Busting the Myth About Size - Small Cap outperformance overstated?

Seemingly many of models people use here have a big size component but the study below implies small cap out-performance is overstated. Explanations vary but include it’s just “a poor man’s” value measure, slippage is understated, and especially that databases inaccurately account for delisted stocks.

http://advisorperspectives.com/commentaries/research_121614.php

Even with the expanded “variable commissions”, slippage in the small-cap oriented portfolio I trade appears significantly larger than the model states.

Apparently small-cap out-performance has been shrinking (?), but their conclusion of it can best be realized by using rules-based value and momentum strategies, at least, do “play well here”.

I have not found that to be the case with a port where the Liquidity of the bottom 20% is equal to $1,522,000 (as done for R2G testing), making $3,000 trades using Folio.

Where do you trade, what is the Liquidity of your bottom 20% and what percentage of the ADT do you trade?

The small cap effect lost a lot of its outperformance after the original study’s findings became public many years ago. Comparing small cap outperformance before and after the original study publication date supports this hypothesis.

Not sure where you’re getting this. It’s always tricky to make performance comparisons because so much depends on the time period chosen. But
I’ve not seen academic research demonstrating the death of the small cap effect. Meanwhile, we can see at least something on P123. Here’s a simple ETF screen:

ETFAssetClass=equity
ETFCountry=usa
ETFMethod=stanlong
ETFStyle=genstyle

version a: ETFSize=smallcap version b: ETFSize=largcap

Max backtest with all default assumptions in place except choose iShares 3000 ETF as benchmark (a common inclusive universe against which to compute alpha):

                    verson a                  version b

Annl Ret 6.80% 5.71%
Annl SD 22.98% 20.13%
Annl Alpha 1.49% 0.63%

Obviously, anyone could define sub-intervals that would show different results, especially now, given that smaller caps have been experiencing the negative aspects of the increased volatility that is just as much a part of the conventional wisdom as the returns.

Most importantly, neither the return nor risk characteristics are numerical phenomenon per se. Both are common-sense fundamental phenomenon being confirmed by numerical measures.

Small caps are different. The stocks differ in terms of the coverage and attention, making for less efficient information flow. The companies differ in terms of their ability to grow, not just due to company lifecycle issues but also due to operating characteristics (less diversified business portfolios, more streamlined company structures and heavier operating leverage). Such factors contribute to small-cap outperformance during normal times but undferperformance, sometimes sharp underperformance, when the market becomes more risk averse. The main period of structural large-cap outperform,ance was the 1980s-90s, when large firms get very aggressive about streamlining unwieldy corproate structures and attacking costs, thus addressing years of laxness in that area. None of this is new.

As to slippage and trading problems, for most investors, these are not troublesome for investors who simply pursue small-caps, as opposed to those who pursue the most extreme . The main exception would be those who pursue the teeny-weeny itt-bitty sub-micro nano-caps which were not the ones considered by those who identified and wrote about the small cap effect.

It depends on which study you (want to) read. You can find publications and “proof” for almost any stock market thesis.
There is no black and white. For just about any cross-section of expected stock returns you can produce a thesis, a timeframe and a market that matches your hypothesis.

Here’s just one brief summary of the history of the size effect, taken from a 2006 article by Marqueringa/Nissera/Valla.
This excerpt covers the period from when people started to believe that the size effect has vanished - only to conclude that the effect has now resurrected.

As to your question, the basic (what some call scientific) approach would be to slice the S&P 500 into buckets and choose the highest ranked according to whatever effect you want to study. Here is a very simple test for a 50 stocks long portfolio ranked by the (lowest) market cap, rebalanced yearly:

The tests also yields positive results with 300 stocks from the Russell 3000.

I think what counts in the end is to have a core understanding of how the market and stock valuation works and to stick to your conviction - without closing your eyes to the things around you (that’s the tricky part).

That said, I am pretty happy with the performance of some of my R2Gs that partly incorporate the upside of smaller sized stocks, like RoT Munich and RoT Osaka.

Best,
fips

See, for a whole bunch of charts on this:
http://people.stern.nyu.edu/adamodar/pdfiles/invphiloh/growthN.pdf

See charts pages 7-10. And page 18 for international effects.

Huge variations over time in the return attributable simply to size. There are cycles where small caps underperform (the 1970’s) and times when they crush it. The premium volatility and range out outcomes is much higher since the publication.

However, having said that, the biggest winners and losers are still likely to come from this cap range. If you have a long time horizon and real skill (or have built a system with real skill), still likely to make the most in the smallest corners of market cap over most rolling 10-15 year periods. However, over 5 year or less periods, anything can happen. If you lack real skill, likely to lose the most here. Many quant ‘factors’ also show much greater outperformance in the small cap space - there are many papers on this. And manager skill has also been shown to pay out the most in the least liquid markets.

True for every thesis in most fields. Anything can be proven using logic.

Truth 1: inexpensive HMOs are rare
Truth 2: everything that is rare is expensive
Therefore: inexpensive HMOs are expensive

This type of reasoning is common. Lawyers and lobbyists are masters in the craft.

Jrinne = ~“What’s my liquidity?” = All my sims/ports require at least $100mm market cap and $5 price minimum, but I’m showing my age here, because this use to be the minimum definition of small cap… perhaps I need to move it up, but a lot of what I see, at least where people are “competing for show”, is minimums of >$1 and >$50mm. I originally considered $500,000 ADT sufficient, but had to move it up as well as create formulas to account for 1-time spikes in volume. This improved liquidity but my slippage is still higher than expected (something not only the study I cite mentions but also the study Tomyani cites).

Mgerstein = “Not sure where you’re getting this.” = Research Affiliates is Rob Arnott’s firm and if you don’t know him, then you and I may have been “in the business” during the same period, but it was different businesses (fwiw: I’ve been on the buy-side my whole career, until I retired in 2004).

Tomyami = I saw both Rob Arnott (~”my” study) and Aswath Damodaran (“your” study) at the same quant conference in NY years ago and they were the two people that impressed me the most (besides Peter Bernstein, who did not present but was there in a “Chairman Emeritus” role). And Aswath’s study, at least to me, supports the questionability of small cap performance.

Aurelaulel = your 2nd truth contradicts your 1st truth (if everything that is rare is expensive, there can be no rare, inexpensive HMOs). And from my meager education in logic, it is my understanding it works more like the scientific method in that it does NOT prove anything per se, rather it disproves “the null hypothesis”, leaving you to ASS-U-ME your hypothesis is (or may be) correct. That’s why I am devoted to the logic-loving (libertarian) view of the Austrian school of economics (I don’t believe anything in economics can be proven, only logically deduced), and most importantly Schumpeter, who emphasized entrepreneurship and innovation (forte’ of small business, aka small-caps), and especially Schumpeter’s belief that he could be “proven” wrong because, after all, things change (constantly, and change was central to his belief that economic systems were “organic”, not static).
“Edit” PS: Schumpeter, IMHO, is being proven more and more right every passing day (unfortunately).

And to all, I posted this here as “food for thought”, but I feel like Ben Bernanke at a goldbug conference. I’m relatively new to P123, but I’ve known quant before it was quant, and there are cycles to all models. It just happens I joined P123 near a top (perhaps), and currently my results do not show the results that long-term testing show. However, that is exactly a point made by Damodaran, who shows a slide that it takes, what, 15-20 years for small-cap outperformance to “be definitive”. For those of you who don’t understand what impact this has, I would urge you to read an instablog I wrote on Seeking Alpha, just google “What’s Your Decade, Man?”.

Finally, everyone should take note of my last sentence in the original post, which I feel most have missed: “…small-cap out-performance… can best be realized by using rules-based value and momentum strategies.” I don’t know about the rest of you, but that’s why I’m here!

Fips = I’d need to find a free copy and the time to study the study you cite, but the article/study I cite does point out there are a lot of flawed small-cap studies out there. And I know Arnott and Damodaran (by reputation).

Hi, did not read all answers.

I monster heavily go pro on small caps, especially microcaps!

The model I use, uses a liquidity filter that low: AvgDailyTot(20) > 25000

I made 24% (+12% on a strong dollar, but that was a windfall profit!) on small caps (+ value + momentum + 100 Stocks + 3 Month rebalance) the last 12 Months.

Total Return 24.88%
Benchmark Return 6.10%
Active Return 18.78%
Annualized Return 22.87%
Annual Turnover 82.82%
Max Drawdown -4.63% (!!!)
Benchmark Max Drawdown -13.18% (!!!)
Overall Winners (118/194) 60.82%
Sharpe Ratio 2.41 (!!!)
Correlation with Russell2000 0.64

Comission and Slipage was no problem at all, slippage overall was positive (using limit and GTC orders, 70% fills first day, it takes 3 Weeks until all
orders (with some tweaking the limit prices) are filled) so far, I could replicate the above in a 300k (e.g. 3k per Stock) Tradestation account!

Regards

Andreas

Judge - the true test isn’t when times are good but when they are bad. If it takes 3 weeks to execute a trade now, then what will happen in a strong bear market? You may not be able to exit the market. Many of these submicrocap stocks will completely disappear.

But anyways, good luck with your investments.

Steve

Ah yes, I should have been more focused when I saw Research Affiliates. I’m definitely familiar with Arnott’s work. He’s an interesting guy but to me, his “brand” is starting with his ultimate conclusion (fundamental weighting is the way to go) and then working backwards. So yeah, it make sense that he’d slice-and-dice in a way that suggests small caps aren’t so great. He has to. He’s selling bigger-is-better, albeit with the spin that he’s using factors other than market cap to measure size. (FWIW, I like fundamental weighting – it is a good idea to eliminate market-induced distortions from measurement of size – but even the better definition he offers doesn’t change the inherent differences between large and small.)

Steve,
Time will tell :slight_smile:
Regards
Andreas