVUV's ER to be charitable, and assume it has a similar exposure as DFSVX. Using DFSVX's ER, the HmL return drops to 0.17%
Nathan Drake
Posts: 1536
Joined: Mon Apr 11, 2011 12:28 am

Re: Small Cap Value heads Rejoice !!!

Post by Nathan Drake »

countmein wrote: Mon Jul 19, 2021 6:51 pm
Nathan Drake wrote: Mon Jul 19, 2021 4:09 pm
countmein wrote: Mon Jul 19, 2021 3:25 pm
Nathan Drake wrote: Mon Jul 19, 2021 2:33 pm
countmein wrote: Mon Jul 19, 2021 1:02 pm

Ok, let's go with 1.13%. Your actual premium is 1.13 * .5 (long only) * .66 (DFSVX exposure) = 0.37% - 0.22 (AVUV ER minus VTI ER) = 0.15%

If we use AQR's 0.35%, the same calculation gives a premium of -0.10%

If you think that's worth double the volatility then you might like these things called BaB (8.15% over the same period) and QmJ (4.76%).
Why are you multiplying by .5 and .66?

Portfoliovisualizer shows us the difference and it’s not as low as you suggest even in an incredibly ideal back test
You're right, we should only multiply by 0.66 (the HmL weight of DFSVX, intended to exemplify real world factor exposure).

So, for the past 35 years, on the high side (FF) we've got an HmL premium of (1.13 * 0.66) - 0.22 = 0.66%

On the low side (AQR), we've got an HmL premium of (0.35 * 0.66) - 0.22 = 0.01%

Seems fair to average them out: (.66 + .01)/2 = 0.34% is our value premium for the past 35 years.
The backtest of 1.13% premium shows us the real world returns. I am missing why we need to multiply the real world premium by the factor loading?
0.35% and 1.13% are the pure (long-short) factor returns. If your fund's loading were 1.0, you'd get these returns. Best real-world example of a deep small value fund with a long history is DFSVX. It captures about 66% of the HmL factor. I used AVUV's ER to be charitable, and assume it has a similar exposure as DFSVX. Using DFSVX's ER, the HmL return drops to 0.17%
Portfolio visualizer:

Portfolio Returns
Portfolio 1 $10,000 $154,779 10.47%
Portfolio 2 $10,000 $194,942 11.41%

This is the longest it goes back (1994). That's a near 1% premium for DFSVX over VTSMX for 27 years, with an end date leading to the biggest spreads in growth vs value in recorded history.

Shows to me that the premium is alive and well and likely to do extremely well going forward. You can argue all you want about whether it's worth the increased volatility. Nobody is claiming this is a free lunch. But the benefits go beyond the mere returns; sequence of returns as a diversification benefit against a largely market based portfolio. 5 year periods show significantly higher average returns and less "dead decades". In many ways it looks a lot better from a sequencing standpoint over 5 & 10 year periods than a purely market based portfolio, and certainly in an environment with lower expected market + bond returns.
countmein
Posts: 564
Joined: Fri Dec 06, 2013 9:10 pm

Re: Small Cap Value heads Rejoice !!!

Post by countmein »

I'm going back to 1986 using the factor statistics tool in PV. It goes back to 1960-something (where the premium used to be). It's useful to look back as far as we can. When you do, it's clear that a premium used to exist. Then, for whatever combination of reasons, it stopped working.
Nathan Drake
Posts: 1536
Joined: Mon Apr 11, 2011 12:28 am

Re: Small Cap Value heads Rejoice !!!

Post by Nathan Drake »

countmein wrote: Mon Jul 19, 2021 10:19 pm I'm going back to 1986 using the factor statistics tool in PV. It goes back to 1960-something (where the premium used to be). It's useful to look back as far as we can. When you do, it's clear that a premium used to exist. Then, for whatever combination of reasons, it stopped working.
But it doesn’t show that it doesn’t exist. It may show a lowering trend, using an extremely favorable end date, but it doesn’t show that it’s gone. Not sure why you continue to claim that

And a more consistent 10% over 20 years is better than 0% one decade and 14% the next, for obvious sequence reasons
YRT70
Posts: 1055
Joined: Sat Apr 27, 2019 8:51 am

Re: Small Cap Value heads Rejoice !!!

Post by YRT70 »

countmein wrote: Mon Jul 19, 2021 10:01 am 5- Diversification is overstated. Unless you're holding long-short factors, you're not getting much: DFSVX has 0.85 ave historical monthly correlation to VTSAX.
It's longer term diversification that matters more to me. The decade from 2000-2009 VTSMX returned -0.3% while DFSVX returned +9.1%.
countmein wrote: Mon Jul 19, 2021 10:01 am 7- Why the obsession with small value when quality and low vol have historical premia that are many multiples larger (technically infinitely larger if HmL = 0)?
Many small cap value funds also have significant profitability exposure.

PS. most people who believe in tilting to factors also believe in international diversification. An internationally diversified small value portfolio outperformed an international total market portfolio by ~1.9% over the last 25 years. For those that can stomach the greater volatility I'd say it's a worthwhile trade off.
https://www.portfoliovisualizer.com/bac ... tion4_2=45
muffins14
Posts: 633
Joined: Wed Oct 26, 2016 4:14 am

Re: Small Cap Value heads Rejoice !!!

Post by muffins14 »

countmein wrote: Mon Jul 19, 2021 10:19 pm I'm going back to 1986 using the factor statistics tool in PV. It goes back to 1960-something (where the premium used to be). It's useful to look back as far as we can. When you do, it's clear that a premium used to exist. Then, for whatever combination of reasons, it stopped working.
This is a non-quantitative and also incorrect statement. There was an academic paper on this measurement and the premium is statistically consistent with the original measurement. (Yes, there is variance around the point estimates)
countmein
Posts: 564
Joined: Fri Dec 06, 2013 9:10 pm

Re: Small Cap Value heads Rejoice !!!

Post by countmein »

muffins14 wrote: Tue Jul 20, 2021 7:45 am
countmein wrote: Mon Jul 19, 2021 10:19 pm I'm going back to 1986 using the factor statistics tool in PV. It goes back to 1960-something (where the premium used to be). It's useful to look back as far as we can. When you do, it's clear that a premium used to exist. Then, for whatever combination of reasons, it stopped working.
This is a non-quantitative and also incorrect statement. There was an academic paper on this measurement and the premium is statistically consistent with the original measurement. (Yes, there is variance around the point estimates)
Quantitatively, the value "premium" was 0.17% in the 35 years to present using DFSVX as a simulant. I get that Nathan thinks that means "exists" and I think it means "doesn't exist" (and that's fine) but the 0.17 is what it is.
countmein
Posts: 564
Joined: Fri Dec 06, 2013 9:10 pm

Re: Small Cap Value heads Rejoice !!!

Post by countmein »

YRT70 wrote: Tue Jul 20, 2021 7:00 am
countmein wrote: Mon Jul 19, 2021 10:01 am 5- Diversification is overstated. Unless you're holding long-short factors, you're not getting much: DFSVX has 0.85 ave historical monthly correlation to VTSAX.
It's longer term diversification that matters more to me. The decade from 2000-2009 VTSMX returned -0.3% while DFSVX returned +9.1%.
countmein wrote: Mon Jul 19, 2021 10:01 am 7- Why the obsession with small value when quality and low vol have historical premia that are many multiples larger (technically infinitely larger if HmL = 0)?
Many small cap value funds also have significant profitability exposure.

PS. most people who believe in tilting to factors also believe in international diversification. An internationally diversified small value portfolio outperformed an international total market portfolio by ~1.9% over the last 25 years. For those that can stomach the greater volatility I'd say it's a worthwhile trade off.
Interesting point about 2000-2009. I guess the trade off there is no "lost decades" vs many years of trailing TSM with slightly less positive returns. It's not a bad trade. That is a more convincing reason to tilt than pretending 0.17% is anything other than zero + a little noise.
Nathan Drake
Posts: 1536
Joined: Mon Apr 11, 2011 12:28 am

Re: Small Cap Value heads Rejoice !!!

Post by Nathan Drake »

countmein wrote: Tue Jul 20, 2021 7:46 pm
muffins14 wrote: Tue Jul 20, 2021 7:45 am
countmein wrote: Mon Jul 19, 2021 10:19 pm I'm going back to 1986 using the factor statistics tool in PV. It goes back to 1960-something (where the premium used to be). It's useful to look back as far as we can. When you do, it's clear that a premium used to exist. Then, for whatever combination of reasons, it stopped working.
This is a non-quantitative and also incorrect statement. There was an academic paper on this measurement and the premium is statistically consistent with the original measurement. (Yes, there is variance around the point estimates)
Quantitatively, the value "premium" was 0.17% in the 35 years to present using DFSVX as a simulant. I get that Nathan thinks that means "exists" and I think it means "doesn't exist" (and that's fine) but the 0.17 is what it is.
I can find periods nearly as long where bonds beat stocks dramatically.

Guess the equity premium doesn’t exist

viewtopic.php?t=352394
muffins14
Posts: 633
Joined: Wed Oct 26, 2016 4:14 am

Re: Small Cap Value heads Rejoice !!!

Post by muffins14 »

countmein wrote: Tue Jul 20, 2021 7:46 pm
muffins14 wrote: Tue Jul 20, 2021 7:45 am
countmein wrote: Mon Jul 19, 2021 10:19 pm I'm going back to 1986 using the factor statistics tool in PV. It goes back to 1960-something (where the premium used to be). It's useful to look back as far as we can. When you do, it's clear that a premium used to exist. Then, for whatever combination of reasons, it stopped working.
This is a non-quantitative and also incorrect statement. There was an academic paper on this measurement and the premium is statistically consistent with the original measurement. (Yes, there is variance around the point estimates)
Quantitatively, the value "premium" was 0.17% in the 35 years to present using DFSVX as a simulant. I get that Nathan thinks that means "exists" and I think it means "doesn't exist" (and that's fine) but the 0.17 is what it is.
DFSVX Is not the same as the value premium, obviously. You don’t get to just make up your own definitions. It’s not even a value fund, it’s a small-value fund, and that matters because there is no small-value factor. There is a small factor and value factor, and the endpoints you chose have been disfavorable to both, yet even then DFSVX outperformed in your example.
TheDoctor91
Posts: 75
Joined: Thu Feb 25, 2021 12:43 pm

Re: Small Cap Value heads Rejoice !!!

Post by TheDoctor91 »

One of the more interest parts of SCV is the advantages in rolling returns and more diversification benefits as well.
countmein
Posts: 564
Joined: Fri Dec 06, 2013 9:10 pm

Re: Small Cap Value heads Rejoice !!!

Post by countmein »

muffins14 wrote: Wed Jul 21, 2021 5:57 am
countmein wrote: Tue Jul 20, 2021 7:46 pm
muffins14 wrote: Tue Jul 20, 2021 7:45 am
countmein wrote: Mon Jul 19, 2021 10:19 pm I'm going back to 1986 using the factor statistics tool in PV. It goes back to 1960-something (where the premium used to be). It's useful to look back as far as we can. When you do, it's clear that a premium used to exist. Then, for whatever combination of reasons, it stopped working.
This is a non-quantitative and also incorrect statement. There was an academic paper on this measurement and the premium is statistically consistent with the original measurement. (Yes, there is variance around the point estimates)
Quantitatively, the value "premium" was 0.17% in the 35 years to present using DFSVX as a simulant. I get that Nathan thinks that means "exists" and I think it means "doesn't exist" (and that's fine) but the 0.17 is what it is.
DFSVX Is not the same as the value premium, obviously. You don’t get to just make up your own definitions. It’s not even a value fund, it’s a small-value fund, and that matters because there is no small-value factor. There is a small factor and value factor, and the endpoints you chose have been disfavorable to both, yet even then DFSVX outperformed in your example.
The endpoints aren't arbitrary. They come from the question of HmL and SmB persistence, namely: how long has it been since we've seen a small value premium? If you look back 5 years, no premium. 10 years, still none. 15 years? nope. Et cetera until you get past 35 years. That was the exercise.

DFSVX is simply a way to translate historical long-short factor premia into real-world terms. i.e. if you had pure factor premia of X from 1986 to present (1% HmL, 0% SmB), how much of that can/could an investor actually capture? To answer that you have to assume some fund factor exposures. DFSVX seems like the most reasonable proxy, but you could also look at IJS or perhaps some others.
countmein
Posts: 564
Joined: Fri Dec 06, 2013 9:10 pm

Re: Small Cap Value heads Rejoice !!!

Post by countmein »

Nathan Drake wrote: Wed Jul 21, 2021 12:33 am
countmein wrote: Tue Jul 20, 2021 7:46 pm
muffins14 wrote: Tue Jul 20, 2021 7:45 am
countmein wrote: Mon Jul 19, 2021 10:19 pm I'm going back to 1986 using the factor statistics tool in PV. It goes back to 1960-something (where the premium used to be). It's useful to look back as far as we can. When you do, it's clear that a premium used to exist. Then, for whatever combination of reasons, it stopped working.
This is a non-quantitative and also incorrect statement. There was an academic paper on this measurement and the premium is statistically consistent with the original measurement. (Yes, there is variance around the point estimates)
Quantitatively, the value "premium" was 0.17% in the 35 years to present using DFSVX as a simulant. I get that Nathan thinks that means "exists" and I think it means "doesn't exist" (and that's fine) but the 0.17 is what it is.
I can find periods nearly as long where bonds beat stocks dramatically.

Guess the equity premium doesn’t exist

viewtopic.php?t=352394
You're confusing the risk free rate with bonds. The thread you link to says it right there: "both stocks and bonds are risk assets, capable of outperforming or underperforming over any human time horizon." I think the longest Rm-Rf has gone in the red is about 13 years or so. Still pretty scary.
muffins14
Posts: 633
Joined: Wed Oct 26, 2016 4:14 am

Re: Small Cap Value heads Rejoice !!!

Post by muffins14 »

countmein wrote: Thu Jul 22, 2021 12:06 am
muffins14 wrote: Wed Jul 21, 2021 5:57 am
countmein wrote: Tue Jul 20, 2021 7:46 pm
muffins14 wrote: Tue Jul 20, 2021 7:45 am
countmein wrote: Mon Jul 19, 2021 10:19 pm I'm going back to 1986 using the factor statistics tool in PV. It goes back to 1960-something (where the premium used to be). It's useful to look back as far as we can. When you do, it's clear that a premium used to exist. Then, for whatever combination of reasons, it stopped working.
This is a non-quantitative and also incorrect statement. There was an academic paper on this measurement and the premium is statistically consistent with the original measurement. (Yes, there is variance around the point estimates)
Quantitatively, the value "premium" was 0.17% in the 35 years to present using DFSVX as a simulant. I get that Nathan thinks that means "exists" and I think it means "doesn't exist" (and that's fine) but the 0.17 is what it is.
DFSVX Is not the same as the value premium, obviously. You don’t get to just make up your own definitions. It’s not even a value fund, it’s a small-value fund, and that matters because there is no small-value factor. There is a small factor and value factor, and the endpoints you chose have been disfavorable to both, yet even then DFSVX outperformed in your example.
The endpoints aren't arbitrary. They come from the question of HmL and SmB persistence, namely: how long has it been since we've seen a small value premium? If you look back 5 years, no premium. 10 years, still none. 15 years? nope. Et cetera until you get past 35 years. That was the exercise.

DFSVX is simply a way to translate historical long-short factor premia into real-world terms. i.e. if you had pure factor premia of X from 1986 to present (1% HmL, 0% SmB), how much of that can/could an investor actually capture? To answer that you have to assume some fund factor exposures. DFSVX seems like the most reasonable proxy, but you could also look at IJS or perhaps some others.
You’ve chosen to ignore my comments and just restate your point, so OK
YRT70
Posts: 1055
Joined: Sat Apr 27, 2019 8:51 am

Re: Small Cap Value heads Rejoice !!!

Post by YRT70 »

countmein wrote: Tue Jul 20, 2021 7:46 pm Quantitatively, the value "premium" was 0.17% in the 35 years to present using DFSVX as a simulant.
Two small things: DFSVX started in 1993 so I think we're talking about 28 years or so. According to PV the outperformance vs. vtsmx has been 0.94%, although that can change any day of course.

https://www.portfoliovisualizer.com/bac ... ion2_2=100
countmein
Posts: 564
Joined: Fri Dec 06, 2013 9:10 pm

Re: Small Cap Value heads Rejoice !!!

Post by countmein »

YRT70 wrote: Thu Jul 22, 2021 8:03 am
countmein wrote: Tue Jul 20, 2021 7:46 pm Quantitatively, the value "premium" was 0.17% in the 35 years to present using DFSVX as a simulant.
Two small things: DFSVX started in 1993 so I think we're talking about 28 years or so. According to PV the outperformance vs. vtsmx has been 0.94%, although that can change any day of course.

https://www.portfoliovisualizer.com/bac ... ion2_2=100
Sorry I haven't been clearer. I'm not looking at DFSVX's performance at all. In order to reach further back in history, I'm looking at raw long-short factor performance, found here

The point of going further back is to try to get a sense of when these two premia started to dry up, and to ponder whether the premium is still considered "existing" if it can go 35 years yielding next to nothing (0.17%).

On that factor statistics page, you can see that the raw factor performance for HmL and SmB using the AQR model are 0.35% and -0.07%, respectively. (Alternatively, 1.13% and 0.11% using the FF model)

Translating raw factor returns into would-be real world returns requires accounting for the partial factor exposure captured by funds (e.g. DFSVX captures 66% of HmL) plus management fees and other frictions (negative alpha) which I have waived (e.g. assumed perfect trading efficiency).

Since SmB is so close to zero, I dropped it from the calculation.

If you want to look at performance of funds, note that DFLVX lags VTSMX since inception (march 1993), thus a negative real-world value premium. FF HmL was +0.56% over the period. So even if the raw HmL premium is able to hang on to +0.5% going forward, can your fund capture it? DFSVX did a better job of keeping the negative alpha at bay. How will AVUV perform in that regard? We don't know, it's a gamble (with no upside).

It's nice to hear AVUV is picking up QmJ exposure because I wouldn't want to rely on it capturing what's left of HmL and SmB with DFA-like efficiency.
User avatar
Robert T
Posts: 2761
Joined: Tue Feb 27, 2007 9:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Re: Small Cap Value heads Rejoice !!!

Post by Robert T »

.
Unpacking the Fama-French HML factor over the 05/1986 - 05/2021 timeframe gives the following:

FF-HML = 1.13% annualized, as per the Portfolio Visualizer link

FF Small Value component of FF-HML = 12.9% annualized
FF Market = 11.0% annualized
Difference = 1.9%

Implied growth of $1 from start of 05/1986
FF Small Value = $70
FF Market = $39

And the FF-HML in "International ex US" and "Emerging markets" on the Portfolio Visualizer website, for the longest period they have data, give greater than zero numbers.

Robert
.
YRT70
Posts: 1055
Joined: Sat Apr 27, 2019 8:51 am

Re: Small Cap Value heads Rejoice !!!

Post by YRT70 »

countmein wrote: Thu Jul 22, 2021 11:26 am Sorry I haven't been clearer. I'm not looking at DFSVX's performance at all.
I see. My bad, I hadn't read the whole discussion.
countmein
Posts: 564
Joined: Fri Dec 06, 2013 9:10 pm

Re: Small Cap Value heads Rejoice !!!

Post by countmein »

Robert T wrote: Thu Jul 22, 2021 5:11 pm .
Unpacking the Fama-French HML factor over the 05/1986 - 05/2021 timeframe gives the following:

FF-HML = 1.13% annualized, as per the Portfolio Visualizer link

FF Small Value component of FF-HML = 12.9% annualized
FF Market = 11.0% annualized
Difference = 1.9%

Implied growth of $1 from start of 05/1986
FF Small Value = $70
FF Market = $39

And the FF-HML in "International ex US" and "Emerging markets" on the Portfolio Visualizer website, for the longest period they have data, give greater than zero numbers.

Robert
.
Interesting. So you could say there are two HmLs-- a large (HmL-L) and a small (HmL-s). The former is dead, the latter is hanging on. And if you target your portfolio to HmL and not specifically to HmL-s, you will be missing out to some degree.

Is it also true that the large and small halves of the other factors have significantly different returns?

Do factor returns change monotonically through the deciles? I could see the answer being 'yes' for value and quality but 'no' for mom and low vol (imagining a lull in midcap returns for those two).
muffins14
Posts: 633
Joined: Wed Oct 26, 2016 4:14 am

Re: Small Cap Value heads Rejoice !!!

Post by muffins14 »

countmein wrote: Sat Jul 24, 2021 4:39 pm
Robert T wrote: Thu Jul 22, 2021 5:11 pm .
Unpacking the Fama-French HML factor over the 05/1986 - 05/2021 timeframe gives the following:

FF-HML = 1.13% annualized, as per the Portfolio Visualizer link

FF Small Value component of FF-HML = 12.9% annualized
FF Market = 11.0% annualized
Difference = 1.9%

Implied growth of $1 from start of 05/1986
FF Small Value = $70
FF Market = $39

And the FF-HML in "International ex US" and "Emerging markets" on the Portfolio Visualizer website, for the longest period they have data, give greater than zero numbers.

Robert
.
Interesting. So you could say there are two HmLs-- a large (HmL-L) and a small (HmL-s). The former is dead, the latter is hanging on. And if you target your portfolio to HmL and not specifically to HmL-s, you will be missing out to some degree.

Is it also true that the large and small halves of the other factors have significantly different returns?

Do factor returns change monotonically through the deciles? I could see the answer being 'yes' for value and quality but 'no' for mom and low vol (imagining a lull in midcap returns for those two).
You’re again making inferences to suit your confirmation bias.

He wrote: HmL = 1.13%, small part of HmL= 1.9%.

How could the large part be zero (I.e.dead) if the average of that and the small part is 1.13%? It can’t.

If it were zero, then the overall HmL annualized value would have been 0.95%, but it was 1.13%.
garlandwhizzer
Posts: 3215
Joined: Fri Aug 06, 2010 3:42 pm

Re: Small Cap Value heads Rejoice !!!

Post by garlandwhizzer »

The premiums the models produce are a joke unless real factor funds can capture those premiums. The case for capture after cost in real funds has not been impressive since the inception of factor funds. The market has changed (professionally dominated, algorithms, essentially all investors know about factors), the economy has changed (tech dominated information economy instead of manufacturing/ bricks and mortar). Is it reasonable to expect long term outperformance going forward of SCV when P/B is outmoded and there is no agreement of exactly how to properly measure value.

IMO there is considerably more uncertainty about future factor returns of real funds than the models may suggest. The models are designed to show positive premiums, magnifying positives and totally neglecting negatives. Real funds have to deal with real markets and real economies which are vastly different than they were 1929 - 1992 when most stock buyers were inexperienced amateurs, the economy was based on manufacturing/bricks and mortar in which P/B had clout as predictor of the future returns. There were many clearly exploitable overlooked companies in the SCV space that could be identified by simple parameters like P/B back then. Trying to pick the few big winners from among the vast legions of small struggling companies which are now heavily scrutinized by many SCV funds is quite simply not as easy or as reliable as it was 1929 - 1992.

Garland Whizzer
Nathan Drake
Posts: 1536
Joined: Mon Apr 11, 2011 12:28 am

Re: Small Cap Value heads Rejoice !!!

Post by Nathan Drake »

garlandwhizzer wrote: Sun Jul 25, 2021 3:02 pm The premiums the models produce are a joke unless real factor funds can capture those premiums. The case for capture after cost in real funds has not been impressive since the inception of factor funds. The market has changed (professionally dominated, algorithms, essentially all investors know about factors), the economy has changed (tech dominated information economy instead of manufacturing/ bricks and mortar). Is it reasonable to expect long term outperformance going forward of SCV when P/B is outmoded and there is no agreement of exactly how to properly measure value.

IMO there is considerably more uncertainty about future factor returns of real funds than the models may suggest. The models are designed to show positive premiums, magnifying positives and totally neglecting negatives. Real funds have to deal with real markets and real economies which are vastly different than they were 1929 - 1992 when most stock buyers were inexperienced amateurs, the economy was based on manufacturing/bricks and mortar in which P/B had clout as predictor of the future returns. There were many clearly exploitable overlooked companies in the SCV space that could be identified by simple parameters like P/B back then. Trying to pick the few big winners from among the vast legions of small struggling companies which are now heavily scrutinized by many SCV funds is quite simply not as easy or as reliable as it was 1929 - 1992.

Garland Whizzer
Technological change isn't something new. The factor premium has been alive and well throughout all of these technological shifts. Buying into many of the hot new coming industries has often led to poor long-term results as investors overpay for that growth. I would argue that even with an end-date in mind that is not favorable to SCV, the premium has been quite significant (1% annualized over 35 years is no joke). Shift the end date 10 years and you have an even larger premium.

Many SCV value funds don't use a purely P/B strategy, and it is generally agreed upon within the space that you have to account for various measures and also adjust your valuation models for certain industries.

If you don't believe in SCV, you also don't believe in market risk. So stocks should not be expected to perform better than bonds. Of course we all know that this is not true. It's a risk story - not a free lunch, so the concept of SCV makes complete sense and likely isn't going away anytime soon. If it does go away, then that means you won't see a huge discount in SCV going forward -- so you should see a huge rise in pricing models for these types of stocks.
countmein
Posts: 564
Joined: Fri Dec 06, 2013 9:10 pm

Re: Small Cap Value heads Rejoice !!!

Post by countmein »

muffins14 wrote: Sun Jul 25, 2021 10:35 am
countmein wrote: Sat Jul 24, 2021 4:39 pm
Robert T wrote: Thu Jul 22, 2021 5:11 pm .
Unpacking the Fama-French HML factor over the 05/1986 - 05/2021 timeframe gives the following:

FF-HML = 1.13% annualized, as per the Portfolio Visualizer link

FF Small Value component of FF-HML = 12.9% annualized
FF Market = 11.0% annualized
Difference = 1.9%

Implied growth of $1 from start of 05/1986
FF Small Value = $70
FF Market = $39

And the FF-HML in "International ex US" and "Emerging markets" on the Portfolio Visualizer website, for the longest period they have data, give greater than zero numbers.

Robert
.
Interesting. So you could say there are two HmLs-- a large (HmL-L) and a small (HmL-s). The former is dead, the latter is hanging on. And if you target your portfolio to HmL and not specifically to HmL-s, you will be missing out to some degree.

Is it also true that the large and small halves of the other factors have significantly different returns?

Do factor returns change monotonically through the deciles? I could see the answer being 'yes' for value and quality but 'no' for mom and low vol (imagining a lull in midcap returns for those two).
You’re again making inferences to suit your confirmation bias.

He wrote: HmL = 1.13%, small part of HmL= 1.9%.

How could the large part be zero (I.e.dead) if the average of that and the small part is 1.13%? It can’t.

If it were zero, then the overall HmL annualized value would have been 0.95%, but it was 1.13%.
Implies about 0.36% for HmL-L. Accounting for fund friction, that is zero (dead).
muffins14
Posts: 633
Joined: Wed Oct 26, 2016 4:14 am

Re: Small Cap Value heads Rejoice !!!

Post by muffins14 »

Zero is not 0.36. Would you rather pay zero in expenses or 0.36? 0.36 compounded over a lifetime is higher than 0 compounded
countmein
Posts: 564
Joined: Fri Dec 06, 2013 9:10 pm

Re: Small Cap Value heads Rejoice !!!

Post by countmein »

muffins14 wrote: Mon Jul 26, 2021 11:34 am Zero is not 0.36. Would you rather pay zero in expenses or 0.36? 0.36 compounded over a lifetime is higher than 0 compounded
0.36 is the implied raw factor return. In real funds you have to subtract out 1) trading cost (neg alpha), 2) management fee (neg alpha), 3) partial exposure to the factor.

For example, look at DFLVX since inception. It earned 0.64% a year in HmL-attributed returns but has a -1.03% alpha. So it lags the total market fund by 20-some bps. According to some recent posters, that's an earth-shattering difference which should make you never want to mess around with inefficient non-TSM funds. To me, it's about a tie.

Also don't forget, DFLVX investors have paid somewhere in the neighborhood of probably 1% a year in cap gains distributions. Not an issue for total market. ETFs save the day on that issue though.
Post Reply