Quantitative Value - Gray & Carlisle

[ericd1]

[amazonsearch]Quantitative Value[/amazonsearch]

 

I'm left-brained, so this book caught my attention. It is an interesting read that attempts to dissect value investing based on a number of well-know value investors including; yes, Graham, Buffett and others. 

 

The book covers four main areas...

 

How to avoid stocks that can cause a permanent loss of capital

How to find stocks with the highest quality

The secret to finding deeply undervalued stocks

The five signals sent by smart money.

 

The authors develop a screening technique to identify potential candidates and they offer some back-tested models and there's a website that offers a monthly-updated screening tool to find stocks using the model and a blog about recent developments in quant value investing.

 

The authors start with Greenblatt's Magic Formula and build on it using their spin on Piotroski's F-Score, valuation ratios, smart money signals and other concepts. The resulting model when back-tested beat the market, which is the premises of the book.

 

I'm sure most on this board would argue with one of the strategy's main elements -- removing individual judgement from the process, but the book provides a compelling argument that most of the time individual judgment hurts performance. I would suggest the authors are addressing the "average investor" and not the "knowledgeable and highly experienced value investors" found here on the premier valuing investing hangout in the world!

 

I really enjoyed the reasoning presented behind several of Buffett's purchases. Although you can't know for certain if their analysis is totally correct, they make reference to his annual letters and their arguments make sense (although hindsight provides remarkable vision).

 

I don't think I would recommend it for a novice, but my guess is most reading this board would enjoy and benefit from reading Quantitative Value.  I read it quickly the first ime through and intend to re-read it again--starting tonight.

 

The book pulled together

 

 

[rukawa]

I am reading their book but there are some things I don't like. They have a strong tendency to just believe models. Their argument appears to be that models are more correct than experts so we should just trust models. And we shouldn't fiddle with models because that is always bad. But where exactly did these sanctified models come from? God?

 

But there are many problems with quantitative models. Models can be extremely bad. They never discuss the dangers of over-fitting a model. For instance their PROBM model they discuss is hugely complicated. Its difficult to understand the justification for the coefficients in the PROBM formula and its not clear if these coefficients remain stable if the model was fit with a different historical sample of data. I am hugely skeptical that this formula will work. They are way too credulous about all of this.

 

In addition models can be easily gamed and will be gamed as soon as they become popular. This appears to be the case with some of the earnings quality measures they discuss. Generally any econometric type model is highly suspect because the basic coefficients derived from the regression used can be unstable over time. This is because the economic system does not have a stable structure....its not like an atom. Its an organic evolving entity and so relationships that hold at one time can fail to hold at another. This point was made by Maynard Keynes but completely ignored by econometric researchers because it threatened their research which of course proved to be mostly garbage.

 

A good example is the Black-Scholes equation which is the single greatest social scientific model ever developed. I have tremendous respect for this model and think its quite remarkable. However its wrong because implied vols should be flat for all option times to maturity and strike prices. They aren't. There is a volatility smile observed in most markets where extreme strike prices have higher than expected volatilities. But here's the thing. It wasn't always so! When the model was originally developed the smile did not exist. Implied vols were constant across all strikes. The vol smile appeared I believe after the 1987 crash. The economy evolved. The model no longer worked. Traders realized that selling out of the money puts and calls could be hugely risky because huge market moves could occur more frequently than they expected.

[innerscorecard]

I am reading their book but there are some things I don't like. They have a strong tendency to just believe models. Their argument appears to be that models are more correct than experts so we should just trust models. And we shouldn't fiddle with models because that is always bad. But where exactly did these sanctified models come from? God?

 

But there are many problems with quantitative models. Models can be extremely bad. They never discuss the dangers of over-fitting a model. For instance their PROBM model they discuss is hugely complicated. Its difficult to understand the justification for the coefficients in the PROBM formula and its not clear if these coefficients remain stable if the model was fit with a different historical sample of data. I am hugely skeptical that this formula will work. They are way too credulous about all of this.

 

In addition models can be easily gamed and will be gamed as soon as they become popular. This appears to be the case with some of the earnings quality measures they discuss. Generally any econometric type model is highly suspect because the basic coefficients derived from the regression used can be unstable over time. This is because the economic system does not have a stable structure....its not like an atom. Its an organic evolving entity and so relationships that hold at one time can fail to hold at another. This point was made by Maynard Keynes but completely ignored by econometric researchers because it threatened their research which of course proved to be mostly garbage.

 

A good example is the Black-Scholes equation which is the single greatest social scientific model ever developed. I have tremendous respect for this model and think its quite remarkable. However its wrong because implied vols should be flat for all option times to maturity and strike prices. They aren't. There is a volatility smile observed in most markets where extreme strike prices have higher than expected volatilities. But here's the thing. It wasn't always so! When the model was originally developed the smile did not exist. Implied vols were constant across all strikes. The vol smile appeared I believe after the 1987 crash. The economy evolved. The model no longer worked. Traders realized that selling out of the money puts and calls could be hugely risky because huge market moves could occur more frequently than they expected.

 

Tobias said to me on reddit that he's now come full circle into advocating just the simple Magic Formula.

[CorpRaider]

I don't think he's totally abandoned the methods set forth in QV.  From the same reddit:

 

"A good screen can make the process better. I have customized screening software that uses a model built in the writing of Quantitative Value (QV) that sifts opportunities for me. I stick closely to the output of that model. The best idea is to figure out the big drivers of investment performance, build them into your model, and then stick to the model's output.

There's an idea in QV that the application of statistical prediction rules (SPRs) outperforms the best expert intuition. We think that the SPRs are the floor to which we add, but they're really the ceiling from which we detract. A model/screen is an SPR, and for the best performance we shouldn't "cherry pick" the output. The "Broken Leg" problem is an idea I discuss in Deep Value best illustrated by an example. Say we have an SPR that predicts when John goes to the cinema on Friday. He breaks his leg. Should we be allowed to override the SPR and incorporate this information into our decision? The answer is "no," and the reason is that, while it might be valid in any individual instance, we find more broken legs than there really are, and exercise our discretion to override the SPR too frequently. In investment, lots of undervalued stocks have broken legs, but that's often the reason that they're cheap. If the SPR says 'buy,' I do."

 

 

[CorpRaider]

Just thought I would share a link to the valueshares site.  Wes Gray launched the first of his ETF based on the methodology in the book which is the subject of this thread.  Looks like a lot of MF stocks in the portfolio.

 

http://www.valueshares.com/

[innerscorecard]

Just thought I would share a link to the valueshares site.  Wes Gray launched the first of his ETF based on the methodology in the book which is the subject of this thread.  Looks like a lot of MF stocks in the portfolio.

 

http://www.valueshares.com/

 

If I'm not mistaken, I believe this is separate from any ETF that Tobias Carlisle may or may not launch with him as the fund manager.

[peter1234]

Just thought I would share a link to the valueshares site.  Wes Gray launched the first of his ETF based on the methodology in the book which is the subject of this thread.  Looks like a lot of MF stocks in the portfolio.

 

http://www.valueshares.com/

 

If I'm not mistaken, I believe this is separate from any ETF that Tobias Carlisle may or may not launch with him as the fund manager.

 

I think you are right.

 

They wrote the book together but lead separate investment firms.

[CorpRaider]

Yeah just Wes Gray.  The professor one with the PHD from Chicago.  The one that runs the turnkey analyst and alpha architect sites.

[KCLarkin]

Are there any similar magic formula style ETFs? Greenblatt's funds look great but they are expensive.

[CorpRaider]

Not that I'm aware of.  Well, Blackrock has a couple that are quality-value active etfs with low expense ratios: IELG, IEIL, IEIS and one other (domestic and foreign large and small cap ones).  They don't give a ton of detail on the methodology but if you nose around a bit you will find a paper published by one of the guys from Blackrock's quant research department that sam lee from M* said was the basis for the funds.  If memory serves, they use low price to book and low accruals/high cash component of earnings, for the value and quality proxies. 

 

I've just been tracking them so far.  QVAL is going to be funded with $50MM from Alpha architect that they have been running as seperately managed accounts which appear to be based on the same strategy and they have some short performance history records for the strategy on their website. 

 

I'm ok with the Gotham fees, I mean I would have given my pinky to get into his prior fund and pay him 2 and 20 and let him compound at 50% per year (David Geffen has said Greenblatt made him more money than he made for himself, hah), but I'm still not comfortable with the long short strategy.  I mean I actually heard him say (or maybe read it) that they looked at doing that in the initial MF research and it did great until it "blew up" or something to that effect. 

[KCLarkin]

The long-short strategy is much safer than a long-short Magic Formula. With only 20 names, it is easy for one bad short to blow up the fund.

[CorpRaider]

Yeah, maybe that's the difference but my recollection was that he talks about it after he shows the breakdown between deciles in the backtest data.  Like "why not buy decile 1 and short decile 10?"  That sort of a thing.

 

I haven't tried to find the quote I am thinking about because, as you note, the turnover is insane and its hard to get $250K built up in a tax advantaged account (or at least it has been for me so far).

[KCLarkin]

Your recollection is correct. I do remember Greenblatt saying that about MF specifically.

[crastogi]

Yeah, maybe that's the difference but my recollection was that he talks about it after he shows the breakdown between deciles in the backtest data.  Like "why not buy decile 1 and short decile 10?"  That sort of a thing.

 

I haven't tried to find the quote I am thinking about because, as you note, the turnover is insane and its hard to get $250K built up in a tax advantaged account (or at least it has been for me so far).

 

It is correct.  That is what is puzzles me too - he is doing what he panned in the past

[EliG]

I took a quick peek at QVAL holdings. The number of retailers is disturbing. 7 out of 40 equally weight positions, or 17.5% of assets.

[west]

I took a quick peek at QVAL holdings. The number of retailers is disturbing. 7 out of 40 equally weight positions, or 17.5% of assets.

 

Uh oh.  Maybe someone isn't capitalizing operating leases?

 

But then again, I wonder if his back-testing capitalized operating leases when it got its market beating returns?...

[CorpRaider]

You could go to the turnkey analyst list and see if you can reverse engineer the EBIT/TEV ratios for some of the retailers.  Lots of healthcare, retailers, energy and unloved chemical companies.  Interesting!

 

I would also point out he's got some real world performance data on the strategy (coming up on two years worth) from the SMAs on the advisor's site.

[rainman]

Where can I find this ETF's holdings.

 

Thanks

[CorpRaider]

There's a downloadable file/spreadsheet on the website.

[ni-co]

I took a quick peek at QVAL holdings. The number of retailers is disturbing. 7 out of 40 equally weight positions, or 17.5% of assets.

 

Uh oh.  Maybe someone isn't capitalizing operating leases?

 

But then again, I wonder if his back-testing capitalized operating leases when it got its market beating returns?...

 

What's the point in doing this for a simple formula? Don't forget that operating leases are not only a liability. There is an RE asset against them that you'd have to capitalize too.