R2G DESIGNERS - READ THIS!!!!!!!!!!!!!

Research
1

This is a recent featured column on the Advisor Perspectives web site. We’re not in it because the authors/researchers haven’t yet found us. But it is an attack on what we do.

http://www.advisorperspectives.com/articles/2015/08/18/why-you-shouldn-t-trust-most-financial-research

We here have had many discussions, debates, arguments, etc. on the strategy-design process in the area of quant practices vs reliance on fundamental theory, etc. You all know what I think and there’s no need to re-open these debates right now.

But as we enhance and improve R2G (and as quiet as it may seem to many of you, there’s a heck of a lot of activity going on in Chicago with Marco and our regular team and with terrific work being done by three summer interns and frankly, i think it’s terrific stuff – be patient) with he idea of scaling it (making it more useful and attractive to more people hopefully to translate to more subscribers), there is going to be an important cost: A lot of the kinds of conversations etc. we’ve been having in house are going to be brought out into the outside world and we have to expect discussion of P123 and R2G not only in our own forums but in other venues as well.

We hope the bulk of the discussion will cast us in a positive light, and I’m already working with Forbes (a very valuable friend of p123) to try as best we can to shape the agenda (see http://www.forbes.com/sites/marcgerstein/), and my Forbes content is also reproduced on hvst.com, talkmarkets.com and in my Seeking Alpha InstaBlog. But no matter what I or anyone else does, we cannot ever hope to control the agenda. So we have to be prepared for others to publish and promote unfriendly views. This is normal. The bigger and better a company is, the more detractors it attracts. Amazon, Apple, Microsoft, the New England Patriots . . . that’s life.

The Adviser Perspectives article is one, and just the most recent, example of opinion we have to expect. And as we grow, the odds are about 100% that those who feel this way will start looking at and commenting publicly on us.

Personally, I find the Adviser Perspectives article and the research on which its based to be horrifically flawed – in fact, downright ridiculous.

I’ll contact Adviser Perspectives this week to offer a rebuttal. I’ve never worked with them, but an RIA I know well writes for them and hopefully can point me in the right direction. If I am unable to rebut there, then I’ll do so elsewhere: I certainly can and will on Forbes and other sites to which I syndicate my Forbes content and I’m now in the very early stages of starting to build a relationship between P123 and Institutional Investor Journals (publisher of “Institutional Investor” and many high-end journals that publish serious research such as “Journal of Portfolio Management,” “Journal of Trading,” etc.), a very prestigious outfit that has discovered us. So we have and are getting allies in the outside world and rest assured we won’t have to sit back passively if spitballs come our way, as we know they will.

But beyond what I can do, am doing, and will do to enhance our broader stature, I encourage you to make yourself aware of what’s being said out there on the sort of things we do with the aim of being able to produce and describe models that can withstand skeptical scrutiny on the part of knowledgeable outsiders with access to big soapboxes. If anybody wants to discuss any aspect of this an would feel more comfortable doing so privately, that’s fine too.

2

Marc,
I have published many articles at AP. The editor is Robert Huebscher. email: rhuebscher@advisorperspectives.com. I am sure he will publish your comments.

I have read this article myself. There was no attack on P123 in the article. This statement in the article cannot apply to large-cap P123 models - there are simply not enough subscribers, however it is true for small-cap models, in my opinion:

“If an investment strategy that beats the market is discovered and verified – even if it is not a spurious discovery – it will not work for everybody. It cannot; it is tautologically obvious that not everybody can beat the market. Hence, a strategy that is identified as effective, correctly or not, must, at some point, if it becomes popular and widely adopted, stop working and may even reverse to become a bad strategy.”

3

This has been discussed at length, as you mentioned. The best rebuttal I’ve read so far has been from Cliff Asness: https://www.aqr.com/cliffs-perspective/it-is-not-data-mining-not-even-close

To be fair, there are certainly many factors that are the result of data mining, as illustrated by article (e.g., 17th monthly lag in U.S. Industrial Production). But the record shows that a handful of factors (e.g., value, momentum) have demonstrated robustness over a very long period of time, across markets, and across asset classes.

4

Criticism of or argument for P123?

After reading this article I want to do 2 things:

  1. examine the statistics of a strategy carefully with full knowledge of potential statistical pitfalls

  2. use strategies that are not widely used by others

Now where can I go to do this? Hmmm. That’s right: P123!

BTW, this is a good argument for not breaking any of the statistical abilities of P123. There is much truth in the article and it is hard enough the way it is.

Of course, this article is a strong argument for P123 only if major changes are not planned for P123.

5

If P123 wants to compete on a larger stage, they must be promoting large-cap models (only), with very simple concepts / well defined stock universes and models should be documented in the same fashion as RAFI indices http://www.ftse.com/analytics/factsheets/Home/Search
No more info than that (certainly not in-sample backtest).

If backtest stats are provided as R2G currently does, P123 will be ridiculed/criticized. This has happened at some investment review sites. It is not good for P123’s reputation.

Take care
Steve

6

All:

  I agree with Stitts and geov.  R2G's should be limited to large-cap, high liquidity stocks and ETF's.

Bill

7

I disagree.
Why not give “the little guy” the opportunity to benefit from “low liquidity” models. It is simply a tradeoff: We know that low liquidity has its severe limitations (slippage etc.). But on the other side there is the chance to make exceptional gains which cannot be made with large liquidity models.

I have 2 ports with low liquidity. They are my best models in terms of alpha. Difficult to trade but still possible.
Therefore, everybody has a choice here. And that is great asset of P123!!

8

I think what P123 does is very useful and I put my money where my mouth is on this.

Furthermore, I will let the R2G designers decide what the future P123 will look like.

With regard to the article, it is not much different than the efficient-market hypothesis. It is actually less strong than the efficient-market hypothesis in that it allows for the possibility of factors that can beat the market but argues that they are nearly impossible to find and that they will not persist.

Even if we get a few Nobel Prize winners on our side, I don’t think we will make much headway in converting those who believe in the efficient-market hypothesis – they have their own Nobel Prize winners. And yes, those Nobel Prize winners do admit that a few factors may have a small effect.

9

I’m a bit time constrained today but wanted to offer a few comments on what’s been said by others:

RE: large cap models and micros - I want to focus manily here for now. We won’t be excluding anybody. Those who are satisfied maintaining and even offering new low-liquidity models will still be able to do so. Realistically, though, looking at R2G in its entirety, economic reality dictates that if it truly meets its potential, the lion’s share of subscriber revenue will be coming from high-liquidity models. I don’t want to speculate now on what the boundary will be, but common sense suggests large-cap offerings will be critical to our success. We hope designers will choose to fill the product gap, and our new platform will make it easy for designers to see where the product-line gaps are so they can jump in with offerings. The good news with large cap is that because it is what it is (no constraints on number of subs), we’ll be fine is only a few choose to offer product. Steve is a large-cap advocate and from his comment here,it’s obvious he’ll be participating in our growth. I’ve seen other interesting large-cap R2Gs that haven’t gotten the attention they deserve that will, I expect, also be part of the growth and hopefully, their less-outspoken designers will be inspired to offer more along those lines. I intend to offer high liquidity product as well. Bottom line: R2G designers will still be able to choose to do as they wish, View the coming emphasis on large cap as an invitation to participate in growth, not as a rule you must follow.

The Efficient Market Hypothesis is out there and will always remain out there. But it’s not nearly as widely revered as it once was, even in academia. The opposition to it has become much more effectively articulated and at this point, its main champions are those marketing so-called passive funds (see, e.g. Vanguard). I’m not at all bothered by the persistence of those debates and am happy to continue playing my role in them. Debates aside, though, EMH advocates are not something p123 and r2g need lose sleep over.

Geov, thank you for that contact and Alan thank you for the Asness link. I will;most definitely use both.

The main threat posed by the article and others like it is the attack on a research process, data mining. Those who do it are putting themselves into a firing line, not from cranky old me and some other forum participants, but from a heck of a lot more in the outside world.

10

I’ve read quite a few academic research papers, and there are many cases where I’ve been unable to replicate the results on P123. I suspect that this is because the authors were evaluating market data over different time periods when different factors appeared to work, or perhaps they weren’t viewing the results in the same way that I do. I’d be surprised if a lot of the papers were statistically flawed in the simple way the article suggests. Something I cherish about P123 is that it gives me the ability to re-check the work of others - I don’t have to blindly trust most financial research. Also, I have a great deal more faith in the results I get from P123 than a lot of the guff that’s written about the stock market around the web.

I note that the article stops short of saying that ALL factors are bogus, and that the market is completely efficient. Are the authors saying that all the investigations that have highlighted the effectiveness of value factors is bogus? I think they should clarify their own stance.

11

Marc & All - the main problem is that if P123 offers smallcap models alongside largecap models, the uneducated investor will choose smallcap every time even though that environment is pretty saturated. It becomes very difficult for largecap models to thrive.

The reason I mentioned RAFI fact sheets is because they don’t discuss their fundamental algorithms at all. (I believe) the methodology is proprietary. The only text I have found so far is “These factors include dividends, cash flow, sales and book value.” Despite not revealing the underlying fundamental algorithms, the ETFs based on RAFI are apparently quite popular. From a mainstream perspective, it might be wise to document our models in the same fashion as RAFI does, although probably a lot less detailed.

Steve

P.S. My 11 year old daughter is preparing for school and she just made up a binder with a quote on the front cover. I hadn’t heard this one before but I think it is appropriate for the efficient market hypothesists…

“Everyone is a genius. But if you judge a fish by its ability to climb a tree it will live its whole life believing that it is stupid” - Albert Einstein

12

We need 2 kinds of people: those who share the same view (potential clients) and those who don’t (to take the other side of the trades). I prefer a few in the first category and a lot in the second.

13

Hi Marc,

thanks for pointing out the article and your efforts to a constructive reply. As with all things in life, there will be a range of opinions - pro and contra. I’m not surprised that the first author is a mathematician - and as such has a difficulty with the rather irrational behaviour of the stock market.

Also good to know that you have a voice at Forbes; they are a far more professional shop than where the article was published.

Cheers,
Florian

14

I’m tempted to agree; i’ve said often in the forums that there’s a difference between a personal model and a public model. But I think the reality is that we can and should do as much as we can to protect people from hurting themselves, but ultimately, no matter what we or anybody else does, the market kamikazes are going to find ways to do what they want to do. Did i ever mention the theme of the novel I’m trying to write? My protagonist is a Bernard Madoff type, but my guy does something the real guy stupidly didn’t do. My character fully and in elaborate detail explains the details of his Ponzi scheme in a Registration Statement filed with the SEC. (This is not a spoiler, this comes out in the first half of the novel.) I honestly believe that had Madoff done this, he would still have been able to raise just as much money as he actually did. That’s the way people are.

R2G’s original design was way too kamikazee-friendly. That’s been well discussed in the forums. No need to rehash now. Going forward, in the spirit of never say never, i can’t promise that a kamikaze won;t still be able to find ways to crash their portfolios. But they are really going to have to put in some serious effort to accomplish that. The sort of searches they will be doing will be completely different. Patience.

As a side note, i showed Forbes what R2G is, they’re intrigued, and I have carte blanche to talks about it and link directly to model pages. Rest assured, though, is that I won’t discuss any model that isn’t scalable (i prefer that word to large cap, although I suspect the correlation will be high, albeit not necessarily 1.00). That goes for models by p123 members. It also goes for my own models. (I’m getting my feet wet by talking publicly about Cherrypicking the Blue Chips and when I discuss other models, I may contact designers privately to discuss method and how much they’d be comfortable with me saying publicly – and I hope descriptions will be improved to RAFI-like approaches so as to make it less problematic.)

Again, for those who are into small and micro cap, i don;t have a bias against this. I do it with my own money, and even write about the small cap effect not as a statistical phenomenon (which it may or may not be) but as a set of substantive fundamental characteristics. But as for the R2G business, scalability is important. Small caps are find, if they are sufficiently liquid. It means my low-priced stock model, which has delivered significant alpha even in five years out of sample, is a no-go for public discussion and we agreed with Forbes to close the newsletter and focus instead on things that can be disseminated more broadly. So I’m practicing what I preach about R2G scalability.

Love it. I guess that’a why he gets to actually be Einstein.

15

Marc - how do you find the time to write novels?

Here is a dumb thought. (Some relief from today’s market troubles): With all the stem cell / DNA research going on, and Einstein’s brain in a bottle, do you think a big pharmaceutical company will eventually come up with an Einstein pill? ie replicate Einstein’s brain cells and find a way to introduce them into another’s brain.

16

So what’s the big deal about that article? The author makes valid points. It’s definitely not an attack on P123. The backtest overfitting issue is very real, and P123 has never encouraged running thousands of meaningless simulations until one gets lucky.

17

Marc,

In terms of Forbes and ‘Institutional’ types or family offices. Likely gonna need a 3 year out-of-sample track record with no ‘survivorship bias’ - and likely an audited record of trading money on this with some ‘seed account’ P123 ran. And a whole new set of rules. Things like:
a) Min. 30 (or 50) holdings
b) Class that turns over once a year (all LT gains), but has dynamic hedging overlays.
c) Class that turns over however much designers want.
d) Will likely need to upgrade the book functionality significantly (so people do things like set Vol. targets and change expsoure rates based on rules and/or use and vary leverage).

As far as retail investors, they will likely buy anything marketed well, and promising huge returns for no risk.

Best,
Tom

18

It’s not like I can confine all my fiction to the forums. :slight_smile:

If it’s done, it would be cool to see how many commentators dismiss it saying pharma research can’t be trusted.

We know that. But by portraying the situation as if it’s what everybody does, it erroneously casts us in a bad light? That needs to be corrected and that’s what I’m doing.

19

This goes into the “If any of this really worked, everyone would do it and it would no longer work” line of thinking by the original author. Well, can everyone really do it? Can everyone really trade low liquidity models, especially? Seems as an individual retail investor this is the one advantage I have over institutional investors. I can trade in lower liquidity equities without blowing out the price on the trade for the volumes needed to move the needle for larger institutional investor. Obviously it’s not a good idea to put a disproportionate amount of total net worth into highly volatile assets like a port of micro/small cap equities, but if anyone could take advantage of it it should be “the little guy”.

In general, “If it really worked, everyone would do it” could be applied to value investing. “If buying low and selling high really worked, everyone would do it”. Well, everyone can’t do it, because it’s generally not in most people’s temperament. Most people don’t have the temperament to faithfully follow a fully automated modeling processs, especially if the fully automated modeling process has periods of underperformance to rattle confidence and give people reason to abandon the process. This is anecdotal observation I have in my career as a software developer, dealing with users and their inclination to distrust the “machine”. If an automated software process works thousands of times without error, but bombs out 1 time every 20,000 tasks, the user will often declare it unreliable, useless and scraped. Meanwhile, humans manually doing the same task will be lauded for their consistency, reliability and thoroughness with just a fraction of the reliability. No matter how many hours Google successfully logs on their self-driving car without incident, people will still find it too unnerving to just passively trust a machine to drive. This will be an psychological barrier it could take generations to fully overcome, especially as we experience anomalies in which the self driving car technology has accidents (and it will have accidents). Yet we will reflexively jump into the back of a cab with some sleep deprived (possibly innebriated) human cab driver, without any information or background on their driving history, and mindlessly play on our smartphone while turning over life and limb to a human driver without giving it second thought.

Data mining and curve fitting is no doubt the bane of my existence, and should be rightfully highlighted warned against, but at least I’m aware (almost hyper aware) of it and do everything I can to account for it and hopefully guard against it. This is opposed to the legion of innate biases and logical fallacies I’m vulnerable to if I’m just flipping through pages of ValueLine evaluating companies, reading annual reports/balance sheets and other “sound” methods of conventional stock picking.

20

I also allocate a portion of my total portfolio to low liquidity ports. My thought is that low liquidity stocks are ignored by the hedge funds with armies of smart analysts. The “little guy” can go where the big boys can’t, and p123 becomes the analyst for the little guy.

Overall, my use of P123 for small stocks has outperformed the market, so I’m happy. However, I have abandoned most of the models, both private and R2G, that I put real money into. Most of these models lost me money, but one outperformed by so much that I’m up overall. Definately high-risk high-reward. This is important and I think newbies that just stumbled onto this site might not understand.