Why don't you factor tilt?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Nathan Drake
Posts: 6201
Joined: Mon Apr 11, 2011 12:28 am

Re: Why don't you factor tilt?

Post by Nathan Drake »

burritoLover wrote: Mon Oct 03, 2022 7:34 am Have you all checked out the latest podcast from Rational Reminder? Couple of academic guests that present an argument that only CAPM matters and that factors are not a risk story:
https://rationalreminder.ca/podcast/220
Jules Van Binsbergen wrote:To summarize it very simply, we just ask the question, do investors view Fama-French factors as outperformance or as risk? And the answer is they view it as outperformance, not as risk. And so it's an alpha. They view it ... They count it as alpha, not as a risk premium that you can ... Because ... Another way of saying it is a risk ... You shouldn't reward a manager with more money if all they did was take more risk according to that risk premium. You should only reward them if it was outperformance. And so clearly they count it as outperformance.

Now, in the beginning there was some criticism, and people said, "Oh, but this is just because it's unsophisticated investors." But then there were some people that did it for hedge funds. Where I think people generally agreed that the investors are quite more sophisticated and they found the same result there. And then there were even some people that tried to do it for real investment decisions inside firms to figure out what firms themselves uses the risk model. And again, the same ... No, I say that quite incorrectly. How firms decide to repurchase their stock or issue more stock. And so that's also an investment decision that's made inside the firm by managers. And again, the same result shows up. It's again going to be the CAPM. CAPM seems to be quite a dominant model that seems to be showing up all over the place.
There are risk based measures that you can quantitatively track, though. So their theory doesn’t seem consistent.

Even if it’s not all (or mostly) risk based, I don’t believe the behavioral aspects are able to be arbitraged away
20% VOO | 20% VXUS | 20% AVUV | 20% AVDV | 20% AVES
abc132
Posts: 2411
Joined: Thu Oct 18, 2018 1:11 am

Re: Why don't you factor tilt?

Post by abc132 »

Apathizer wrote: Mon Oct 03, 2022 10:32 am Based on the data available the higher returns of the 4 non-market factors have occurred about 70-80% of the time. That seems pretty consistent. Rigorous analysis by multiple sources reduces the likely these results are attributable to data mining or random chance. It could be, but it's unlikely. Remember, we're talking about probability not certainty.
That would be expected on average if they carry more risk.

It is more complicated than simply finding something that outperforms more than 50% of the time, as that such outperformance can be attributed to additional risk which may or may not offer a benefit to a portfolio. We can dump bonds and our stocks will outperform bonds about 70% of the time. That alone does not make it a good idea to dump bonds.

The argument needs to be on getting something risk-free or at a better risk vs reward ratio and not on a simple performance basis.
Apathizer
Posts: 2507
Joined: Sun Sep 26, 2021 2:56 pm

Re: Why don't you factor tilt?

Post by Apathizer »

abc132 wrote: Mon Oct 03, 2022 12:08 pm
Apathizer wrote: Mon Oct 03, 2022 10:32 am Based on the data available the higher returns of the 4 non-market factors have occurred about 70-80% of the time. That seems pretty consistent. Rigorous analysis by multiple sources reduces the likely these results are attributable to data mining or random chance. It could be, but it's unlikely. Remember, we're talking about probability not certainty.
That would be expected on average if they carry more risk.

It is more complicated than simply finding something that outperforms more than 50% of the time, as that such outperformance can be attributed to additional risk which may or may not offer a benefit to a portfolio. We can dump bonds and our stocks will outperform bonds about 70% of the time. That alone does not make it a good idea to dump bonds.

The argument needs to be on getting something risk-free or at a better risk vs reward ratio and not on a simple performance basis.
I think we agree. To me it's about having more potential return sources than a cap weight TSM. If the market does poorly or underperforms TSM will as well. But if small value and other factors do well while TSM lags a factor slanted portfolio will mitigate this underperformance.

So to me it's about increasing the likelihood of better performance or at least consistent performance. But likelihood all five factors will underperform simultaneously is much lower than the likelihood any one of them will underperform at any given time.
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
User avatar
Taylor Larimore
Posts: 32839
Joined: Tue Feb 27, 2007 7:09 pm
Location: Miami FL

Re: Why don't you factor tilt?

Post by Taylor Larimore »

vineviz wrote: Unfortunately, mathematics couldn’t save Norstad from being incorrect (about TSM being an efficient portfolio).
vineviz:

I would appreciate learning why you think John Norstad, a prestigious mathematician at Northwestern University, is incorrect.

Thank you and best wishes
Taylor
Jack Bogle's Words of Wisdom: “It’s very difficult for any particular segment of the stock market to sustain superior performance. The watch word for our financial markets is ‘reversion to the mean.’”
"Simplicity is the master key to financial success." -- Jack Bogle
abc132
Posts: 2411
Joined: Thu Oct 18, 2018 1:11 am

Re: Why don't you factor tilt?

Post by abc132 »

Apathizer wrote: Mon Oct 03, 2022 1:14 pm
abc132 wrote: Mon Oct 03, 2022 12:08 pm
Apathizer wrote: Mon Oct 03, 2022 10:32 am Based on the data available the higher returns of the 4 non-market factors have occurred about 70-80% of the time. That seems pretty consistent. Rigorous analysis by multiple sources reduces the likely these results are attributable to data mining or random chance. It could be, but it's unlikely. Remember, we're talking about probability not certainty.
That would be expected on average if they carry more risk.

It is more complicated than simply finding something that outperforms more than 50% of the time, as that such outperformance can be attributed to additional risk which may or may not offer a benefit to a portfolio. We can dump bonds and our stocks will outperform bonds about 70% of the time. That alone does not make it a good idea to dump bonds.

The argument needs to be on getting something risk-free or at a better risk vs reward ratio and not on a simple performance basis.
I think we agree. To me it's about having more potential return sources than a cap weight TSM. If the market does poorly or underperforms TSM will as well. But if small value and other factors do well while TSM lags a factor slanted portfolio will mitigate this underperformance.
That makes sense until you say this...
Apathizer wrote: Mon Oct 03, 2022 1:14 pm So to me it's about increasing the likelihood of better performance or at least consistent performance. But likelihood all five factors will underperform simultaneously is much lower than the likelihood any one of them will underperform at any given time.
You are investing in market plus other factor risk - you are loading up on risk factors.

You don't need all 5 to go bad to underperform, nor is that comparison valid. You would see 1/2 of the standard deviation if you had factors operating in this manner, reducing risk through numbers of factors. You see increased standard deviation because of the increased risks of loading additional factors, which means if even 1 or 2 of your 5 factors end up doing poorly you can end up with a worse portfolio.

This 5 vs 1 narrative is a gross misrepresentation.

A better narrative would be you speed while driving because it gets you there faster and you also code on your phone while driving, both increasing productivity while increasing risks. That may or may not be a good idea, depending on the risk and rewards. Maybe you get more work done but get a speeding ticket, with only one risk showing up. That might be beneficial but the real disaster would be if 5 separate risks show up at the same time.
User avatar
Beensabu
Posts: 5618
Joined: Sun Aug 14, 2016 3:22 pm

Re: Why don't you factor tilt?

Post by Beensabu »

abc132 wrote: Mon Oct 03, 2022 3:53 pm
Apathizer wrote: Mon Oct 03, 2022 1:14 pm So to me it's about increasing the likelihood of better performance or at least consistent performance. But likelihood all five factors will underperform simultaneously is much lower than the likelihood any one of them will underperform at any given time.
You are investing in market plus other factor risk - you are loading up on risk factors.
They are different sources of risk.
A better narrative would be you speed while driving because it gets you there faster and you also code on your phone while driving, both increasing productivity while increasing risks. That may or may not be a good idea, depending on the risk and rewards.
The "driving a car" analogy was terribly inapplicable when you first trotted it out, and this version is no better.

I shall proffer a quite possibly also terrible analogy that is at least slightly more applicable:

Different forms of transportation have different likelihoods of resulting in bodily injury and/or death. Instead of always driving a car, sometimes you walk, or ride a bike, or take a train, or fly. Different sources of risk.
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
abc132
Posts: 2411
Joined: Thu Oct 18, 2018 1:11 am

Re: Why don't you factor tilt?

Post by abc132 »

Beensabu wrote: Mon Oct 03, 2022 4:05 pm
abc132 wrote: Mon Oct 03, 2022 3:53 pm
Apathizer wrote: Mon Oct 03, 2022 1:14 pm So to me it's about increasing the likelihood of better performance or at least consistent performance. But likelihood all five factors will underperform simultaneously is much lower than the likelihood any one of them will underperform at any given time.
You are investing in market plus other factor risk - you are loading up on risk factors.
They are different sources of risk.
A better narrative would be you speed while driving because it gets you there faster and you also code on your phone while driving, both increasing productivity while increasing risks. That may or may not be a good idea, depending on the risk and rewards.
The "driving a car" analogy was terribly inapplicable when you first trotted it out, and this version is no better.

I shall proffer a quite possibly also terrible analogy that is at least slightly more applicable:

Different forms of transportation have different likelihoods of resulting in bodily injury and/or death. Instead of always driving a car, sometimes you walk, or ride a bike, or take a train, or fly. Different sources of risk.
My analogy was very good. Speeding and distracted driving were independent risk factors. You can choose one without the other and when pooled together they increase risk.

If I actually did what you said I would get a lower standard deviation and not a higher standard deviation. Your example does not match what happens when you invest in multiple factors. Mine does. If you can come up with an example where deviation is increased and not decreased, you might produce something better.
Last edited by abc132 on Mon Oct 03, 2022 4:17 pm, edited 1 time in total.
User avatar
Beensabu
Posts: 5618
Joined: Sun Aug 14, 2016 3:22 pm

Re: Why don't you factor tilt?

Post by Beensabu »

abc132 wrote: Mon Oct 03, 2022 4:13 pm If you can come up with an example where deviation is increased and not decreased, you might produce something better.
You need to add fixed income too. But you didn't like that answer.
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
abc132
Posts: 2411
Joined: Thu Oct 18, 2018 1:11 am

Re: Why don't you factor tilt?

Post by abc132 »

Beensabu wrote: Mon Oct 03, 2022 4:16 pm
abc132 wrote: Mon Oct 03, 2022 4:13 pm If you can come up with an example where deviation is increased and not decreased, you might produce something better.
You need to add fixed income too. But you didn't like that answer.
That was actually my earlier recommendation - that fixed income is so good of a diversifier that factors may not be needed.

So I do like the fixed income answer.

I hate the 5 vs 1 factor diversification answer because it is fixed income that is providing the diversification that reduces risk, while the factors add risk.

I wont argue that you might be able to benefit from both (factors + fixed income), but arguing 5 vs 1 factors for safety is the opposite of what is happening with respect to factors. Using 5 factors loads up on risk and we should not be thinking in terms of the chance of all 5 doing poorly. That narrative shows major misunderstanding.
User avatar
Beensabu
Posts: 5618
Joined: Sun Aug 14, 2016 3:22 pm

Re: Why don't you factor tilt?

Post by Beensabu »

abc132 wrote: Mon Oct 03, 2022 4:21 pm
Beensabu wrote: Mon Oct 03, 2022 4:16 pm
abc132 wrote: Mon Oct 03, 2022 4:13 pm If you can come up with an example where deviation is increased and not decreased, you might produce something better.
You need to add fixed income too. But you didn't like that answer.
That was actually my earlier recommendation - that fixed income is so good of a diversifier that factors may not be needed.

So I do like the fixed income answer.

I hate the 5 vs 1 factor diversification answer (loading risk) because it is fixed income that is providing the diversification that reduces risk.

I wont argue that you might be able to benefit from both (factors + fixed income), but arguing 5 vs 1 factors for safety is the opposite of what is happening with respect to factors. Using 5 factors loads up on risk and we should not be thinking in terms of the chance of all 5 doing poorly.
Okay. But how about this perspective (which totally looks at risk as measured by standard deviation, just the way you like to) for a minute?

1. A has standard deviation of X

2. A + B + C + D has standard deviation of >X

3. A + B + C + D + E has standard deviation of ≤X

4. A + E has standard deviation of <X

You are saying that you would prefer #4 over #3, correct?

And if that is your preference, is it still impossible for you to see why someone else might prefer #3? Especially if they look at risk as something that cannot just be measured by standard deviation?
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
User avatar
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Why don't you factor tilt?

Post by vineviz »

abc132 wrote: Mon Oct 03, 2022 4:21 pm

I hate the 5 vs 1 factor diversification answer because it is fixed income that is providing the diversification that reduces risk, while the factors add risk.
This falsely frames the discussion in two ways.

1) Decisions about diversification are not mutually exclusive: the investor can diversify between stocks and bonds and within stocks.

2) "Reducing risk" is not a function or goal of diversification, at least not in absolute terms.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
abc132
Posts: 2411
Joined: Thu Oct 18, 2018 1:11 am

Re: Why don't you factor tilt?

Post by abc132 »

vineviz wrote: Mon Oct 03, 2022 4:42 pm
abc132 wrote: Mon Oct 03, 2022 4:21 pm

I hate the 5 vs 1 factor diversification answer because it is fixed income that is providing the diversification that reduces risk, while the factors add risk.
This falsely frames the discussion in two ways.

1) Decisions about diversification are not mutually exclusive: the investor can diversify between stocks and bonds and within stocks.
Your comment indicates a failure to understand what I said, or a failure to read the words when I said one might benefit from doing exactly that.
vineviz wrote: Mon Oct 03, 2022 4:42 pm 2) "Reducing risk" is not a function or goal of diversification, at least not in absolute terms.
If that commented is directed at the 5 factors is better than 1 because all 5 are less likely to happen at once, then I will say I am glad to agree with you.
User avatar
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Why don't you factor tilt?

Post by vineviz »

abc132 wrote: Mon Oct 03, 2022 5:10 pm
vineviz wrote: Mon Oct 03, 2022 4:42 pm
abc132 wrote: Mon Oct 03, 2022 4:21 pm

I hate the 5 vs 1 factor diversification answer because it is fixed income that is providing the diversification that reduces risk, while the factors add risk.
This falsely frames the discussion in two ways.

1) Decisions about diversification are not mutually exclusive: the investor can diversify between stocks and bonds and within stocks.
Your comment indicates a failure to understand what I said, or a failure to read the words when I said one might benefit from doing exactly that.
I understood what you said. Maybe you didn't say what you meant?
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
abc132
Posts: 2411
Joined: Thu Oct 18, 2018 1:11 am

Re: Why don't you factor tilt?

Post by abc132 »

vineviz wrote: Mon Oct 03, 2022 5:43 pm
abc132 wrote: Mon Oct 03, 2022 5:10 pm
vineviz wrote: Mon Oct 03, 2022 4:42 pm
abc132 wrote: Mon Oct 03, 2022 4:21 pm

I hate the 5 vs 1 factor diversification answer because it is fixed income that is providing the diversification that reduces risk, while the factors add risk.
This falsely frames the discussion in two ways.

1) Decisions about diversification are not mutually exclusive: the investor can diversify between stocks and bonds and within stocks.
Your comment indicates a failure to understand what I said, or a failure to read the words when I said one might benefit from doing exactly that.
I understood what you said. Maybe you didn't say what you meant?
Maybe you can quote where I said diversification was mutually exclusive or that diversification must reduce risk.

My prior comments indicate an understanding that diversification can in fact increase or decrease risk, and that one would need to include factors and fixed income to get a reduction in risk.

Investing in factors is so obviously diversifying within stocks, I'm having trouble with your failure in comprehension.

It would be interesting to see how you resolve how I can agree with your statements while you are telling me I am incorrect.
User avatar
Beensabu
Posts: 5618
Joined: Sun Aug 14, 2016 3:22 pm

Re: Why don't you factor tilt?

Post by Beensabu »

abc132 wrote: Mon Oct 03, 2022 5:10 pm If that commented is directed at the 5 factors is better than 1 because all 5 are less likely to happen at once, then I will say I am glad to agree with you.
abc132 wrote: Mon Oct 03, 2022 6:00 pm My prior comments indicate an understanding that diversification can in fact increase or decrease risk, and that one would need to include factors and fixed income to get a reduction in risk.
Cool. I think you get it. :D That's awesome :beer
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
Apathizer
Posts: 2507
Joined: Sun Sep 26, 2021 2:56 pm

Re: Why don't you factor tilt?

Post by Apathizer »

Beensabu wrote: Mon Oct 03, 2022 6:55 pm
abc132 wrote: Mon Oct 03, 2022 5:10 pm If that commented is directed at the 5 factors is better than 1 because all 5 are less likely to happen at once, then I will say I am glad to agree with you.
abc132 wrote: Mon Oct 03, 2022 6:00 pm My prior comments indicate an understanding that diversification can in fact increase or decrease risk, and that one would need to include factors and fixed income to get a reduction in risk.
Cool. I think you get it. :D That's awesome :beer
Exactly. The likelihood of six different asset classes underperforming is lower than the likelihood of only two underperforming over the same period.

With cap weight stocks and bonds you only hold two asset classes. With a factor slanted portfolio you hold six if you include bonds: market, the other four equity factors, and bonds. The likelihood that all six of these asset classes will underperform is lower than the likelihood of only two underperforming over the same period.
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
rkhusky
Posts: 17654
Joined: Thu Aug 18, 2011 8:09 pm

Re: Why don't you factor tilt?

Post by rkhusky »

Apathizer wrote: Mon Oct 03, 2022 8:51 pm Exactly. The likelihood of six different asset classes underperforming is lower than the likelihood of only two underperforming over the same period.

With cap weight stocks and bonds you only hold two asset classes. With a factor slanted portfolio you hold six if you include bonds: market, the other four equity factors, and bonds. The likelihood that all six of these asset classes will underperform is lower than the likelihood of only two underperforming over the same period.
No, you only own two asset classes: stocks and bonds. For example, small value stocks are stocks. If you want a different asset class, you need precious metals, or commodities, or physical real estate, or micro-loans, or a small business, etc.
User avatar
Beensabu
Posts: 5618
Joined: Sun Aug 14, 2016 3:22 pm

Re: Why don't you factor tilt?

Post by Beensabu »

rkhusky wrote: Mon Oct 03, 2022 9:03 pm
Apathizer wrote: Mon Oct 03, 2022 8:51 pm Exactly. The likelihood of six different asset classes underperforming is lower than the likelihood of only two underperforming over the same period.

With cap weight stocks and bonds you only hold two asset classes. With a factor slanted portfolio you hold six if you include bonds: market, the other four equity factors, and bonds. The likelihood that all six of these asset classes will underperform is lower than the likelihood of only two underperforming over the same period.
No, you only own two asset classes: stocks and bonds. For example, small value stocks are stocks. If you want a different asset class, you need precious metals, or commodities, or physical real estate, or micro-loans, or a small business, etc.
Sub-asset classes, then. Stuff that's just different from each other that it tends to behave differently at times.

Long-term treasuries and cash are both fixed income. Would you say they have behaved differently, lately?

Same deal.
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
JSPECO9
Posts: 457
Joined: Thu Nov 24, 2016 11:34 pm

Re: Why don't you factor tilt?

Post by JSPECO9 »

I own the S&P 500 index fund.

"That's not diversified enough, you need exposure to mid and small cap stocks."

Ok I'll buy Total Stock Market.

"That doesn't give you enough diversification, it's almost the same thing as the S&P 500."

Ok fine, I'll diversify and buy Vanguard Small Cap Value. Diversified, cheap, index fund.

"Nope, that doesn't work, that's not small enough."

Sigh, ok, I'll buy the Vanguard S&P 600 Small Cap Value index fund then, a little more expensive than I want, but it's an index fund. Fine.

"Nope, that's still not enough, something something factor loading beta something something. Momentum, profitability, something something else. You need DFA or Avantis"

Ok just forget it then, I'll just stick to my S&P 500 index fund.
Apathizer
Posts: 2507
Joined: Sun Sep 26, 2021 2:56 pm

Re: Why don't you factor tilt?

Post by Apathizer »

rkhusky wrote: Mon Oct 03, 2022 9:03 pm
Apathizer wrote: Mon Oct 03, 2022 8:51 pm Exactly. The likelihood of six different asset classes underperforming is lower than the likelihood of only two underperforming over the same period.

With cap weight stocks and bonds you only hold two asset classes. With a factor slanted portfolio you hold six if you include bonds: market, the other four equity factors, and bonds. The likelihood that all six of these asset classes will underperform is lower than the likelihood of only two underperforming over the same period.
No, you only own two asset classes: stocks and bonds. For example, small value stocks are stocks. If you want a different asset class, you need precious metals, or commodities, or physical real estate, or micro-loans, or a small business, etc.
The data available shows factors are distinct potential risk/return sources from the cap weight TSM. Factors seem imperfectly correlated with the cap weight TSM, so integrating them is likely to improve return consistency.
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
Apathizer
Posts: 2507
Joined: Sun Sep 26, 2021 2:56 pm

Re: Why don't you factor tilt?

Post by Apathizer »

JSPECO9 wrote: Mon Oct 03, 2022 9:34 pm I own the S&P 500 index fund.

"That's not diversified enough, you need exposure to mid and small cap stocks."

Ok I'll buy Total Stock Market.

"That doesn't give you enough diversification, it's almost the same thing as the S&P 500."

Ok fine, I'll diversify and buy Vanguard Small Cap Value. Diversified, cheap, index fund.

"Nope, that doesn't work, that's not small enough."

Sigh, ok, I'll buy the Vanguard S&P 600 Small Cap Value index fund then, a little more expensive than I want, but it's an index fund. Fine.

"Nope, that's still not enough, something something factor loading beta something something. Momentum, profitability, something something else. You need DFA or Avantis"

Ok just forget it then, I'll just stick to my S&P 500 index fund.
If you just want to keep it really simple, a cap weight TSM portfolio is just as simple as and more well diversified than the S&P 500.
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
JSPECO9
Posts: 457
Joined: Thu Nov 24, 2016 11:34 pm

Re: Why don't you factor tilt?

Post by JSPECO9 »

Apathizer wrote: Mon Oct 03, 2022 9:38 pm
JSPECO9 wrote: Mon Oct 03, 2022 9:34 pm I own the S&P 500 index fund.

"That's not diversified enough, you need exposure to mid and small cap stocks."

Ok I'll buy Total Stock Market.

"That doesn't give you enough diversification, it's almost the same thing as the S&P 500."

Ok fine, I'll diversify and buy Vanguard Small Cap Value. Diversified, cheap, index fund.

"Nope, that doesn't work, that's not small enough."

Sigh, ok, I'll buy the Vanguard S&P 600 Small Cap Value index fund then, a little more expensive than I want, but it's an index fund. Fine.

"Nope, that's still not enough, something something factor loading beta something something. Momentum, profitability, something something else. You need DFA or Avantis"

Ok just forget it then, I'll just stick to my S&P 500 index fund.
If you just want to keep it really simple, a cap weight TSM portfolio is just as simple as and more well diversified than the S&P 500.
S&P 500 is good enough. Thank you though.
abc132
Posts: 2411
Joined: Thu Oct 18, 2018 1:11 am

Re: Why don't you factor tilt?

Post by abc132 »

Apathizer wrote: Mon Oct 03, 2022 8:51 pm
Exactly. The likelihood of six different asset classes underperforming is lower than the likelihood of only two underperforming over the same period.

With cap weight stocks and bonds you only hold two asset classes. With a factor slanted portfolio you hold six if you include bonds: market, the other four equity factors, and bonds. The likelihood that all six of these asset classes will underperform is lower than the likelihood of only two underperforming over the same period.
Do you have any data to show this, like a graph of each of the 5 factors vs time showing a correlation of zero between each factor?

What is the percent of time
- all 5 underperform
- 4 underperform
- 3 underperform
- 2 underperform
- 1 underperform
- 0 underperform

What are you using to measure underperform?
Apathizer
Posts: 2507
Joined: Sun Sep 26, 2021 2:56 pm

Re: Why don't you factor tilt?

Post by Apathizer »

abc132 wrote: Mon Oct 03, 2022 10:43 pm
Apathizer wrote: Mon Oct 03, 2022 8:51 pm
Exactly. The likelihood of six different asset classes underperforming is lower than the likelihood of only two underperforming over the same period.

With cap weight stocks and bonds you only hold two asset classes. With a factor slanted portfolio you hold six if you include bonds: market, the other four equity factors, and bonds. The likelihood that all six of these asset classes will underperform is lower than the likelihood of only two underperforming over the same period.
Do you have any data to show this, like a graph of each of the 5 factors vs time showing a correlation of zero between each factor?

What is the percent of time
- all 5 underperform
- 4 underperform
- 3 underperform
- 2 underperform
- 1 underperform
- 0 underperform

What are you using to measure underperform?
The Ben Felix link provides details. Based on the data available here is the approximate probability each factor will under-perform 1-month T-bills over rolling ten year periods in decimals to keep the math simple. Also to keep things simple I'm using only the US market, though the figure are similar for ex-US markets.

Cap Weight TSM: 0.20
Size: 0.30
Value: 0.15
Profitability: 0.15
Reinvestment: 0.05.

So the likelihood a cap weight TSM index will under-perform is about 20%. For all factors combined we multiply all of them together:
0.02*0.30*0.15*0.15*0.05=0.00000675 or 0.675%. A 20% likelihood of under-performance is low, but still plausible. A 0.675% chance is infinitesimally low; so low I would say it's implausible.

Admittedly this is a gross over-simplification based on general historical data. The actual math is significantly more complicated. I'm just using this to illustrate how diversifying across different risk/return sources significantly reduces the likelihood of under-performance.
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
abc132
Posts: 2411
Joined: Thu Oct 18, 2018 1:11 am

Re: Why don't you factor tilt?

Post by abc132 »

Apathizer wrote: Mon Oct 03, 2022 11:23 pm
abc132 wrote: Mon Oct 03, 2022 10:43 pm
Apathizer wrote: Mon Oct 03, 2022 8:51 pm
Exactly. The likelihood of six different asset classes underperforming is lower than the likelihood of only two underperforming over the same period.

With cap weight stocks and bonds you only hold two asset classes. With a factor slanted portfolio you hold six if you include bonds: market, the other four equity factors, and bonds. The likelihood that all six of these asset classes will underperform is lower than the likelihood of only two underperforming over the same period.
Do you have any data to show this, like a graph of each of the 5 factors vs time showing a correlation of zero between each factor?

What is the percent of time
- all 5 underperform
- 4 underperform
- 3 underperform
- 2 underperform
- 1 underperform
- 0 underperform

What are you using to measure underperform?
The Ben Felix link provides details. Based on the data available here is the approximate probability each factor will under-perform 1-month T-bills over rolling ten year periods in decimals to keep the math simple. Also to keep things simple I'm using only the US market, though the figure are similar for ex-US markets.

Cap Weight TSM: 0.20
Size: 0.30
Value: 0.15
Profitability: 0.15
Reinvestment: 0.05.

So the likelihood a cap weight TSM index will under-perform is about 20%. For all factors combined we multiply all of them together:
0.02*0.30*0.15*0.15*0.05=0.00000675 or 0.675%. A 20% likelihood of under-performance is low, but still plausible. A 0.675% chance is infinitesimally low; so low I would say it's implausible.

Admittedly this is a gross over-simplification based on general historical data. The actual math is significantly more complicated. I'm just using this to illustrate how diversifying across different risk/return sources significantly reduces the likelihood of under-performance.
That's very helpful but this data is still a case of using your own assumption to reach your own conclusion. I was interested in seeing the correlation between factors and backtracking out how combinations of factors actually performed, and also seeing how often factors worked well when the market did not.

To be very clear I understand how non-correlated assets would reduce risk (as shown in your example) but I also know that non-correlated assets would reduce volatility if they were actually randomly offsetting each other. So I'm interested in the actual historical relationships between factors to investigate these two separate issues (how random are they acting and how specifically they increase deviation).
Apathizer
Posts: 2507
Joined: Sun Sep 26, 2021 2:56 pm

Re: Why don't you factor tilt?

Post by Apathizer »

abc132 wrote: Mon Oct 03, 2022 11:38 pm
Apathizer wrote: Mon Oct 03, 2022 11:23 pm
abc132 wrote: Mon Oct 03, 2022 10:43 pm
Apathizer wrote: Mon Oct 03, 2022 8:51 pm
Exactly. The likelihood of six different asset classes underperforming is lower than the likelihood of only two underperforming over the same period.

With cap weight stocks and bonds you only hold two asset classes. With a factor slanted portfolio you hold six if you include bonds: market, the other four equity factors, and bonds. The likelihood that all six of these asset classes will underperform is lower than the likelihood of only two underperforming over the same period.
Do you have any data to show this, like a graph of each of the 5 factors vs time showing a correlation of zero between each factor?

What is the percent of time
- all 5 underperform
- 4 underperform
- 3 underperform
- 2 underperform
- 1 underperform
- 0 underperform

What are you using to measure underperform?
The Ben Felix link provides details. Based on the data available here is the approximate probability each factor will under-perform 1-month T-bills over rolling ten year periods in decimals to keep the math simple. Also to keep things simple I'm using only the US market, though the figure are similar for ex-US markets.

Cap Weight TSM: 0.20
Size: 0.30
Value: 0.15
Profitability: 0.15
Reinvestment: 0.05.

So the likelihood a cap weight TSM index will under-perform is about 20%. For all factors combined we multiply all of them together:
0.02*0.30*0.15*0.15*0.05=0.00000675 or 0.675%. A 20% likelihood of under-performance is low, but still plausible. A 0.675% chance is infinitesimally low; so low I would say it's implausible.

Admittedly this is a gross over-simplification based on general historical data. The actual math is significantly more complicated. I'm just using this to illustrate how diversifying across different risk/return sources significantly reduces the likelihood of under-performance.
That's very helpful but this data is still a case of using your own assumption to reach your own conclusion. I was interested in seeing the correlation between factors and backtracking out how combinations of factors actually performed, and also seeing how often factors worked well when the market did not.

To be very clear I understand how non-correlated assets would reduce risk (as shown in your example) but I also know that non-correlated assets would reduce volatility if they were actually randomly offsetting each other. So I'm interested in the actual historical relationships between factors to investigate these two separate issues (how random are they acting and how specifically they increase deviation).
We can do a back-test comparing the US TSM with US small value. Also, correlation isn't the only thing to consider. Dispersion is also important. Even if different assets are highly correlated, the dispersion can be significant. For instance recently US and ex-US markets have been highly correlated, but there's still significant dispersion. Anyway, here's a comparison of US TSM and US SV.
https://www.portfoliovisualizer.com/bac ... tion2_3=20
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
User avatar
Beensabu
Posts: 5618
Joined: Sun Aug 14, 2016 3:22 pm

Re: Why don't you factor tilt?

Post by Beensabu »

abc132 wrote: Mon Oct 03, 2022 11:38 pm So I'm interested in the actual historical relationships between factors to investigate these two separate issues (how random are they acting and how specifically they increase deviation).
About 1/3 of the way down the page of this thing for the "factor premia is dead" crowd to seize upon and wave about with glee is a chart that I think sort of shows part of what you are asking for:

https://www.man.com/maninstitute/the-fu ... ant-equity

Feel free to run down your own rabbit holes. It can be quite fun.
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
abc132
Posts: 2411
Joined: Thu Oct 18, 2018 1:11 am

Re: Why don't you factor tilt?

Post by abc132 »

Beensabu wrote: Tue Oct 04, 2022 12:30 am
abc132 wrote: Mon Oct 03, 2022 11:38 pm So I'm interested in the actual historical relationships between factors to investigate these two separate issues (how random are they acting and how specifically they increase deviation).
About 1/3 of the way down the page of this thing for the "factor premia is dead" crowd to seize upon and wave about with glee is a chart that I think sort of shows part of what you are asking for:

https://www.man.com/maninstitute/the-fu ... ant-equity

Feel free to run down your own rabbit holes. It can be quite fun.
Amazing info, and actually much worse recently for factors than I expected.

The average of all five factors has been negative over the last 10 years before fees. That should be nearly impossible from 5 independent factors with positive expected premiums and the rate of overperformance in the Ben Felix data from Apathizer posted earlier. This maninstitute data seems shows that factors are not helping through their independent actions but rather helping or hurting through cyclical preferences for or against all factors. The cyclical preference for (or against) factors argument makes much more sense to me than the 5 factor diversification argument.

I would go so far as to say this data disproves the 5 factors vs 1 argument, while still leaving potentially good (cyclical) reasons to consider factor investing. We should believe that historical preference for factors will return and not disappear completely if we choose to factor invest.

Thanks, that was very helpful.
rkhusky
Posts: 17654
Joined: Thu Aug 18, 2011 8:09 pm

Re: Why don't you factor tilt?

Post by rkhusky »

Apathizer wrote: Mon Oct 03, 2022 10:32 am Based on the data available the higher returns of the 4 non-market factors have occurred about 70-80% of the time. That seems pretty consistent. Rigorous analysis by multiple sources reduces the likely these results are attributable to data mining or random chance. It could be, but it's unlikely. Remember, we're talking about probability not certainty.
Perhaps you can share the details of your calculations. I find that the Vanguard Total Stock Market fund (VTSMX) has outperformed the Vanguard Small Cap Value fund (VISVX) 51% of the time over the last 20 years. I also find that the Vanguard Total Stock Market fund (VTSMX) has outperformed the iShares Momentum Factor ETF (MTUM) 47% of the time from the inception of MTUM. So, I am not sure where your 70-80% numbers are coming from or what factors you are talking about.

Edit: I also find that VTSMX outperforms DFSVX (DFA Small Value) 49.5% of the time, since the inception of DFSVX in 2/1993.
Last edited by rkhusky on Tue Oct 04, 2022 9:43 am, edited 1 time in total.
rkhusky
Posts: 17654
Joined: Thu Aug 18, 2011 8:09 pm

Re: Why don't you factor tilt?

Post by rkhusky »

Beensabu wrote: Mon Oct 03, 2022 9:32 pm
rkhusky wrote: Mon Oct 03, 2022 9:03 pm
Apathizer wrote: Mon Oct 03, 2022 8:51 pm Exactly. The likelihood of six different asset classes underperforming is lower than the likelihood of only two underperforming over the same period.

With cap weight stocks and bonds you only hold two asset classes. With a factor slanted portfolio you hold six if you include bonds: market, the other four equity factors, and bonds. The likelihood that all six of these asset classes will underperform is lower than the likelihood of only two underperforming over the same period.
No, you only own two asset classes: stocks and bonds. For example, small value stocks are stocks. If you want a different asset class, you need precious metals, or commodities, or physical real estate, or micro-loans, or a small business, etc.
Sub-asset classes, then. Stuff that's just different from each other that it tends to behave differently at times.

Long-term treasuries and cash are both fixed income. Would you say they have behaved differently, lately?

Same deal.
It's not the same deal. Long-term treasuries and cash are different things, which behave differently. When you pile factor upon factor, you are not adding different things together, you are making your single group of stocks more and more alike. So you end up with one thing - a single group of stocks that all have the same factor characteristics and all behave the same way.
rkhusky
Posts: 17654
Joined: Thu Aug 18, 2011 8:09 pm

Re: Why don't you factor tilt?

Post by rkhusky »

Apathizer wrote: Mon Oct 03, 2022 1:14 pm So to me it's about increasing the likelihood of better performance or at least consistent performance. But likelihood all five factors will underperform simultaneously is much lower than the likelihood any one of them will underperform at any given time.
You are also increasing the likelihood of worse performance. You need the majority of the factors to outperform in order for the whole to outperform. The likelihood of all 5 factors outperforming is the same as the likelihood of all 5 factors underperforming. The likelihood of 4 factors outperforming is the same as the likelihood of 4 factors underperforming. The likelihood of 3 factors outperforming is the same as the likelihood of 3 factors underperforming.
Da5id
Posts: 5058
Joined: Fri Feb 26, 2016 7:20 am

Re: Why don't you factor tilt?

Post by Da5id »

rkhusky wrote: Tue Oct 04, 2022 8:20 am
Apathizer wrote: Mon Oct 03, 2022 1:14 pm So to me it's about increasing the likelihood of better performance or at least consistent performance. But likelihood all five factors will underperform simultaneously is much lower than the likelihood any one of them will underperform at any given time.
You are also increasing the likelihood of worse performance. You need the majority of the factors to outperform in order for the whole to outperform. The likelihood of all 5 factors outperforming is the same as the likelihood of all 5 factors underperforming. The likelihood of 4 factors outperforming is the same as the likelihood of 4 factors underperforming. The likelihood of 3 factors outperforming is the same as the likelihood of 3 factors underperforming.
I'm not on team factor. I find it interesting and even plausible, but not compelling enough for me to invest that way. I'm dubious about anyone arguing that it increases risk adjusted performance, but mostly I think the arguments are about attempts to mitigate risk.

But I find your statements not totally logical. e.g. If one factor massively outperforms the market (say small or value does) and the others modestly underperform you could outperform the market with your portfolio. So your idea that you need "the majority of the factors to outperform" isn't reasonable to me as it depends on the magnitude of the out/under performance of each factor.
rkhusky
Posts: 17654
Joined: Thu Aug 18, 2011 8:09 pm

Re: Why don't you factor tilt?

Post by rkhusky »

Da5id wrote: Tue Oct 04, 2022 10:00 am But I find your statements not totally logical. e.g. If one factor massively outperforms the market (say small or value does) and the others modestly underperform you could outperform the market with your portfolio. So your idea that you need "the majority of the factors to outperform" isn't reasonable to me.
What are the odds of that happening? And what are the odds of the market massively outperforming the factors? Most everyone invests via long-only funds, which have been highly correlated with the market in the past, and with each other. So, while we can't compute the odds for the future, it is very unlikely that one of the long-only factors is going to massively outperform the market, while the others underperform the market

And when you are looking at the probability, you need to look at all the probabilities. You can't just say that the likelihood of all factors underperforming is small. You also need to consider that the likelihood of all factors overperforming is also small.
Last edited by rkhusky on Tue Oct 04, 2022 10:11 am, edited 1 time in total.
User avatar
Beensabu
Posts: 5618
Joined: Sun Aug 14, 2016 3:22 pm

Re: Why don't you factor tilt?

Post by Beensabu »

rkhusky wrote: Tue Oct 04, 2022 8:13 am
Beensabu wrote: Mon Oct 03, 2022 9:32 pm
rkhusky wrote: Mon Oct 03, 2022 9:03 pm
Apathizer wrote: Mon Oct 03, 2022 8:51 pm Exactly. The likelihood of six different asset classes underperforming is lower than the likelihood of only two underperforming over the same period.

With cap weight stocks and bonds you only hold two asset classes. With a factor slanted portfolio you hold six if you include bonds: market, the other four equity factors, and bonds. The likelihood that all six of these asset classes will underperform is lower than the likelihood of only two underperforming over the same period.
No, you only own two asset classes: stocks and bonds. For example, small value stocks are stocks. If you want a different asset class, you need precious metals, or commodities, or physical real estate, or micro-loans, or a small business, etc.
Sub-asset classes, then. Stuff that's just different from each other that it tends to behave differently at times.

Long-term treasuries and cash are both fixed income. Would you say they have behaved differently, lately?

Same deal.
It's not the same deal. Long-term treasuries and cash are different things, which behave differently. When you pile factor upon factor, you are not adding different things together, you are making your single group of stocks more and more alike. So you end up with one thing - a single group of stocks that all have the same factor characteristics and all behave the same way.
Who's piling factors upon factors? I'm talking about diversification within an asset class by having a non-negligible holding in sub-asset classes that behave differently at times. Whether or not you wish to do it by factor tilting, it is possible to do within your equity allocation. I don't factor tilt, but I spread the risk around to different areas than just large cap growth. Because if the opposite of large growth can languish for a long period of time (after having clearly done quite well for a long period of time), so can large growth.

I don't want just one kind of thing. I don't want to pin everything on a large growth momentum strategy. That doesn't seem wise to me. If I survive all the things that could get me out of here along the way, I'm looking at 50+ more years based on family longevity. That's a long time, during which lots of stuff could happen.

Also, those people who do factor tilt will hit two factors with one fund, or try to avoid additional loading on a particular factor on purpose in order to avoid exactly what you're talking about. That's why you'll see the "X load on Y" thing from them. It's personalized - balanced to taste.

You don't have to want to do it yourself. But it's silly to avoid understanding why other people do. And if you're going to argue against it, then go read the link I posted upthread a bit. Because that argument has weight to it, but your current one doesn't.
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
Da5id
Posts: 5058
Joined: Fri Feb 26, 2016 7:20 am

Re: Why don't you factor tilt?

Post by Da5id »

rkhusky wrote: Tue Oct 04, 2022 10:08 am
Da5id wrote: Tue Oct 04, 2022 10:00 am But I find your statements not totally logical. e.g. If one factor massively outperforms the market (say small or value does) and the others modestly underperform you could outperform the market with your portfolio. So your idea that you need "the majority of the factors to outperform" isn't reasonable to me.
What are the odds of that happening? And what are the odds of the market massively outperforming the factors? Most everyone invests via long-only funds, which have been highly correlated with the market in the past, and with each other. So, while we can't compute the odds for the future, it is very unlikely that one of the long-only factors is going to massively outperform the market, while the others underperform the market

And when you are looking at the probability, you need to look at all the probabilities. You can't just say that the likelihood of all factors underperforming is small. You also need to consider that the likelihood of all factors overperforming is also small.
You are focused too much on the outperformance aspect in my opinion. If I were to choose to invest in factors, higher risk adjusted performance wouldn't be my motivation.

It feels like your logic would work for 100% US investing advocacy. While I don't do factors, I do have some ex-US. You actually appear to be arguing against diversification as I understand it. IF it is the case that the various factors aren't strictly correlated with each other, investing that way *may* decrease risk/variance compared to the market portfolio is what I understand to be the strongest point in favor of factors. Now that is a big IF, who knows if the features that give factors diversification benefits will persist into the future?

There are arguments in favor of factors that are stronger than you make them seem IMO.
User avatar
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Why don't you factor tilt?

Post by vineviz »

rkhusky wrote: Tue Oct 04, 2022 10:08 am
Da5id wrote: Tue Oct 04, 2022 10:00 am But I find your statements not totally logical. e.g. If one factor massively outperforms the market (say small or value does) and the others modestly underperform you could outperform the market with your portfolio. So your idea that you need "the majority of the factors to outperform" isn't reasonable to me.
What are the odds of that happening? And what are the odds of the market massively outperforming the factors? Most everyone invests via long-only funds, which have been highly correlated with the market in the past, and with each other. So, while we can't compute the odds for the future, it is very unlikely that one of the long-only factors is going to massively outperform the market, while the others underperform the market
What are the odds? On average, about 60% of the time a diversified group of factors has had higher returns than just the market factor alone.

There is no such thing as a "long-only factor". All factors are measured as long-short portfolios (i.e. the difference between the return of portfolio X and the return of portfolio Y).

And you can't say "we can't compute the odds for the future" and then pretend you've competed the odds for the future (i.e. "it is very unlikely"). Pick a side.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
rkhusky
Posts: 17654
Joined: Thu Aug 18, 2011 8:09 pm

Re: Why don't you factor tilt?

Post by rkhusky »

Beensabu wrote: Tue Oct 04, 2022 10:11 am
rkhusky wrote: Tue Oct 04, 2022 8:13 am
Beensabu wrote: Mon Oct 03, 2022 9:32 pm
rkhusky wrote: Mon Oct 03, 2022 9:03 pm
Apathizer wrote: Mon Oct 03, 2022 8:51 pm Exactly. The likelihood of six different asset classes underperforming is lower than the likelihood of only two underperforming over the same period.

With cap weight stocks and bonds you only hold two asset classes. With a factor slanted portfolio you hold six if you include bonds: market, the other four equity factors, and bonds. The likelihood that all six of these asset classes will underperform is lower than the likelihood of only two underperforming over the same period.
No, you only own two asset classes: stocks and bonds. For example, small value stocks are stocks. If you want a different asset class, you need precious metals, or commodities, or physical real estate, or micro-loans, or a small business, etc.
Sub-asset classes, then. Stuff that's just different from each other that it tends to behave differently at times.

Long-term treasuries and cash are both fixed income. Would you say they have behaved differently, lately?

Same deal.
It's not the same deal. Long-term treasuries and cash are different things, which behave differently. When you pile factor upon factor, you are not adding different things together, you are making your single group of stocks more and more alike. So you end up with one thing - a single group of stocks that all have the same factor characteristics and all behave the same way.
Who's piling factors upon factors? I'm talking about diversification within an asset class by having a non-negligible holding in sub-asset classes that behave differently at times. Whether or not you wish to do it by factor tilting, it is possible to do within your equity allocation. I don't factor tilt, but I spread the risk around to different areas than just large cap growth. Because if the opposite of large growth can languish for a long period of time (after having clearly done quite well for a long period of time), so can large growth.

I don't want just one kind of thing. I don't want to pin everything on a large growth momentum strategy. That doesn't seem wise to me. If I survive all the things that could get me out of here along the way, I'm looking at 50+ more years based on family longevity. That's a long time, during which lots of stuff could happen.

Also, those people who do factor tilt will hit two factors with one fund, or try to avoid additional loading on a particular factor on purpose in order to avoid exactly what you're talking about. That's why you'll see the "X load on Y" thing from them. It's personalized - balanced to taste.

You don't have to want to do it yourself. But it's silly to avoid understanding why other people do. And if you're going to argue against it, then go read the link I posted upthread a bit. Because that argument has weight to it, but your current one doesn't.
If you are using more than one factor, such as a small value fund, you are applying a small filter on top of a value filter. If you apply a momentum filter, that further reduces the stocks in your fund.

I have no problem with people investing in factors, but I object to misleading descriptions of it. Factor investing is not investing in different asset classes or different asset sub-classes or having different sources of return. A factor fund has one source of return, which is the basket of stocks in the fund. The stocks in that basket all have the same factor characteristics and should therefore behave the same.

If you want to add your factor fund(s) to a TSM fund, which dilutes the factor characteristics of your portfolio, that is fine. Diluting the factor characteristics lowers the risk.

If you want to overweight certain components of your portfolio, by, for example, equal weighting LG, SG, LV, and SV funds, that would be fine too. Those four are true examples of sub-classes of stocks, which, when added together, increase your portfolio diversification.
rkhusky
Posts: 17654
Joined: Thu Aug 18, 2011 8:09 pm

Re: Why don't you factor tilt?

Post by rkhusky »

vineviz wrote: Tue Oct 04, 2022 10:22 am There is no such thing as a "long-only factor". All factors are measured as long-short portfolios (i.e. the difference between the return of portfolio X and the return of portfolio Y).
So no one who invests in long-only funds is factor investing? I would be fine with that definition.
vineviz wrote: Tue Oct 04, 2022 10:22 am And you can't say "we can't compute the odds for the future" and then pretend you've competed the odds for the future (i.e. "it is very unlikely"). Pick a side.
I never provided specific odds.
Last edited by rkhusky on Tue Oct 04, 2022 10:34 am, edited 1 time in total.
User avatar
Beensabu
Posts: 5618
Joined: Sun Aug 14, 2016 3:22 pm

Re: Why don't you factor tilt?

Post by Beensabu »

rkhusky wrote: Tue Oct 04, 2022 10:23 am If you want to overweight certain components of your portfolio, by, for example, equal weighting LG, SG, LV, and SV funds, that would be fine too. Those four are true examples of sub-classes of stocks, which, when added together, increase your portfolio diversification.
Those four are examples of combinations of factors...

Small is SmB

Value is HmL

Large is BmS - still a factor, just not one for which a premium was found looking backwards

Growth is LmH - still a factor, just not one for which a premium was found looking backwards
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
rkhusky
Posts: 17654
Joined: Thu Aug 18, 2011 8:09 pm

Re: Why don't you factor tilt?

Post by rkhusky »

vineviz wrote: Tue Oct 04, 2022 10:22 am What are the odds? On average, about 60% of the time a diversified group of factors has had higher returns than just the market factor alone.
Would you provide the details where that number comes from? Fund name, time frame, etc.
rkhusky
Posts: 17654
Joined: Thu Aug 18, 2011 8:09 pm

Re: Why don't you factor tilt?

Post by rkhusky »

Beensabu wrote: Tue Oct 04, 2022 10:29 am
rkhusky wrote: Tue Oct 04, 2022 10:23 am If you want to overweight certain components of your portfolio, by, for example, equal weighting LG, SG, LV, and SV funds, that would be fine too. Those four are true examples of sub-classes of stocks, which, when added together, increase your portfolio diversification.
Those four are examples of combinations of factors...

Small is SmB

Value is HmL

Large is BmS - still a factor, just not one for which a premium was found looking backwards

Growth is LmH - still a factor, just not one for which a premium was found looking backwards
We need to decide if small, growth etc are factors, or are only SmB and HmL factors?
rkhusky
Posts: 17654
Joined: Thu Aug 18, 2011 8:09 pm

Re: Why don't you factor tilt?

Post by rkhusky »

Da5id wrote: Tue Oct 04, 2022 10:17 am
rkhusky wrote: Tue Oct 04, 2022 10:08 am
Da5id wrote: Tue Oct 04, 2022 10:00 am But I find your statements not totally logical. e.g. If one factor massively outperforms the market (say small or value does) and the others modestly underperform you could outperform the market with your portfolio. So your idea that you need "the majority of the factors to outperform" isn't reasonable to me.
What are the odds of that happening? And what are the odds of the market massively outperforming the factors? Most everyone invests via long-only funds, which have been highly correlated with the market in the past, and with each other. So, while we can't compute the odds for the future, it is very unlikely that one of the long-only factors is going to massively outperform the market, while the others underperform the market

And when you are looking at the probability, you need to look at all the probabilities. You can't just say that the likelihood of all factors underperforming is small. You also need to consider that the likelihood of all factors overperforming is also small.
You are focused too much on the outperformance aspect in my opinion. If I were to choose to invest in factors, higher risk adjusted performance wouldn't be my motivation.

It feels like your logic would work for 100% US investing advocacy. While I don't do factors, I do have some ex-US. You actually appear to be arguing against diversification as I understand it. IF it is the case that the various factors aren't strictly correlated with each other, investing that way *may* decrease risk/variance compared to the market portfolio is what I understand to be the strongest point in favor of factors. Now that is a big IF, who knows if the features that give factors diversification benefits will persist into the future?

There are arguments in favor of factors that are stronger than you make them seem IMO.
Adding US and Int'l stocks is true diversification. Investing in a small corner of the stock market is not diversification, it is concentrating/compounding risk in the hope of achieving higher returns.
Da5id
Posts: 5058
Joined: Fri Feb 26, 2016 7:20 am

Re: Why don't you factor tilt?

Post by Da5id »

rkhusky wrote: Tue Oct 04, 2022 10:36 am
Da5id wrote: Tue Oct 04, 2022 10:17 am
rkhusky wrote: Tue Oct 04, 2022 10:08 am
Da5id wrote: Tue Oct 04, 2022 10:00 am But I find your statements not totally logical. e.g. If one factor massively outperforms the market (say small or value does) and the others modestly underperform you could outperform the market with your portfolio. So your idea that you need "the majority of the factors to outperform" isn't reasonable to me.
What are the odds of that happening? And what are the odds of the market massively outperforming the factors? Most everyone invests via long-only funds, which have been highly correlated with the market in the past, and with each other. So, while we can't compute the odds for the future, it is very unlikely that one of the long-only factors is going to massively outperform the market, while the others underperform the market

And when you are looking at the probability, you need to look at all the probabilities. You can't just say that the likelihood of all factors underperforming is small. You also need to consider that the likelihood of all factors overperforming is also small.
You are focused too much on the outperformance aspect in my opinion. If I were to choose to invest in factors, higher risk adjusted performance wouldn't be my motivation.

It feels like your logic would work for 100% US investing advocacy. While I don't do factors, I do have some ex-US. You actually appear to be arguing against diversification as I understand it. IF it is the case that the various factors aren't strictly correlated with each other, investing that way *may* decrease risk/variance compared to the market portfolio is what I understand to be the strongest point in favor of factors. Now that is a big IF, who knows if the features that give factors diversification benefits will persist into the future?

There are arguments in favor of factors that are stronger than you make them seem IMO.
Adding US and Int'l stocks is true diversification. Investing in a small corner of the stock market is not diversification, it is concentrating/compounding risk in the hope of achieving higher returns.
What are the odds of ex-US outperforming US? What percentage of the time does that happen? How often does it matter? Very unlikely etc. That feels like how your argument would apply to that realm.

And you are assuming the conclusion that factors are hooey. And that the goal is outperformance. I don't assume that conclusion or that goal.

I think the *idea* that one can identify groups of stocks (like small, value, momentum, quality, whatever) that behave *differently* than the market and *differently* than each other isn't inherently unreasonable. And that combining investing those stocks can give a diversification benefit compared to the market portfolio isn't inherently unreasonable either. At least to me. I just don't choose to do it as I'm not totally sold on factor behaviors persisting into the future.
User avatar
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Why don't you factor tilt?

Post by vineviz »

rkhusky wrote: Tue Oct 04, 2022 10:28 am
vineviz wrote: Tue Oct 04, 2022 10:22 am There is no such thing as a "long-only factor". All factors are measured as long-short portfolios (i.e. the difference between the return of portfolio X and the return of portfolio Y).
So no one who invests in long-only funds is factor investing? I would be fine with that definition.
No.

Factors are long-short returns.

Investors can gain exposure to those factors using long-short portfolios OR long-only portfolios.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
Beensabu
Posts: 5618
Joined: Sun Aug 14, 2016 3:22 pm

Re: Why don't you factor tilt?

Post by Beensabu »

rkhusky wrote: Tue Oct 04, 2022 10:33 am
Beensabu wrote: Tue Oct 04, 2022 10:29 am
rkhusky wrote: Tue Oct 04, 2022 10:23 am If you want to overweight certain components of your portfolio, by, for example, equal weighting LG, SG, LV, and SV funds, that would be fine too. Those four are true examples of sub-classes of stocks, which, when added together, increase your portfolio diversification.
Those four are examples of combinations of factors...

Small is SmB

Value is HmL

Large is BmS - still a factor, just not one for which a premium was found looking backwards

Growth is LmH - still a factor, just not one for which a premium was found looking backwards
We need to decide if small, growth etc are factors, or are only SmB and HmL factors?
We don't need to decide anything. They are.

SmB (small minus big) is small

LmH (low minus high) is growth

SmB + LmH (small growth) just doesn't happen to have displayed a past premium over the market (rather, the opposite).

You can take information and do what you want with it. But information is information.
"The only thing that makes life possible is permanent, intolerable uncertainty; not knowing what comes next." ~Ursula LeGuin
Apathizer
Posts: 2507
Joined: Sun Sep 26, 2021 2:56 pm

Re: Why don't you factor tilt?

Post by Apathizer »

rkhusky wrote: Tue Oct 04, 2022 8:08 am
Apathizer wrote: Mon Oct 03, 2022 10:32 am Based on the data available the higher returns of the 4 non-market factors have occurred about 70-80% of the time. That seems pretty consistent. Rigorous analysis by multiple sources reduces the likely these results are attributable to data mining or random chance. It could be, but it's unlikely. Remember, we're talking about probability not certainty.
Perhaps you can share the details of your calculations. I find that the Vanguard Total Stock Market fund (VTSMX) has outperformed the Vanguard Small Cap Value fund (VISVX) 51% of the time over the last 20 years. I also find that the Vanguard Total Stock Market fund (VTSMX) has outperformed the iShares Momentum Factor ETF (MTUM) 47% of the time from the inception of MTUM. So, I am not sure where your 70-80% numbers are coming from or what factors you are talking about.

Edit: I also find that VTSMX outperforms DFSVX (DFA Small Value) 49.5% of the time, since the inception of DFSVX in 2/1993.
While Vanguard is great for straight index funds they don't do factors well. Also until the last couple years small value has lagged well the market led by large growth has done well.

My portfolio is in the signature section. There isn't much track record but it's outperformed the capweight market by about 2% annually. Again I know that's only about 3 years so not much of a track record, but it's structured in such a way that won the market does well the performance seems to be about the same as cap weight TSM. And when the market or large growth underperforms, my Factor slants mitigate this underperformance and in some cases there can be a positive return while the market is flat or negative.
Last edited by Apathizer on Tue Oct 04, 2022 1:09 pm, edited 1 time in total.
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
Apathizer
Posts: 2507
Joined: Sun Sep 26, 2021 2:56 pm

Re: Why don't you factor tilt?

Post by Apathizer »

rkhusky wrote: Tue Oct 04, 2022 8:20 am
Apathizer wrote: Mon Oct 03, 2022 1:14 pm So to me it's about increasing the likelihood of better performance or at least consistent performance. But likelihood all five factors will underperform simultaneously is much lower than the likelihood any one of them will underperform at any given time.
You are also increasing the likelihood of worse performance. You need the majority of the factors to outperform in order for the whole to outperform. The likelihood of all 5 factors outperforming is the same as the likelihood of all 5 factors underperforming. The likelihood of 4 factors outperforming is the same as the likelihood of 4 factors underperforming. The likelihood of 3 factors outperforming is the same as the likelihood of 3 factors underperforming.
Yes Factor slants do increase the likelihood of underperformance, but based on historical data in our understanding of markets they are more likely to improve performance and consistency.

With a standard cap weight portfolio you're relying entirely on the market dominated by large growth. And I don't think you understand the math. The likelihood of three things happening simultaneously is lower than one thing happening on its own. So if your portfolio is spread out among five factors and two underperform your returns are likely to be higher than a cap weight portfolio, but of course this depends on the degree of underperformance.
ROTH: 50% AVGE, 10% DFAX, 40% BNDW. Taxable: 50% BNDW, 40% AVGE, 10% DFAX.
Ependytis
Posts: 612
Joined: Mon Oct 17, 2016 11:10 am

Re: Why don't you factor tilt?

Post by Ependytis »

To increase the amount of money you had to spend in retirement there are three things you could do. One is to invest in more stocks. Two is to factor tilt. Three is to save more. Investing in more stocks is a risk especially with sequence of returns risk. Factor tilting and saving more would take a long time to come to fruition therefore most retirees may not have the time.
HeavyChevy
Posts: 516
Joined: Sun Oct 31, 2021 7:07 am
Location: Below the Mac bridge (Troll)

Re: Why don't you factor tilt?

Post by HeavyChevy »

I guess I factor tilt without ever intending to.

My entire stock portfolio could be summarized as 50% large cap value, 50% mid cap (broad)

I took this direction at retirement after benefitting greatly from the general market runup, particularly large cap growth.

I care nothing about academic factor investing analysis. I wanted to get the historical relative stability of large cap value stocks along with some dividend cash. I wanted the historically greater growth potential of smaller cap stocks while avoiding the friction and inefficiencies of truly small cap stocks. I backtested this fifty-fifty mix just to ensure that historically it hasn't been a disaster, and was pleased to see the results.

That is my portfolio. It obviously will largely track TSM, yet satisfies my own unique intentions. It has been fine so far (couple years in).

I am NOT trying to outperform the market, but perhaps trying not to grossly underperform. I don't think this portfolio is magical in any sense. Just fundamentally believed going in that possible inflation was underemphasized and certain growth stocks (TSLA) had little rational basis for their values unless we got to negative interest rates.

I'm a factor investor who never read the literature on factor investing, nor do I care about mathematical valuation models. I shunned academia to work in an applied research lab for my entire career. I solved real problems that I didn't get to formulate. They came to me with myriad uncertainties and were often poorly defined. I guess that is why I find personal investing interesting. It is mostly behavior/psychology that people try to solve with math. Good luck.
"It's not the best move, but it is a move." - GMHikaru
km91
Posts: 1373
Joined: Wed Oct 13, 2021 12:32 pm

Re: Why don't you factor tilt?

Post by km91 »

Apathizer wrote: Tue Oct 04, 2022 1:08 pm With a standard cap weight portfolio you're relying entirely on the market dominated by large growth.
There's an important point about diversification in here that deserves more attention. With a cap weighted TSM portfolio, the market decides how much of each stock you buy, and naturally it is biased towards large and growth. By factor tilting, you are diversifying your exposure into companies that are underrepresented in a cap weighted index. Even if we assume that the there is no premium at all from holding the factors and that large, small, value, and growth all should have more or less the same expected returns in aggregate, there is still a diversification benefit to be had by overweighting towards the factors. The ten largest companies make up almost 25% of the market cap of VTI. If there truly is no difference between large and small companies, why wouldn't an investor lessen their concentration risk to these top 10 companies by holding more small companies? A dollar spread across the 2000 smallest companies is more diversified than a dollar spread across the largest 500, where a quarter of the concentration is to the top 10 largest.
Post Reply