I don’t know about crushing dreams. Means a lot of people were wrong but at the same time many funds use a multiple factor approach such as value, profitability/quality and then use momentum for trading decisions. Most are very mindful of transaction costs. Where Andrew’s study did not examine the implications of multiple factors working in combination…. Which of course would be a gigantic task.
Hey this is Andrew. My co-authors Mihail Velikov and Tom Zimmermann do look at this, and they both find that there are returns net of costs with ML combinations of predictors in recent years. But in the bigger picture, wouldn't these net returns get traded away, if many funds are pursuing them?
@@achenfinance-z2k I am no expert but I don’t think so. Many people pursue the market factor and it is still around. So anything that is risk based has to exist over long periods of time or at least we have to think it does. If you could get the same returns in your checking account as the bonds and stock markets those markets would not exist.
Hey @@achenfinance-z2k I enjoyed the podcast! Depending on whether your believe the multi-factor returns are due to miss-pricing vs risk, that would determine whether they could become arbitraged away. So if I am understanding you right, multi factor approaches like DFA or Avantis still generate excess returns after transaction costs, as long as the factors are risk based and not miss pricing? I’m going to look more into these papers
Really looking forward to a breakdown of this and how to better understand the implications. It just intrinsically makes sense that higher discount rate=more risk=higher expected returns
That's true, unless and it's just my opinion... As markets are efficient, the price you pay for that high discount rate creates a match between the return and the probability of you capturing that extra return for the extra risk. If all the world invests in the same value tilt portfolio, the return is driven away by the fact that the price of the components move higher, reducing the expected return. At the end, value was just a temporary misspricing that the efficient market already solved.
The answer seems to be all asset prices are, with various degree, the result of data mining and being fast about making trades. Additionally, the market is getting more efficient due to better tech. Thus, factor investing doesn't really have as much impact as was previously thought. During the interview it is mentioned that data mining is perceived in a poor light. But the argument Chen provides is that they should be using Econ philosophy rather than the amalgamation of data. My counter would be: an economist wants elegance, a trader wants cents.
I would love to see another intelligent and well spoken person debate Andrew, this all seems great but I feel like it would be useful to have someone highly knowledgeable debate these points.
@@rationalreminder I don't mean to sound demanding, but would you mind giving a timestamp or general idea of where in that 8 hour video he talks about this paper?
The interview I've waited for the past 6 months, as until now Ben's videos and Andrew's papers have been the most insightful sources of knowledge on investing in stocks for me. After reading Andrew's papers I'm still torn between investing in an equally-weighted portfolio of stocks with a composite of strong data-mined accounting signals to scrape the remaining arbitrage off the floor or just buying the global market equally for the rest of my life. Maybe I'll do a bit of data-mined signal investing just for the future me who would second-guess this decision anyway.
What a beast! We Humans love Stories (and factors). Statistitians brutally destroying our nice logical theories and dreams of scientific persistent overperformance. Im finally going to consolidate my 20 etfs in just 1 😅
Great interview...lots to think about in my own investing strategy. it does increasingly seem like the markets are driven by liquidity and positioning more so than risk factors.
In timing the factor zoo (2023) they also find that timing factors by some data mined characteristics increases returns by another 1-2% if you take this into account plus returns by efficient transactions is is at least possible to outperform consistently. After a brief existential crisis, I now have higher expectations of myself and the complexity of the task. Really really distrupting results nevertheless, thanks for the Interview.
Most of the factor return decay is specifically in cap weighted US stocks. If you go equal weight US or international, the vast majority of factors are pretty much as strong as they've ever been. A very poignant example is EBITDA/EV (historically one of the strongest value factors). Cap weighted last 10 year alpha in the US is -1.8% per year. Switch to equal weight and it's 8.4% per year. Explanations can be that either S&P 500 specifically got way more efficient in the mid 2000s or that the crazy bull market and multiple expansion that the S&P 500 has seen in the past 15 years has somehow suppressed factor premiums. This plays somewhat into the Mike Green school of thought where factors can't really work if most capital just mindlessly flows to mega caps. Whatever the reason may be, the lesson is stay away from cap weighted funds. Go US small and go international for factor exposure.
@@matiasnicolaspizmeny I don't know much about how efficient firms like DFA or Avantis can be at maintaining low slippage with high turnover and capacity. When you are managing tens of billions I imagine it becomes a very difficult problem. But herein lies the greatest advantage of the retail investor. We are not capacity constrained. We can go after the biggest premiums in the smallest most illiquid microcaps. AVUV has over 12 billion in assets. They cannot invest in some amazing factor premiums in a 10 million illiquid microcap. But you and I can, with virtually no cost and without impacting the price in any meaningful way. DIY is the future for the retail factor investor.
Ok so it doesnt make sense to include funds that invest in factors/predictors if youre looking for excess returns, but does it make sense to invest in them if youre looking to diversify? Can you be more diversified than a market cap weighted index with all the investable market? Would overweighting factors smooth out returns / shrink the distribution of outcomes, reducing the probability of ending up in either tail of the distribution? Great episode btw
Gas Station Sushi Crew is concerned after this episode. ⛽️🍣 Cross-sectional asset pricing predictor replication meta-research keeping Ben's hair looking good.
He sounds super intelligent, but some of the arguments he makes against others can be made against him. He talks a lot about the stories authors want to make while he tries to make his own story about why the numbers look the they do. I wouldn't change anything just because there is one guy that sounds smart published a paper that contradicts 100 different papers. Next week there will be a new super smart guy that comes and hell contradict this paper . When we started factor investing we knew there will be a lot of noise out there and there will be a lot of years where factors will underperform the market. To me all of this is noise to see if you are strong enough to stay on course.
The most successful factor is always changing, and some go extinct. You can't predict any factor will be more successful than another. He's just rehashing that what worked in the past isn't guaranteed to work in the future and that time in the market is better than timing the market. So stay your course if thats what works for you.
If factors don’t work why has international small value produced a premium for all these years? International as a whole has performed poorly and there is no reversion to fall back on. Nevertheless I’m too deep to exit now, but future contributions may go into TSM funds.
Andrew Chen crushes your dreams...
I'm glad I listened to this but the finding is mildly depressing. 😄
Whatbis the conclusion?
I don’t know about crushing dreams. Means a lot of people were wrong but at the same time many funds use a multiple factor approach such as value, profitability/quality and then use momentum for trading decisions. Most are very mindful of transaction costs.
Where Andrew’s study did not examine the implications of multiple factors working in combination…. Which of course would be a gigantic task.
Hey this is Andrew. My co-authors Mihail Velikov and Tom Zimmermann do look at this, and they both find that there are returns net of costs with ML combinations of predictors in recent years. But in the bigger picture, wouldn't these net returns get traded away, if many funds are pursuing them?
@@achenfinance-z2k I am no expert but I don’t think so. Many people pursue the market factor and it is still around. So anything that is risk based has to exist over long periods of time or at least we have to think it does.
If you could get the same returns in your checking account as the bonds and stock markets those markets would not exist.
Hey @@achenfinance-z2k I enjoyed the podcast! Depending on whether your believe the multi-factor returns are due to miss-pricing vs risk, that would determine whether they could become arbitraged away.
So if I am understanding you right, multi factor approaches like DFA or Avantis still generate excess returns after transaction costs, as long as the factors are risk based and not miss pricing? I’m going to look more into these papers
Wow: the stronger the theoretical underpinning for the factor, the worse the out of sample performance.
Really looking forward to a breakdown of this and how to better understand the implications. It just intrinsically makes sense that higher discount rate=more risk=higher expected returns
That's true, unless and it's just my opinion... As markets are efficient, the price you pay for that high discount rate creates a match between the return and the probability of you capturing that extra return for the extra risk. If all the world invests in the same value tilt portfolio, the return is driven away by the fact that the price of the components move higher, reducing the expected return. At the end, value was just a temporary misspricing that the efficient market already solved.
The answer seems to be all asset prices are, with various degree, the result of data mining and being fast about making trades. Additionally, the market is getting more efficient due to better tech. Thus, factor investing doesn't really have as much impact as was previously thought.
During the interview it is mentioned that data mining is perceived in a poor light. But the argument Chen provides is that they should be using Econ philosophy rather than the amalgamation of data.
My counter would be: an economist wants elegance, a trader wants cents.
I would love to see another intelligent and well spoken person debate Andrew, this all seems great but I feel like it would be useful to have someone highly knowledgeable debate these points.
You can see John Campbell discuss the paper here: th-cam.com/users/livezf1h3TpKk-s?si=2rSDVMuKZ2Qmfmjt
@@rationalreminder I don't mean to sound demanding, but would you mind giving a timestamp or general idea of where in that 8 hour video he talks about this paper?
To answer my own question: Andrew's talk starts at 7:06 and the response starts at 7:27
Freaking great interview
The interview I've waited for the past 6 months, as until now Ben's videos and Andrew's papers have been the most insightful sources of knowledge on investing in stocks for me.
After reading Andrew's papers I'm still torn between investing in an equally-weighted portfolio of stocks with a composite of strong data-mined accounting signals to scrape the remaining arbitrage off the floor or just buying the global market equally for the rest of my life. Maybe I'll do a bit of data-mined signal investing just for the future me who would second-guess this decision anyway.
Cameron’s lookin good today. Real good.
Is it his hair?
@@ousefk5476can’t quite put my finger on it, but something’s different about him today
there is a reason why we started audio only ...
@@cameronpassmore1561 hahah I’m just messing around. Hope you enjoyed your time off!
Tremendous work RR team.
What a beast! We Humans love Stories (and factors). Statistitians brutally destroying our nice logical theories and dreams of scientific persistent overperformance. Im finally going to consolidate my 20 etfs in just 1 😅
At 32.4 he drops the bombshell.
Great interview...lots to think about in my own investing strategy. it does increasingly seem like the markets are driven by liquidity and positioning more so than risk factors.
In timing the factor zoo (2023) they also find that timing factors by some data mined characteristics increases returns by another 1-2% if you take this into account plus returns by efficient transactions is is at least possible to outperform consistently. After a brief existential crisis, I now have higher expectations of myself and the complexity of the task. Really really distrupting results nevertheless, thanks for the Interview.
Textile strap with the green is an incredible pairing
Most of the factor return decay is specifically in cap weighted US stocks. If you go equal weight US or international, the vast majority of factors are pretty much as strong as they've ever been. A very poignant example is EBITDA/EV (historically one of the strongest value factors). Cap weighted last 10 year alpha in the US is -1.8% per year. Switch to equal weight and it's 8.4% per year.
Explanations can be that either S&P 500 specifically got way more efficient in the mid 2000s or that the crazy bull market and multiple expansion that the S&P 500 has seen in the past 15 years has somehow suppressed factor premiums. This plays somewhat into the Mike Green school of thought where factors can't really work if most capital just mindlessly flows to mega caps.
Whatever the reason may be, the lesson is stay away from cap weighted funds. Go US small and go international for factor exposure.
Are you counting the turnover cost of maintaining an equal weight portfolio?
@@matiasnicolaspizmeny I don't know much about how efficient firms like DFA or Avantis can be at maintaining low slippage with high turnover and capacity. When you are managing tens of billions I imagine it becomes a very difficult problem. But herein lies the greatest advantage of the retail investor. We are not capacity constrained. We can go after the biggest premiums in the smallest most illiquid microcaps.
AVUV has over 12 billion in assets. They cannot invest in some amazing factor premiums in a 10 million illiquid microcap. But you and I can, with virtually no cost and without impacting the price in any meaningful way. DIY is the future for the retail factor investor.
Ok so it doesnt make sense to include funds that invest in factors/predictors if youre looking for excess returns, but does it make sense to invest in them if youre looking to diversify? Can you be more diversified than a market cap weighted index with all the investable market? Would overweighting factors smooth out returns / shrink the distribution of outcomes, reducing the probability of ending up in either tail of the distribution?
Great episode btw
Gas Station Sushi Crew is concerned after this episode. ⛽️🍣
Cross-sectional asset pricing predictor replication meta-research keeping Ben's hair looking good.
Factors also reduce overall volatility of a portfolio dont they? That has value.
Yeah I thought the same they where only Talking about returns not even once die they say SR or volatility.
He sounds super intelligent, but some of the arguments he makes against others can be made against him.
He talks a lot about the stories authors want to make while he tries to make his own story about why the numbers look the they do.
I wouldn't change anything just because there is one guy that sounds smart published a paper that contradicts 100 different papers.
Next week there will be a new super smart guy that comes and hell contradict this paper .
When we started factor investing we knew there will be a lot of noise out there and there will be a lot of years where factors will underperform the market. To me all of this is noise to see if you are strong enough to stay on course.
The most successful factor is always changing, and some go extinct. You can't predict any factor will be more successful than another. He's just rehashing that what worked in the past isn't guaranteed to work in the future and that time in the market is better than timing the market. So stay your course if thats what works for you.
It is difficult to make predictions, especially about the future.
If factors don’t work why has international small value produced a premium for all these years? International as a whole has performed poorly and there is no reversion to fall back on. Nevertheless I’m too deep to exit now, but future contributions may go into TSM funds.
I’d wait for the adtershow next week before changing anything.
-Ben
@@rationalreminderwhere can I find this?
@@rationalreminder which episode do you mean by aftershow?