How Much To Tilt To Small Value?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
larryswedroe
Posts: 15636
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: How Much To Tilt To Small Value?

Post by larryswedroe » Sat Jan 14, 2017 8:57 pm

hilink
If referring to AQR style premium, definitely not, and as I said it's not even an equity only fund with exposure to factors in bonds and commodities and currencies as well, across four factors of value, defensive (quality), MOM and carry
Larry

User avatar
GreatOdinsRaven
Posts: 527
Joined: Thu Apr 23, 2015 8:47 pm

Re: How Much To Tilt To Small Value?

Post by GreatOdinsRaven » Sat Jan 14, 2017 10:30 pm

chatbotte wrote: Just use the following script. R will download ETF returns from Yahoo, if available. All you need to do is put the correct ticker in the first line of the script, and manually download Fama/French Global 3 Factors. Save the CSV file in R's working directory, give it a name, such as "Global Factors" (see script), then open the CSV file in an editor to delete annual factor returns at the end of the file and the first six or seven rows containing file description, and put the following column names in the first cell of the first row, separated by commas: Date,Mkt.RF,SMB,HML,RF. That's it. If it doesn't work, please let me know, you may need to tweak it a bit, the CSV file in particular.

Code: Select all

library(tseries)
library(zoo)

#download data from yahoo
daily.close <- get.hist.quote("efv", quote="AdjClose", retclass="zoo")
daily.close <- na.locf(daily.close)
monthly.close <- aggregate(daily.close, as.yearmon, tail, 1)

#take logs of prices 
returns<-(exp(diff(log(monthly.close)))-1)*100

#global data from Ken French's library
ffdata<-read.csv("Global Factors.csv",header=TRUE, sep=",")
#US data from Ken French's library
#ffdata<-read.csv("US Factors.csv",header=TRUE,sep=",")

ffdata$Date<-ffdata$Date*100+1
ffdata$Date<-as.Date(as.character(ffdata$Date),format="%Y%m%d")
ffzoo<-zoo(ffdata[,2:5],as.yearmon(ffdata$Date))
all.data<-merge.zoo(ffzoo,returns,all=FALSE,retclass="data.frame")

names(all.data)<-c('Mkt.RF','SMB','HML','RF','Returns')
all.data$Returns<-all.data$Returns-all.data$RF
mod<-lm(Returns~Mkt.RF+SMB+HML,data=all.data)

summary(mod)
Chatbotte,
Thank you. I didn't have success. Downloaded the Fama/French data series and edited them with Excel, deleting the top 7 lines and adding the colum names. Saved on my desktop as Global Factors.csv. Then I tried importing the data with the following function:

Code: Select all

data1 <- read.csv(file.choose(), header=TRUE) 
This allowed me to select the file from my desktop. It correctly loaded the data set. After this point, though, I ran into problems.

Code: Select all

library(tseries)
Error in library(tseries) : there is no package called ‘tseries’
Any suggestions?

Thank you,
GOR
"The greatest enemies of the equity investor are expenses and emotions." -John C. Bogle, Little Book of Common Sense Investing. | | "Winter is coming." Lord Eddard Stark.

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: How Much To Tilt To Small Value?

Post by chatbotte » Sun Jan 15, 2017 7:14 am

GreatOdinsRaven,

You need to install the "tseries" package. Go to Packages -- Install Package(s)... and select the "tseries" package from the list. You may need to install the "zoo" package as well.

XBTC
Posts: 14
Joined: Sat Jan 21, 2017 7:50 am

Re: How Much To Tilt To Small Value?

Post by XBTC » Sun Feb 05, 2017 9:47 am

Very interesting discussion so far. There is one thing i don't completey understand though.

Provided the resulting factor loadings and standard deviation would be the same. Does it make a difference if the small value tilt is realized via

1) A single small value fund

or

2) Combination of large value fund and small cap blend fund

My probably naive thinking would be that with 1) one would not be diversified as well as one could miss a possible value premium occuring in the large caps right? Or are only the final size/value loadings of portfolio relevant for expected risk/return?

Theoretical
Posts: 1287
Joined: Tue Aug 19, 2014 10:09 pm

Re: How Much To Tilt To Small Value?

Post by Theoretical » Sun Feb 05, 2017 11:46 am

The value premium is a lot bigger in small caps than large caps, which is the main reason it's worth using blend for the larges and value for the smalls.

livesoft
Posts: 61498
Joined: Thu Mar 01, 2007 8:00 pm

Re: How Much To Tilt To Small Value?

Post by livesoft » Sun Feb 05, 2017 11:52 am

XBTC wrote:Very interesting discussion so far. There is one thing i don't completey understand though.

Provided the resulting factor loadings and standard deviation would be the same. Does it make a difference if the small value tilt is realized via

1) A single small value fund

or

2) Combination of large value fund and small cap blend fund

My probably naive thinking would be that with 1) one would not be diversified as well as one could miss a possible value premium occuring in the large caps right? Or are only the final size/value loadings of portfolio relevant for expected risk/return?
I don't know if it makes a difference, but I think it is simpler to get value-tilt with an additional small-cap value fund rather than with an additional large-cap value fund. The reason is that I want to own a Total Stock Market index fund anyways (whether US or foreign), so I will get the large-cap stocks in spades with such funds and not so much of the small-caps. Furthermore, it is very often easy to find a large-cap blend fund in a 401(k)/403(b)/457, but not a large-cap value fund.
Wiki This signature message sponsored by sscritic: Learn to fish.

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: How Much To Tilt To Small Value?

Post by chatbotte » Sun Feb 05, 2017 1:59 pm

Theoretical wrote:The value premium is a lot bigger in small caps than large caps, which is the main reason it's worth using blend for the larges and value for the smalls.
Theoretical,
Regressing the five high B/M portfolios out of the 25 Fama French US portfolios on the Fama French 3 factors yields the following HML loadings:

SMALL HiBM = 0.91
ME2 BM5 = 0.88
ME3 BM5 = 0.88
ME4 BM5 = 0.93
BIG HiBM = 0.99

User avatar
willthrill81
Posts: 4651
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: How Much To Tilt To Small Value?

Post by willthrill81 » Sun Feb 05, 2017 7:37 pm

"Start with the S&P 500:

The best 20-year period for that index was 1980 through 1999. Those years produced an annual return of 17.9%.
The worst 20-year period for the index was 1959 through 1978, when the annual return was only 6.5%.
The average return of the 61 20-year periods was 11.6%.
That's not bad, and not even awful over the worst 20-year period.

Now let's look at the small-cap value index:

The best 20 calendar years to have invested in small-cap value stocks were 1977 through 1996; the annual return was 24%.
The worst 20 years for this particular index were 1955 through 1974, when the annual return was 9%.
The average return of the 61 20-year periods was 16.7% — an improvement of 46% when compared with the S&P 500.
Here's one that really surprised me: Of these 61 periods, small-cap value stocks beat the S&P 500 60 times. If that's not consistency, I don't know what would qualify."
http://www.marketwatch.com/story/the-on ... 2014-06-25
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

freyj6
Posts: 206
Joined: Thu Jun 19, 2014 2:09 am

Re: How Much To Tilt To Small Value?

Post by freyj6 » Sun Feb 05, 2017 10:13 pm

willthrill81 wrote:"Start with the S&P 500:

Here's one that really surprised me: Of these 61 periods, small-cap value stocks beat the S&P 500 60 times. If that's not consistency, I don't know what would qualify."
http://www.marketwatch.com/story/the-on ... 2014-06-25
Interestingly, mid-cap value beat small-cap value in like 80% of 5-year periods for the last 45 years.

It sure hasn't in the last 12 months, but interesting nevertheless :)

User avatar
willthrill81
Posts: 4651
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: How Much To Tilt To Small Value?

Post by willthrill81 » Sun Feb 05, 2017 10:40 pm

freyj6 wrote:
willthrill81 wrote:"Start with the S&P 500:

Here's one that really surprised me: Of these 61 periods, small-cap value stocks beat the S&P 500 60 times. If that's not consistency, I don't know what would qualify."
http://www.marketwatch.com/story/the-on ... 2014-06-25
Interestingly, mid-cap value beat small-cap value in like 80% of 5-year periods for the last 45 years.

It sure hasn't in the last 12 months, but interesting nevertheless :)
Would you post a link to that? Thanks.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: How Much To Tilt To Small Value?

Post by chatbotte » Mon Feb 06, 2017 7:15 am

willthrill,

The value premium looks pretty much the same in all the high B/M portfolios (value portfolios) of the 25 Fama French portfolios formed on size and B/M. Hardly any difference in HML loadings due to changes in ME (market equity, or size). See my previous post. Any differences in portfolio returns must be due to different MKT loadings (market beta) and/or SMB loadings (size).

SV (small value) is a convenient vehicle, because it captures both SMB and HML. (Note SMB and HML are long/short, so long-only ETFs and index funds can only exploit the long side of the long/short strategies.) Moving from SV to MV (midcaps) reduces your exposure to SMB without increasing your HML exposure.

dcabler
Posts: 534
Joined: Wed Feb 19, 2014 11:30 am

Re: How Much To Tilt To Small Value?

Post by dcabler » Mon Feb 06, 2017 7:33 am

There was a thread a while back from siamond where he built some tell-tale charts. But you can do this yourself using the latest Simba's spreadsheet where you can compare various tilts against a plain vanilla TSM/TBM portfolio using tell-tale charts. One thing about small caps, when they outperform, they have tended to do so in bursts of 2 to 10 year periods. You'll see that between those periods of outperformance, you mostly see close to parity performance with some periods of underperformance. Those periods of parity/underperformance are what can lead to tracking error and abandoning your plan. During those times, it doesn't mean you're losing money necessarily, just that you're underperforming a more traditional, simpler portfolio. If you don't think you can tolerate that (financially or emotionally), then tilting might not be for you. You can of course take this idea all the way to 100% small value with a higher dose of bonds and you then have the "Larry Portfolio".

As for me, I'm heavily in midcaps with a tilt toward midcap value/small cap value. Midcap value has shown the same bursty behavior as small caps, just not as extreme on either the up or downside.

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: How Much To Tilt To Small Value?

Post by chatbotte » Mon Feb 06, 2017 7:46 am

dcabler wrote:As for me, I'm heavily in midcaps with a tilt toward midcap value/small cap value. Midcap value has shown the same bursty behavior as small caps, just not as extreme on either the up or downside.
dcabler,

Being heavily in midcaps AND a tilt toward midcap value/smallcap value makes no sense to me. IMO you should hold the market AND tilt, not hold part of the market (e.g. midcaps) and tilt.

XBTC
Posts: 14
Joined: Sat Jan 21, 2017 7:50 am

Re: How Much To Tilt To Small Value?

Post by XBTC » Mon Feb 06, 2017 6:27 pm

chatbotte wrote:willthrill,

The value premium looks pretty much the same in all the high B/M portfolios (value portfolios) of the 25 Fama French portfolios formed on size and B/M. Hardly any difference in HML loadings due to changes in ME (market equity, or size). See my previous post. Any differences in portfolio returns must be due to different MKT loadings (market beta) and/or SMB loadings (size).
Interesting that the HML loadings are so consistent over ME. However I was under the impression that the regressed factor loadings in the fama french model only express how much a portfolio was exposed to certain premiums, but not how high these premiums actually were. So if we look at a small and a large cap portfolio each with a hml loading of 0.5 each, that means they both were able to reap 50% percent of the value premium. But how big the permium was for small vs. large is another story I thought? Did I get that wrong?

@willthrill81: That are some interesting observations. Thanks alot for posting.

Theoretical
Posts: 1287
Joined: Tue Aug 19, 2014 10:09 pm

Re: How Much To Tilt To Small Value?

Post by Theoretical » Mon Feb 06, 2017 6:37 pm

XBTC wrote:
chatbotte wrote:willthrill,

The value premium looks pretty much the same in all the high B/M portfolios (value portfolios) of the 25 Fama French portfolios formed on size and B/M. Hardly any difference in HML loadings due to changes in ME (market equity, or size). See my previous post. Any differences in portfolio returns must be due to different MKT loadings (market beta) and/or SMB loadings (size).
Interesting that the HML loadings are so consistent over ME. However I was under the impression that the regressed factor loadings in the fama french model only express how much a portfolio was exposed to certain premiums, but not how high these premiums actually were. So if we look at a small and a large cap portfolio each with a hml loading of 0.5 each, that means they both were able to reap 50% percent of the value premium. But how big the permium was for small vs. large is another story I thought? Did I get that wrong?

@willthrill81: That are some interesting observations. Thanks alot for posting.
Yes, that's the difference. I believe the historical Large Value premium was something like 1.9% and the Small Value premium was something around 5.4%, which are quite different. HML is 1/2 of the Small Value-Small Growth added to the Large Value-Large Growth returns.

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: How Much To Tilt To Small Value?

Post by chatbotte » Mon Feb 06, 2017 8:15 pm

Theoretical wrote:
XBTC wrote:
chatbotte wrote:willthrill,

The value premium looks pretty much the same in all the high B/M portfolios (value portfolios) of the 25 Fama French portfolios formed on size and B/M. Hardly any difference in HML loadings due to changes in ME (market equity, or size). See my previous post. Any differences in portfolio returns must be due to different MKT loadings (market beta) and/or SMB loadings (size).
Interesting that the HML loadings are so consistent over ME. However I was under the impression that the regressed factor loadings in the fama french model only express how much a portfolio was exposed to certain premiums, but not how high these premiums actually were. So if we look at a small and a large cap portfolio each with a hml loading of 0.5 each, that means they both were able to reap 50% percent of the value premium. But how big the permium was for small vs. large is another story I thought?
XBTC,

No. HML, or value premium, is defined as: [1/2*(return on small value portfolio + return on large value portfolio) - 1/2*(return on small growth portfolio + return on large growth portfolio)]. So it's a daily, monthly etc. return on a strategy that's short average growth long average value. That's your HML return. Now, HML daily, monthly etc. returns aren't the same, because HML return is a random variable. You might want to visit Ken French's website if you need more info on how the factor is constructed.

By doing a regression, you basically try to determine how much of your portfolio's variance is due to your predictor (e.g. HML). I regressed five value portfolios ranked by size, and obtained HML regression coefficients (loadings, or betas) of 0.91, 0.88 and so on.

That means if your portfolio has HML beta of 0.91 and HML returns 5 percent in October, you expect your portfolio to return 0.91 * 5 = 4.55 percent. If HML returns -10 percent in November, you expect your portfolio to return 0.91 * (-10) = -9.1 percent. And so on.

XBTC
Posts: 14
Joined: Sat Jan 21, 2017 7:50 am

Re: How Much To Tilt To Small Value?

Post by XBTC » Tue Feb 07, 2017 7:05 am

chatbotte wrote:
No. HML, or value premium, is defined as: [1/2*(return on small value portfolio + return on large value portfolio) - 1/2*(return on small growth portfolio + return on large growth portfolio)]. So it's a daily, monthly etc. return on a strategy that's short average growth long average value. That's your HML return. Now, HML daily, monthly etc. returns aren't the same, because HML return is a random variable. You might want to visit Ken French's website if you need more info on how the factor is constructed.

By doing a regression, you basically try to determine how much of your portfolio's variance is due to your predictor (e.g. HML). I regressed five value portfolios ranked by size, and obtained HML regression coefficients (loadings, or betas) of 0.91, 0.88 and so on.

That means if your portfolio has HML beta of 0.91 and HML returns 5 percent in October, you expect your portfolio to return 0.91 * 5 = 4.55 percent. If HML returns -10 percent in November, you expect your portfolio to return 0.91 * (-10) = -9.1 percent. And so on.
Thanks for the great explanation and the link to the original defintion chatbotte. Just to clarify: This still does not contradict my conclusion, that the HML beta/coefficient does not say anything about the size of HML / the available premium itself right?

When you originally posted:
chatbotte wrote:The value premium looks pretty much the same in all the high B/M portfolios (value portfolios) of the 25 Fama French portfolios formed on size and B/M. Hardly any difference in HML loadings due to changes in ME (market equity, or size). See my previous post. Any differences in portfolio returns must be due to different MKT loadings (market beta) and/or SMB loadings (size).
I thought you wanted to make a point that it doesn't matter from which ME you get your value exposure from. But although the regressed HML beta / coefficient was close between these porfolios, the HML premium was quite different historically it seems. Or how did Swedroe & Co arrive at the conclusion that the value premium was strongest in small caps then?

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: How Much To Tilt To Small Value?

Post by chatbotte » Tue Feb 07, 2017 8:11 am

XBTC wrote: Thanks for the great explanation and the link to the original defintion chatbotte. Just to clarify: This still does not contradict my conclusion, that the HML beta/coefficient does not say anything about the size of HML / the available premium itself right?
Right. HML = return. The coefficient tells you how much your asset return depends on a unit change in your predictor (HML). So if HML goes up 1 percent, the "response" in the asset's return is 0.91 percent (plus random error).

Regressing e.g. the most extreme small value portfolio (of the 25 Fama French portfolios formed on size and BE/ME) on MKT, SMB and HML yields the following result:

Code: Select all

Residuals:
     Min       1Q   Median       3Q      Max 
-12.2831  -1.1034  -0.0997   0.9073  22.3722 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept)  0.08302    0.07049   1.178    0.239    
Mkt.RF       0.99064    0.01399  70.830   <2e-16 ***
SMB          1.31534    0.02294  57.337   <2e-16 ***
HML          0.90969    0.02034  44.722   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 2.291 on 1076 degrees of freedom
Multiple R-squared:  0.9399,    Adjusted R-squared:  0.9397 
F-statistic:  5608 on 3 and 1076 DF,  p-value: < 2.2e-16
Your asset's return each month equals (first column): 0.08302 + Mkt.RF*0.99064 + SMB*1.31534 + HML*0.90969 + residual (random noise).

So each of MKT, SMB and HML is a variable return on a portfolio (or set of portfolios), and you're trying to explain your asset return based on those three variable returns.
XBTC wrote: I thought you wanted to make a point that it doesn't matter from which ME you get your value exposure from. But although the regressed HML beta / coefficient was close between these porfolios, the HML premium was quite different historically it seems.
HML, or value premium, is defined as: [1/2*(return on small value portfolio + return on large value portfolio) - 1/2*(return on small growth portfolio + return on large growth portfolio)]. So it's a unique number. No two different HML returns over the same period. HML is a random variable, so you basically get a different HML return each month, but if you regress LCV and SCV on MKT, SMB and HML and obtain the same coefficients on HML, then each of LCV and SCV returns must respond roughly the same to a unit change in HML.

Look at the following regression of the most extreme LARGE value portfolio (FF 25 again), compare the regression coefficients to those in the above regression, and see what's different (first column again):

Code: Select all

Residuals:
     Min       1Q   Median       3Q      Max 
-23.8327  -1.5044  -0.0599   1.4083  28.4702 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept) -0.18341    0.10786  -1.700 0.089335 .  
Mkt.RF       1.19183    0.02140  55.692  < 2e-16 ***
SMB         -0.13669    0.03510  -3.894 0.000105 ***
HML          0.99344    0.03112  31.918  < 2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 3.506 on 1076 degrees of freedom
Multiple R-squared:  0.8338,    Adjusted R-squared:  0.8333 
F-statistic:  1799 on 3 and 1076 DF,  p-value: < 2.2e-16

XBTC
Posts: 14
Joined: Sat Jan 21, 2017 7:50 am

Re: How Much To Tilt To Small Value?

Post by XBTC » Tue Feb 07, 2017 3:51 pm

Thank you very very much for taking the time to write this awesome explanation chatbotte. Finally everything understood!:-)

Now that I understood why HML is roughly the same over different ME I still wonder why many people refer to a higher value premium in small caps. Maybe the simple reason is that most SC value funds on the market have bigger HML beta/coefficients than LC value funds. So in practice it's just easier to reap the value premium with SC funds.

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: How Much To Tilt To Small Value?

Post by chatbotte » Wed Feb 08, 2017 7:07 am

XBTC wrote:Thank you very very much for taking the time to write this awesome explanation chatbotte. Finally everything understood!:-)

Now that I understood why HML is roughly the same over different ME I still wonder why many people refer to a higher value premium in small caps. Maybe the simple reason is that most SC value funds on the market have bigger HML beta/coefficients than LC value funds. So in practice it's just easier to reap the value premium with SC funds.
XBTC,

Not sure.

HML loadings (= how much of HML return "is" in each portfolio):

Code: Select all

0.43; 0.22; 0.51; 0.56; 0.91 (from left to right: small low BM or growth --- small high BM or value)
-0.22; 0.14; 0.35; 0.56; 0.88
-0.23; 0.05; 0.31; 0.55; 0.88
-0.36; 0.09; 0.36; 0.58; 0.93
-0.26; 0.02; 0.34; 0.64; 0.99 (big growth --- big value)
Mean BE/ME ratios (from 1926 to 2016; sorry about the messy "matrix"):

Code: Select all

SMALL.LoBM    ME1.BM2    ME1.BM3    ME1.BM4 SMALL.HiBM    ME2.BM1    ME2.BM2    ME2.BM3 
 0.3960440  0.7131868  1.0184615  1.4319780  4.1805495  0.3815385  0.6713187  0.9318681 
   ME2.BM4    ME2.BM5    ME3.BM1    ME3.BM2    ME3.BM3    ME3.BM4    ME3.BM5    ME4.BM1 
 1.3275824  3.1687912  0.3502198  0.6372527  0.9203297  1.3207692  2.7648352  0.3358242 
   ME4.BM2    ME4.BM3    ME4.BM4    ME4.BM5   BIG.LoBM    ME5.BM2    ME5.BM3    ME5.BM4 
 0.6315385  0.9037363  1.3383516  2.8213187  0.3062637  0.6115385  0.8935165  1.2942857 
  BIG.HiBM 
 2.4661538 

XBTC
Posts: 14
Joined: Sat Jan 21, 2017 7:50 am

Re: How Much To Tilt To Small Value?

Post by XBTC » Wed Feb 08, 2017 3:34 pm

Thanks for posting the data. I am not sure what you are implying with it though? I mean sure the SMALL.HiBM has twice the mean BE/ME ratio than BIG.HiBM but what are the implications of that? Seems I have to study the work of Fama/French some more... :?

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: How Much To Tilt To Small Value?

Post by chatbotte » Thu Feb 09, 2017 5:15 am

XBTC,

On hindsight, I'm not sure either. :) I'm trying to figure out what it actually means that small caps might have a higher value premium (=HML). If you define the value premium (=HML) as the return on short low BM (averaged across size) long high BM (averaged across size) (i.e. subtract the return on low BM or growth stocks from the return on high BM or value stocks), then that statement doesn't make sense, unless you look at how much of that spread you actually get in small versus big.

larryswedroe
Posts: 15636
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: How Much To Tilt To Small Value?

Post by larryswedroe » Thu Feb 09, 2017 10:18 am

Here's the long term data using FF research indices 7/26-16. Note these are annualized returns, not annual premiums which is way factor returns/premiums are calculated.

LG 9.54 LV 12.06, So large value outperformed large growth by 2.52%, Note the monthly AVERAGE returns are 90 and 120bp

SG 8.56 and SV 14.92, So sv outperformed by 6.46%. The monthly AVERAGE returns are 97 and 148, much wider gap.

Larry

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: How Much To Tilt To Small Value?

Post by chatbotte » Thu Feb 09, 2017 11:01 am

larryswedroe wrote:Here's the long term data using FF research indices 7/26-16. Note these are annualized returns, not annual premiums which is way factor returns/premiums are calculated.

LG 9.54 LV 12.06, So large value outperformed large growth by 2.52%, Note the monthly AVERAGE returns are 90 and 120bp

SG 8.56 and SV 14.92, So sv outperformed by 6.46%. The monthly AVERAGE returns are 97 and 148, much wider gap.

Larry
What are the gaps due to? What are the exposures?

larryswedroe
Posts: 15636
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: How Much To Tilt To Small Value?

Post by larryswedroe » Thu Feb 09, 2017 11:28 am

The larger gap for small is due to the really absymal performance of small growth stocks with high investment and low profitability, those lottery tickets. Also logically should expect it to be larger simply because when you combine two economic risk factors, size and value, you get more economic risk so should have larger premium.

Larry

XBTC
Posts: 14
Joined: Sat Jan 21, 2017 7:50 am

Re: How Much To Tilt To Small Value?

Post by XBTC » Thu Feb 09, 2017 1:04 pm

larryswedroe wrote:The larger gap for small is due to the really absymal performance of small growth stocks with high investment and low profitability, those lottery tickets. Also logically should expect it to be larger simply because when you combine two economic risk factors, size and value, you get more economic risk so should have larger premium.

Larry
Makes sense. Thanks alot for the explanation.

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: How Much To Tilt To Small Value?

Post by chatbotte » Fri Feb 10, 2017 2:48 am

larryswedroe wrote:The larger gap for small is due to the really absymal performance of small growth stocks with high investment and low profitability, those lottery tickets. Also logically should expect it to be larger simply because when you combine two economic risk factors, size and value, you get more economic risk so should have larger premium.

Larry
Larry,

I don't understand your reasoning. Why would you expect a bigger gap in small stocks? SG and SV are the same size segment, aren't they? So the gap must be due to different SG and SV value exposures or different market betas.

Also, FF25 SV and LV portfolios seem to have the same HML exposures, and SV outperformance relative to LV is almost entirely due to SMB.

larryswedroe
Posts: 15636
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: How Much To Tilt To Small Value?

Post by larryswedroe » Fri Feb 10, 2017 10:59 am

chatbotte, already explained why and don't know how to explain further. It's simply that large growth doesn't have the negatives that small growth has. So the gap between large growth returns and large value will be much less than it is between small growth and small value (because a significant part of SG has awful returns). Now you can avoid those stocks by screening them out, don't have to go short. That is what DFA does and Bridgeway, in their small cap funds.
Larry

MrKnight
Posts: 266
Joined: Fri Dec 19, 2014 12:24 pm

Re: How Much To Tilt To Small Value?

Post by MrKnight » Fri Feb 10, 2017 11:34 am

larryswedroe wrote:chatbotte, already explained why and don't know how to explain further. It's simply that large growth doesn't have the negatives that small growth has. So the gap between large growth returns and large value will be much less than it is between small growth and small value (because a significant part of SG has awful returns). Now you can avoid those stocks by screening them out, don't have to go short. That is what DFA does and Bridgeway, in their small cap funds.
Larry
My thoughts are that inefficiency and behavioral investing is magnified in small cap stocks which leads to a larger value premium.

Large-cap stocks are subjected to boosted scrutiny from more parties. Additionally, the fact that large-cap stocks are higher capital better absorbs irrational individual investments.

Small-cap stocks are subjected to less scrutiny which tends to lead to less efficient market coupled with lower capital investments leading to greater impact from irrational individual investments.

In summary, I suspect that value premium is due to 3 these factors:
1. Greater risk demands greater returns (derived from efficient market hypothesis)
2. Less trading and invested capital results in decreased ability to absorb irrational investments (swings both ways)
3. Subject to less scrutiny by investors allowing mispricing

larryswedroe
Posts: 15636
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: How Much To Tilt To Small Value?

Post by larryswedroe » Fri Feb 10, 2017 3:53 pm

MrKnight
I don't think that's it. IMO it's the limits to arbitrage which makes it much more difficult to correct OVERvaluation than under. Under is easy to correct --you just buy. OVER is much harder as have to borrow and go short, which is much more difficult and risky and expensive. And institutions tend to avoid these overpriced stocks and they are the lenders of securities. So there is limited supply. So if large cap is overvalued then easier to correct the overvaluation, but not the case with small caps. We discuss this in Your Complete Guide to Factor Based Investing and cite the literature supporting that belief
Larry

TomCat96
Posts: 469
Joined: Sun Oct 18, 2015 12:18 pm

Re: How Much To Tilt To Small Value?

Post by TomCat96 » Fri Feb 10, 2017 4:32 pm

My AA is ~60% small cap 10% mid cap 30% Large Cap
My portfolio is 100% equities, 0% bonds.


Tilting makes sense if you are of a particular temperament and you have a long time frame remaining. It does not matter to me that the greater gains in small caps may or may not be matched by the greater volatility I am taking on. What matters to me is a reasonable chance small caps will outperform in the long run. Greater risk generally means greater reward.

I am not sure about individuals who have a bond allocation, but want to accept the greater risk from small value. Intuitively, it makes sense to me that if you wanted to take on greater risk, you could simply decrease your bond allocation and increase your diversified equities allocation.

In other words, why be 40% bonds 30% small cap value and 30% total market, when you could just be 20% bonds and 80% total market. Even if it is possible to find some local maxima of risk/reward in your portfolio calculations, that's just splitting hairs to me. It's a lot of complexity and number crunching without any clear benefit. I'm not saying it doesn't exist. I'm just saying, there's a simpler way to increase your gains.

Being already 100% equities, if you desire to further increase your gains at the expense of greater volatility, then tilt small cap. From my number crunching on portfolio visualizer, small caps do better 7/10 times. The additional gain may not be commensurate with the increase in volatility. But again, if you are in it for the long haul, that shouldn't matter to you.

To the extent I still have student loans, my equities allocation actually exceeds 100% theoretically. I would leverage myself further if I could. But other options of levering including deep in the money warrants, warrants on e-minis, sell puts and buying calls on e-minis all have their own complexities and frankly...inefficiencies (transactional costs)

Tilting small caps however has got to be the simplest and easiest way to increase your reward, at the expense of greater risk if you are already 100% equities.

MrKnight
Posts: 266
Joined: Fri Dec 19, 2014 12:24 pm

Re: How Much To Tilt To Small Value?

Post by MrKnight » Fri Feb 10, 2017 5:07 pm

larryswedroe wrote:MrKnight
I don't think that's it. IMO it's the limits to arbitrage which makes it much more difficult to correct OVERvaluation than under. Under is easy to correct --you just buy. OVER is much harder as have to borrow and go short, which is much more difficult and risky and expensive. And institutions tend to avoid these overpriced stocks and they are the lenders of securities. So there is limited supply. So if large cap is overvalued then easier to correct the overvaluation, but not the case with small caps. We discuss this in Your Complete Guide to Factor Based Investing and cite the literature supporting that belief
Larry
I like that explanation. It explains why small-cap value premium persists even if everyone is aware of it.

Growth stocks gets stuck at these overpriced valuations due to lacking market mechanisms to correct it promptly - sticky overpriced valuations.

larryswedroe
Posts: 15636
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: How Much To Tilt To Small Value?

Post by larryswedroe » Fri Feb 10, 2017 5:56 pm

MrKnight
You got it
Larry

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: How Much To Tilt To Small Value?

Post by chatbotte » Sat Feb 11, 2017 4:43 am

MrKnight wrote:
larryswedroe wrote:MrKnight
I don't think that's it. IMO it's the limits to arbitrage which makes it much more difficult to correct OVERvaluation than under. Under is easy to correct --you just buy. OVER is much harder as have to borrow and go short, which is much more difficult and risky and expensive. And institutions tend to avoid these overpriced stocks and they are the lenders of securities. So there is limited supply. So if large cap is overvalued then easier to correct the overvaluation, but not the case with small caps. We discuss this in Your Complete Guide to Factor Based Investing and cite the literature supporting that belief
Larry
I like that explanation. It explains why small-cap value premium persists even if everyone is aware of it.

Growth stocks gets stuck at these overpriced valuations due to lacking market mechanisms to correct it promptly - sticky overpriced valuations.
Larry & MrKnight

When you say "overvalued" and "overpriced", which risk-adjustment model are you talking about? What are they overvalued with respect to? What prices is the market trying to set in your view?

betablocker
Posts: 366
Joined: Mon Jan 11, 2016 1:26 pm

Re: How Much To Tilt To Small Value?

Post by betablocker » Sat Feb 11, 2017 10:57 am

chatbotte wrote:
MrKnight wrote:
larryswedroe wrote:MrKnight
I don't think that's it. IMO it's the limits to arbitrage which makes it much more difficult to correct OVERvaluation than under. Under is easy to correct --you just buy. OVER is much harder as have to borrow and go short, which is much more difficult and risky and expensive. And institutions tend to avoid these overpriced stocks and they are the lenders of securities. So there is limited supply. So if large cap is overvalued then easier to correct the overvaluation, but not the case with small caps. We discuss this in Your Complete Guide to Factor Based Investing and cite the literature supporting that belief
Larry
I like that explanation. It explains why small-cap value premium persists even if everyone is aware of it.

Growth stocks gets stuck at these overpriced valuations due to lacking market mechanisms to correct it promptly - sticky overpriced valuations.
Larry & MrKnight

When you say "overvalued" and "overpriced", which risk-adjustment model are you talking about? What are they overvalued with respect to? What prices is the market trying to set in your view?
Chatbotte,

I don't know the official answer but assume it's the historic returns of stocks with similar characteristics. Same evidence that all modern portfolio theory is based on. That evidence would show a low Sharpe ratio.

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: How Much To Tilt To Small Value?

Post by chatbotte » Sat Feb 11, 2017 11:28 am

betablocker wrote:
chatbotte wrote:
MrKnight wrote:
larryswedroe wrote:MrKnight
I don't think that's it. IMO it's the limits to arbitrage which makes it much more difficult to correct OVERvaluation than under. Under is easy to correct --you just buy. OVER is much harder as have to borrow and go short, which is much more difficult and risky and expensive. And institutions tend to avoid these overpriced stocks and they are the lenders of securities. So there is limited supply. So if large cap is overvalued then easier to correct the overvaluation, but not the case with small caps. We discuss this in Your Complete Guide to Factor Based Investing and cite the literature supporting that belief
Larry
I like that explanation. It explains why small-cap value premium persists even if everyone is aware of it.

Growth stocks gets stuck at these overpriced valuations due to lacking market mechanisms to correct it promptly - sticky overpriced valuations.
Larry & MrKnight

When you say "overvalued" and "overpriced", which risk-adjustment model are you talking about? What are they overvalued with respect to? What prices is the market trying to set in your view?
Chatbotte,

I don't know the official answer but assume it's the historic returns of stocks with similar characteristics.
betablocker,

You need to distinguish between market efficiency and a model of expected returns. If you say "overpriced" and "overvalued", do you mean your model is bad, because it fails to explain market prices despite the fact that you have reason to believe it's correct? So if market prices don't conform to e.g. CAPM, is it possible that a better model might describe market prices better (e.g. FF5)? Or do you observe an inefficient market that doesn't conform to your model (e.g. CAPM) due to some problems in the price setting mechanism (e.g. short sale restrictions), which prevent investors to set e.g. CAPM prices?

betablocker
Posts: 366
Joined: Mon Jan 11, 2016 1:26 pm

Re: How Much To Tilt To Small Value?

Post by betablocker » Sat Feb 11, 2017 12:30 pm

Chatbotte, I'm not sure why the two things are different. Couldn't the model be an explanation of the inefficiencies? CAPM would work if we didn't have people and market problems like limits to arbitrage so academics went and tried to build models that better accounted for people and market problems. But in the end I don't really care either way. Seems excessively academic to me.

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: How Much To Tilt To Small Value?

Post by chatbotte » Sat Feb 11, 2017 12:56 pm

betablocker wrote:Chatbotte, I'm not sure why the two things are different. Couldn't the model be an explanation of the inefficiencies? CAPM would work if we didn't have people and market problems like limits to arbitrage so academics went and tried to build models that better accounted for people and market problems. But in the end I don't really care either way. Seems excessively academic to me.
betablocker,

Do you tilt to e.g. SV? If you do, then you believe in market efficiency and don't believe in the CAPM.

betablocker
Posts: 366
Joined: Mon Jan 11, 2016 1:26 pm

Re: How Much To Tilt To Small Value?

Post by betablocker » Sat Feb 11, 2017 1:03 pm

That's a bit binary. I think the capm explains some things but not all things so it's not the best model available l. How does it follow that I believe in market efficiency because I own scv funds?

chatbotte
Posts: 295
Joined: Thu Jun 05, 2014 2:35 am

Re: How Much To Tilt To Small Value?

Post by chatbotte » Sun Feb 12, 2017 5:47 am

betablocker wrote:That's a bit binary. I think the capm explains some things but not all things so it's not the best model available l. How does it follow that I believe in market efficiency because I own scv funds?
betablocker,

Sorry, you're right. It doesn't follow. My apologies. See? It's a minefield. :) You could buy SCV because you believe they're an arbitrage opportunity with respect to your model of returns, i.e. you expect depressed SCV prices to go up at some point, and make a profit. Not sure why that arbitrage opportunity should persist over the long haul though. Any ideas?

larryswedroe
Posts: 15636
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: How Much To Tilt To Small Value?

Post by larryswedroe » Sun Feb 12, 2017 9:16 am

You could buy SCV because you believe they're an arbitrage opportunity with respect to your model of returns, i.e. you expect depressed SCV prices to go up at some point, and make a profit. Not sure why that arbitrage opportunity should persist over the long haul though. Any ideas?
First, arbitrage implies riskless. So that means you believe that small value premium is all inefficiency/mispricings. While there are many who side on the behavioral story for value that isn't the case for size, which is then a risk premium and not subject to arbitrage. IMO there is sufficiently strong evidence that value is at least partly a risk story. I've cited more than a dozen papers showing clear and simple risk-based explanations. Thus, if you believe that it's not an arbitrage opportunity. IMO it's partly a free lunch, especially in small value, because of limits to arbitrage and costs and fear of shorting. But that's not an arbitrage opportunity.
However if you believe in risk explanation than you have the very same EXPECTATION that the LIKELIHOOD is that you will be rewarded, but not certainty. And in my book we provide the explanations, citing the literature for why one would consider the premiums LIKELY to persist in future.

Investing isn't about certainty, but about putting odds on your side. And certainly since I wrote my first book those that have invested based on the principles have been rewarded for their belief in the likelihood. And it's almost 20 years now. So it's not just some back tested result.

As I showed the persistence of the value premium is almost as strong as it is for market beta, regardless of the time horizon, whether short or long, and for small value relative to the market it's basically a draw.

Best wishes
Larry

groovy9
Posts: 252
Joined: Wed Mar 14, 2007 11:17 am

Re: How Much To Tilt To Small Value?

Post by groovy9 » Sun Feb 12, 2017 11:15 pm

larryswedroe wrote: Investing isn't about certainty, but about putting odds on your side.
I feel like this should be a sticky post in all caps. So many arguments around here amount to "yeah, but it might not." Yes, and my pocket aces will occasionally lose, but I'm still betting the crap out of them 100% of the time. :)

MrKnight
Posts: 266
Joined: Fri Dec 19, 2014 12:24 pm

Re: How Much To Tilt To Small Value?

Post by MrKnight » Mon Feb 13, 2017 1:10 am

groovy9 wrote:
larryswedroe wrote: Investing isn't about certainty, but about putting odds on your side.
I feel like this should be a sticky post in all caps. So many arguments around here amount to "yeah, but it might not." Yes, and my pocket aces will occasionally lose, but I'm still betting the crap out of them 100% of the time. :)
Agreed. It is something that appears to be often forgotten here.

You can draw an analogy to counting cards. Counting cards does not guarantee you win every hand, but it pushes the odds into your favor so that you will come out ahead if you play long enough.

larryswedroe
Posts: 15636
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: How Much To Tilt To Small Value?

Post by larryswedroe » Mon Feb 13, 2017 9:31 am

Like I said, NOTHING in investing is certainty. It's all about taking risks you can accept and putting the odds in your favor. Want certainty? Buy only TIPS. TSM is simply a huge bet on market beta, which certainly isn't guaranteed to matter what your horizon. And like I said, if you do believe markets are efficient then all factors should have the same Sharpe ratio, or risk-adjusted returns, otherwise capital would move to the one with the higher SR. And if factors have low to negative correlation than they reduce portfolio risk if you add them, creating a more efficient portfolio
Larry

KlangFool
Posts: 9191
Joined: Sat Oct 11, 2008 12:35 pm

Re: How Much To Tilt To Small Value?

Post by KlangFool » Mon Feb 13, 2017 10:12 am

larryswedroe wrote:Like I said, NOTHING in investing is certainty. It's all about taking risks you can accept and putting the odds in your favor. Want certainty? Buy only TIPS. TSM is simply a huge bet on market beta, which certainly isn't guaranteed to matter what your horizon. And like I said, if you do believe markets are efficient then all factors should have the same Sharpe ratio, or risk-adjusted returns, otherwise capital would move to the one with the higher SR. And if factors have low to negative correlation than they reduce portfolio risk if you add them, creating a more efficient portfolio
Larry
Larry,

I understand and agree with your point. But, to carry this into the extreme, a person may invest 100/0 into small cap value only. Does this make sense? I do not believe you are advocating this.

KlangFool

MrKnight
Posts: 266
Joined: Fri Dec 19, 2014 12:24 pm

Re: How Much To Tilt To Small Value?

Post by MrKnight » Mon Feb 13, 2017 10:56 am

I'm at 100% equity allocation, and close to 50% vanguard small cap value... but I am also 30, so I have a long time horizon.

User avatar
willthrill81
Posts: 4651
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: How Much To Tilt To Small Value?

Post by willthrill81 » Mon Feb 13, 2017 11:02 am

MrKnight wrote:I'm at 100% equity allocation, and close to 50% vanguard small cap value... but I am also 30, so I have a long time horizon.
Be careful. There are many here who think it extremely unwise to be 100% equities despite its having higher historic returns. And then they'll say if you own your home, or have home equity, or have an EF, etc. that you're not really 100% equities.

I'm 35 and am 100% equities as well when it comes to my retirement assets, and I'm tilted toward small and value. What do I care about volatility when I'm 20-25 years away from needing the funds, more if necessary?
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

larryswedroe
Posts: 15636
Joined: Thu Feb 22, 2007 8:28 am
Location: St Louis MO

Re: How Much To Tilt To Small Value?

Post by larryswedroe » Mon Feb 13, 2017 11:59 am

Klangfool
In fact that is basically what I am invested in, and are many advisors in our firm, and about 10-15% of our clients--they are basically small value, but internationally diversified.

Now some do so to increase expected returns, so they are even 100% equities, and some are like me using the higher tilts to lower beta and cut tail risks.

Both are perfectly reasonable IMO, depending on their circumstances.

Remember, they are certainly well diversified by individual securities, minimizing idiosyncratic risks, owning something like 4-5k stocks. They are globally diversified, typically close to market cap, across US, developed and EM. And they are certainly more diversified across unique sources of risk. Now they have very large tracking error risk. But they accept that as the "price" of either the higher expected returns or the cutting of the tail risk.

I hope that is helpful
Larry

KlangFool
Posts: 9191
Joined: Sat Oct 11, 2008 12:35 pm

Re: How Much To Tilt To Small Value?

Post by KlangFool » Mon Feb 13, 2017 12:01 pm

larryswedroe wrote:Klangfool
In fact that is basically what I am invested in, and are many advisors in our firm, and about 10-15% of our clients--they are basically small value, but internationally diversified.

Now some do so to increase expected returns, so they are even 100% equities, and some are like me using the higher tilts to lower beta and cut tail risks.

Both are perfectly reasonable IMO, depending on their circumstances.

Remember, they are certainly well diversified by individual securities, minimizing idiosyncratic risks, owning something like 4-5k stocks. They are globally diversified, typically close to market cap, across US, developed and EM. And they are certainly more diversified across unique sources of risk. Now they have very large tracking error risk. But they accept that as the "price" of either the higher expected returns or the cutting of the tail risk.

I hope that is helpful
Larry
Larry,

Thanks.

KlangFool

User avatar
willthrill81
Posts: 4651
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: How Much To Tilt To Small Value?

Post by willthrill81 » Mon Feb 13, 2017 12:12 pm

larryswedroe wrote:Klangfool
In fact that is basically what I am invested in, and are many advisors in our firm, and about 10-15% of our clients--they are basically small value, but internationally diversified.

Now some do so to increase expected returns, so they are even 100% equities, and some are like me using the higher tilts to lower beta and cut tail risks.

Both are perfectly reasonable IMO, depending on their circumstances.

Remember, they are certainly well diversified by individual securities, minimizing idiosyncratic risks, owning something like 4-5k stocks. They are globally diversified, typically close to market cap, across US, developed and EM. And they are certainly more diversified across unique sources of risk. Now they have very large tracking error risk. But they accept that as the "price" of either the higher expected returns or the cutting of the tail risk.

I hope that is helpful
Larry
:beer
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

Post Reply