Since robots are cheaper than humans expect smart beta to beat the indices

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
sometimesinvestor
Posts: 1207
Joined: Wed May 13, 2009 6:54 am

Since robots are cheaper than humans expect smart beta to beat the indices

Post by sometimesinvestor » Fri Mar 31, 2017 6:54 pm

https://www.bloomberg.com/gadfly/articl ... -beta-bots

or consider

http://www.marketwatch.com/story/blackr ... 3-30/print



Even john bogle sees a problem with bond index funds. The trick is obviously to decide if a value or growth strategy is best for a given year than program appropriately

User avatar
Taylor Larimore
Advisory Board
Posts: 27405
Joined: Tue Feb 27, 2007 8:09 pm
Location: Miami FL

Robots vs. The Three-Fund Portfolio ?

Post by Taylor Larimore » Fri Mar 31, 2017 7:03 pm

sometimesinvestor:

I linked to MarketWatch and was pleased to learn that Allan Roth's "Second Grader Three-Fund Portfolio" continues to lead Paul Farrell's eight expertly designed portfolios for the past 1-year, 3-years, 5-years and 10-years.

http://www.marketwatch.com/lazyportfolio

Past performance does not forecast future performance.

Best wishes.
Taylor
"Simplicity is the master key to financial success." -- Jack Bogle

User avatar
stemikger
Posts: 4902
Joined: Thu Apr 08, 2010 5:02 am

Re: Since robots are cheaper than humans expect smart beta to beat the indices

Post by stemikger » Sat Apr 01, 2017 4:43 am

Smart Beta is Dumb ~ Jack Bogle
Choose Simplicity ~ Stay the Course!! ~ Press on Regardless!!!

User avatar
k66
Posts: 494
Joined: Sat Oct 27, 2012 1:36 pm

Re: Since robots are cheaper than humans expect smart beta to beat the indices

Post by k66 » Sat Apr 01, 2017 6:13 am

Hmmm... a couple of trains of thought:

A) Evolution of the markets continue; humans are fully replaced by Smart-B 'bots, and they all now routinely "beat the market". Except they are now collectively the market, but continue to outperform by the old standard(s). So the passive indexer (the handful of us that are left) should expect higher returns as a result (except we now score at the 50th percentile rather than our customary 80th) because 'bot fees are now so dramatically low?

B) Will 'bot strategies demonstrate persistence? Would not the presence of multiple strategies all actively playing the market simultaneously still create the necessary "chaos" that prevents generalized persistence of any one strategy? Or are all smart-beta strategies so completely aligned that they would all just converge to the same performance point? In that case, I suppose the passive indexer is still relegated to the 50th percentile expect that there would be smaller dispersion in the field and not much difference between the 1st and 99th percentiles that it doesn't matter much?

C) Will Wall Street really up all those lucrative fees or will they find another way to extract them? Maybe this is the real question!
LOSER of the Boglehead Contest 2015 | lang may yer lum reek

afan
Posts: 3840
Joined: Sun Jul 25, 2010 4:01 pm

Re: Since robots are cheaper than humans expect smart beta to beat the indices

Post by afan » Sat Apr 01, 2017 6:30 am

An active management firm is not in the investment business. It is in the marketing business. Whether individual stock picking, factor bets, smart beta or AI, they make money by convincing people to hire them. So they always need a pitch to get new business in the door and to hold on to what they have.

Factor bets have been around for decades. Why haven't active managers as a whole used them to outperform the indexes? Their expenses are too high. Smart beta is a newer marketing term for the same factor bets. Changing the name does not change the performance. AI is potentially different. The portfolios might be different. But the net of all this will be the market return, available essentially for free from index funds. You could pay the universe of AI firms to give you the index return, minus their fee. Or you could get the index return for free by cutting out the active managers.

Interesting, isn't it that Black Rock has $5 T under management but only $275B in active funds. Yet that is where they make their money, hence all the excitement. Why else would it be news what they are doing with 6% of their assets?
Money in their pocket on the active side is money out of investor pockets.
We don't know how to beat the market on a risk-adjusted basis, and we don't know anyone that does know either | --Swedroe | We assume that markets are efficient, that prices are right | --Fama

avalpert
Posts: 6313
Joined: Sat Mar 22, 2008 4:58 pm

Re: Since robots are cheaper than humans expect smart beta to beat the indices

Post by avalpert » Sat Apr 01, 2017 11:23 am

sometimesinvestor wrote:Since robots are cheaper than humans expect smart beta to beat the indices
The conclusion does not obviously follow from the premise. Why should robots being cheaper then humans impact the returns of 'smart beta' versus 'indices'? (and isn't smart beta often implemented via an index?)

User avatar
sometimesinvestor
Posts: 1207
Joined: Wed May 13, 2009 6:54 am

Re: Since robots are cheaper than humans expect smart beta to beat the indices

Post by sometimesinvestor » Sat Apr 01, 2017 12:37 pm

avalpert wrote:
sometimesinvestor wrote:Since robots are cheaper than humans expect smart beta to beat the indices
The conclusion does not obviously follow from the premise. Why should robots being cheaper then humans impact the returns of 'smart beta' versus 'indices'? (and isn't smart beta often implemented via an index?)
\
Sorry if the articles did not make this clear. Many pros outperform the market but not by more than the amount of their fees. One reason active management was successful in say the early 80s was that managers had access to better info than the public. Changes in the laws have made this less true.If all those in the market are knowledgeable its that much harder to be above average. THe theme of the articles was that computers may be able to do better than humans in many cases and since computers don't have a high salary the expense ratios may go down increasing the chances for outperformance.Obviously the stocks in the S+P 500 are not selected for performance but for market cap and liquidity.. It is therefore reasonable that some mangers armed with computer programs may be able to pick better performing stocks than the staff of Standard and Poor.The criteria they use to select stocks would produce a portfolio that in current usage would be the product of smart beta. No, I don't much like that term either.

avalpert
Posts: 6313
Joined: Sat Mar 22, 2008 4:58 pm

Re: Since robots are cheaper than humans expect smart beta to beat the indices

Post by avalpert » Sat Apr 01, 2017 12:52 pm

sometimesinvestor wrote:
avalpert wrote:
sometimesinvestor wrote:Since robots are cheaper than humans expect smart beta to beat the indices
The conclusion does not obviously follow from the premise. Why should robots being cheaper then humans impact the returns of 'smart beta' versus 'indices'? (and isn't smart beta often implemented via an index?)
\
Sorry if the articles did not make this clear. Many pros outperform the market but not by more than the amount of their fees. One reason active management was successful in say the early 80s was that managers had access to better info than the public.
Except this is false and represents a misunderstanding of why you expect a passive index to outperform active investments. To be more precise, the expected return of active investors is the same as passive investors less fees - but with more volatility. So sure, to the extent that you make the active funds cheaper the expected returns will approach the passive funds (of course they to can execute cheaper due to robots) but because of the volatility you will still be better off with an index of an active fund.

But none of that has anything to do with 'smart beta' which is supposed to be a passive approach anyway.

It is no more obvious that a computer can be expected to consistently outsmart the market than a human can - what makes the computer better at predicting the future?

User avatar
Dale_G
Posts: 3088
Joined: Tue Feb 20, 2007 5:43 pm
Location: Central Florida - on the grown up side of 81

Re: Since robots are cheaper than humans expect smart beta to beat the indices

Post by Dale_G » Sun Apr 02, 2017 1:35 pm

avalpert wrote:It is no more obvious that a computer can be expected to consistently outsmart the market than a human can - what makes the computer better at predicting the future?
Especially when there are thousands of computers with similar, but at least slightly different algorithms, trying to do the same thing. Some of them will be right some of the time.

I'll stick with indexing for equities. Most of my bond funds are actively managed by Vanguard however.

Dale
Volatility is my friend

anil686
Posts: 751
Joined: Thu May 08, 2014 12:33 pm

Re: Since robots are cheaper than humans expect smart beta to beat the indices

Post by anil686 » Sun Apr 02, 2017 4:08 pm

What seems to make sense in theory does not always hold true in practice. Schwab's fundamental index mutual funds seem to have very similar returns over a 10 year period to Vanguard's cap weighted counterparts. maybe that is because growth has done better than value during the fundamental index fund's existence or maybe it is because overtime things balance out. Markets change over time and over time, while some factors may be persistent, they may underperform for periods (sometimes long periods) of time. Unless an investor has conviction to stay with the strategy (in addition to the fancy marketing) they run the risk (like all investors) of bailing out at the wrong times and underperforming a vanilla approach. I have started to believe the biggest difference in investing (when you get to very low ERs that I will define arbitrarily as under 0.40%) seems to be behavioral. The ability to stick with any approach at very low cost will likely lead to good results - jumping on and off different horses will likely not... JMO though...

Starper
Posts: 104
Joined: Sun Oct 26, 2014 10:02 pm

Re: Since robots are cheaper than humans expect smart beta to beat the indices

Post by Starper » Sun Apr 02, 2017 6:00 pm

The Bloomberg article compares S&P500, active management and value strategy.
Now compare that with a simple quantitative value strategy compiled by Dartmouth professor Ken French that selects the cheapest 30 percent of U.S. stocks by price-to-book ratio and equally weights among them. That strategy returned 23 percent annually over the same period, with a standard deviation of 17 percent.
According to the article, value strategy outperforms S&P500 by a huge margin.

I'm not familiar with this strategy. A quick Google search doesn't provide a lot of useful information either, but maybe somebody more knowledgeable can respond to the following comments:

-Has anybody ever done out of time and/or out of sample validation of this strategy? It's easy to claim that your strategy beats the market if you use the same time period to train your model and demonstrate that it works. How about using let's say 1960-1990 data to develop your strategy and then using 1990-present to see if it works? How about using the US data to develop the strategy and then using the UK stock market data to validate it?

-Even if the strategy does work based on out of time and out of sample data, is it guaranteed to work going forward given that the stock market is a random walk? Predictive models work reasonably well in what Nassim Taleb calls "mediocristan": for example, what kind of people are more likely to respond to marketing campaigns given their responses to the same or very similar campaigns in the past. I'm not sure this is the case for "extramistan" environments (another word made up by Taleb) such as the stock market.

-Let's assume for a moment this strategy works perfectly well and can beat passive investing consistently. How does the efficient market hypothesis affect its performance given that the strategy is in the public domain? For example, according to value strategy, if we all invest in the cheapest 30% of stocks by price to book ratio, we are guaranteed to multiply our returns vs the S&P500. If this is the case and we all know it, then according to the efficient market hypothesis, the stock prices of these 30% of stocks will adjust upward until the risk-adjusted return for this strategy matches one for the S&P500 (taking investment fees into account as well). This is why am skeptical of value strategy or any other active/market timing strategy whether it's manual or algorithmic, especially if it's known to the general public.

I'm not an expert in any of this, so please feel free to poke holes in my arguments. I'd like to learn more about this subject. Even though I'm skeptical, I'm keeping my mind open.

avalpert
Posts: 6313
Joined: Sat Mar 22, 2008 4:58 pm

Re: Since robots are cheaper than humans expect smart beta to beat the indices

Post by avalpert » Sun Apr 02, 2017 7:20 pm

Starper wrote:The Bloomberg article compares S&P500, active management and value strategy.
Now compare that with a simple quantitative value strategy compiled by Dartmouth professor Ken French that selects the cheapest 30 percent of U.S. stocks by price-to-book ratio and equally weights among them. That strategy returned 23 percent annually over the same period, with a standard deviation of 17 percent.
According to the article, value strategy outperforms S&P500 by a huge margin.

I'm not familiar with this strategy. A quick Google search doesn't provide a lot of useful information either, but maybe somebody more knowledgeable can respond to the following comments:

-Has anybody ever done out of time and/or out of sample validation of this strategy? It's easy to claim that your strategy beats the market if you use the same time period to train your model and demonstrate that it works. How about using let's say 1960-1990 data to develop your strategy and then using 1990-present to see if it works? How about using the US data to develop the strategy and then using the UK stock market data to validate it?

-Even if the strategy does work based on out of time and out of sample data, is it guaranteed to work going forward given that the stock market is a random walk? Predictive models work reasonably well in what Nassim Taleb calls "mediocristan": for example, what kind of people are more likely to respond to marketing campaigns given their responses to the same or very similar campaigns in the past. I'm not sure this is the case for "extramistan" environments (another word made up by Taleb) such as the stock market.

-Let's assume for a moment this strategy works perfectly well and can beat passive investing consistently. How does the efficient market hypothesis affect its performance given that the strategy is in the public domain? For example, according to value strategy, if we all invest in the cheapest 30% of stocks by price to book ratio, we are guaranteed to multiply our returns vs the S&P500. If this is the case and we all know it, then according to the efficient market hypothesis, the stock prices of these 30% of stocks will adjust upward until the risk-adjusted return for this strategy matches one for the S&P500 (taking investment fees into account as well). This is why am skeptical of value strategy or any other active/market timing strategy whether it's manual or algorithmic, especially if it's known to the general public.

I'm not an expert in any of this, so please feel free to poke holes in my arguments. I'd like to learn more about this subject. Even though I'm skeptical, I'm keeping my mind open.
This is simply describing a way of capturing the value and small risk premiums. It is well studied and has been shown across time and geographies. It is a popular approach on this forum - it can also be accomplished using passive funds.

A search here for value tilt will give you more than enough material to learn about the subject.

Post Reply