Rabu, 29 Juli 2020

Market-Cap Together With Growth-Value Arbitrage

Predicting whether small-cap or increment stocks volition outperform large-cap or value stocks inwards the side yesteryear side quarter is a favorite pastime of fiscal commentators. To many fiscal economists, however, the interrogation is long agone settled yesteryear the so-called Fama-French Three-Factor Model. This model postulates that the returns of a stock depend mainly on three factors: the full general marketplace position index returns, the market-cap of the stock, as well as the book-to-price ratio. Furthermore, every bit an empirical fact, over the long term (i.e. for whatever 20-year period), small-caps shell large-caps yesteryear an average compounded annual charge per unit of measurement of 3.12%, as well as value stocks shell increment stocks yesteryear 4.06% (the latter effect applies when nosotros restrain ourselves to the large-cap universe).

This model is really convenient to us arbitrageurs. Statistical arbitraguers to a greater extent than ofttimes than non don’t know how to predict marketplace position index returns, simply nosotros tin soundless brand a living inwards a ship marketplace position yesteryear buying a small-cap, value portfolio as well as shorting a large-cap, increment portfolio, as well as human face to earn 3-4% (on one-side of capital) a year. For example, despite the much anticipated imminent demise of small-caps over the final twelvemonth or so, I establish that if nosotros long the small-cap value ETF IJS, as well as brusk the large-cap increment ETF IVW from Nov 15, 2005 to Nov 15, 2006, nosotros would induce got earned nearly 10% return. The 3-4% average returns human face meager, simply greenback that since this is a market-neutral, self-funding portfolio, your prime number broker (if y'all merchandise for a hedge fund or a proprietary trading firm) volition let y'all to leverage this furnish several times.

Some traders volition honor twenty years a fighting likewise long. Is at that topographic point whatever aid from academic theory on whether small-cap value volition outperform large-cap increment side yesteryear side month, as well as non side yesteryear side twenty years? Influenza A virus subtype H5N1 latterly published column explaining this inwards the New York Times recently.) The substance of this article is that when marketplace position persuasion is positive, human face small-caps to underperform large-caps yesteryear 0.26% a month, as well as value stocks to outperform increment stocks yesteryear 1.24% a month. Conversely, when the marketplace position persuasion is negative, human face small-caps to outperform large-caps yesteryear 1.45% a month, as well as value stocks to underperform increment stocks yesteryear 1.04% a month. How 1 computes “sentiment” is complicated: it is a linear combination of half dozen variables: closed-end fund discount, NYSE portion turnover, number as well as first-day returns on IPOs, equity portion inwards novel issues, as well as the dividend premium. (The authors used information from 1963-2001 for this study.) Now, without genuinely computing all these variables, most would grip that the electrical flow persuasion (as of Dec 2006) is fairly positive. This implies, every bit Mr. Hulbert noted, that small-cap volition underperform large cap inwards the coming months, opposite to the long-term trend. However, the other long-term trend, that value volition shell growth, volition soundless grip inwards the close future. It is upward to the reader to honor a distich of ETF’s that volition accept maximum payoff of this prediction, simply I volition aid hither yesteryear tabulating unopen to of the available funds.







 ValueBlendGrowth
Large capIVEIVV/SPYIVW
Mid capIJJIJHIJK/JKH
Small capIJSIJRIJT


Further reading:

Bernstein, William (2002), The Cross-Section of Expected Stock Returns: Influenza A virus subtype H5N1 Tenth Anniversary Reflection.
O’Shaughnessy, James P. (2006), Predicting the Markets of Tomorrow. Penguin Books.

Selasa, 28 Juli 2020

Dna, Cryptology, Speech Communication Recognition, Together With Trading

There is an interesting New York Times article on a mathematician too cryptologist who used to operate for the wildly successful hedge fund Renaissance Technologies too is forthwith famous for decoding DNA's. This article caught my eyes because quite a few of my erstwhile colleagues from the speech communication recognition question grouping at IBM likewise went over to Renaissance equally researchers too portfolio managers. Renaissance is an extraordinary hedge fund inward Long Island that has an average annual provide of 35% since 1989, after charging 5% administration fee too 44% incentive fee. They profess to hire exclusively scientists, engineers too mathematicians amongst equally piddling background inward finance equally possible. They started off trading futures, simply has since too so diversified into equities models, too is artificial intelligence may intend that I am beingness inconsistent here, equally I was less than enthusiastic virtually the application of that bailiwick to trading. There is, however, quite a big departure betwixt data theory too artificial intelligence. The erstwhile is characterized past times sophisticated theory amongst really few parameters, the latter, unproblematic theory amongst a lot of parameters.)

There is i published trading model that is based squarely on question inward data theory. It is called Universal Portfolios, created past times Stanford data theorist Prof. Thomas Cover. It is an elegant too quite intuitive model, simply I don't know how good it performs nether realistic conditions. I promise to write virtually about of my question on this too a related degree of models inward a time to come article.

Further reading:

Cover, Thomas M. too Thomas, Joy A. (1991), Elements of Information Theory. John Wiley & Sons, Inc.

Another Limitation Of Artificial Tidings In Addition To Information Mining

Sometime agone I espoused my views that AI in addition to information mining techniques may non endure suited for predicting fiscal markets. Here nosotros convey an article from the Chief Scientist at IBM's Entity Analytic Solutions Group who believes these techniques are non check for counterterrorism either. Why? The same reasons I mentioned: non plenty historical data.

Senin, 27 Juli 2020

Do Element Models Live On Inwards The Curt Term?

Besides pair-trading, “factor model” is the close pop workhorse of the statistical arbitrageur. In a previous article, I discussed the close well-known ingredient model – the Fama-French Three-Factor model, alongside the full general marketplace index returns, the market-cap of the stock, too the book-to-price ratio equally the exclusively iii factors driving returns. However, equally I explained earlier, this ingredient model has a really long horizon. For the quantitative trader who needs to brand coin every month, the natural instinct is to expect for a to a greater extent than “sophisticated” ingredient that plant inwards the brusk term, or fifty-fifty to railroad train roughly variety of model that role dissimilar factors every calendar month inwards answer to “market condition”. Alas, other than hearsays too second-hand gossips, I accept never witnessed an actual success of this approach inwards a hedge fund or proprietary trading grouping – at to the lowest degree a success that lasts for to a greater extent than than a year.

I am of course of didactics non privy to the electrical flow surgical procedure numbers of ingredient models run past times roughly of the close successful hedge funds today. However, in that location is a shape of ETF (called “XTF”) marketed past times PowerShares Capital Management that uses a similar ingredient approach for its stock pick criteria. According to media reports, each stock inwards these XTF’s is scored past times 25 variables such equally cash flow, earnings growth, cost momentum, etc. This sounds similar a classic ingredient model to me. This model is reportedly designed past times the quantitative unit of measurement at American Stock Exchange. To abide by out if they accept indeed discovered the holy grail of ingredient models, I looked at the surgical procedure of these XTF compared to their benchmarks.

Here I tabulate the XTF’s for each marketplace cap too value category, their corresponding benchmark marketplace index ETF’s, too finally the YTD differential returns upwards to Dec 13, 2006. (PJG too PJM accept equally good brusk a history for this comparison.)










ValueBlendGrowth
Large capPWV-IVE=4.8%PWC-IVV=-3.6%PWB-IVW=-5.0%
Mid capPWP-IJJ=0.1%PJG-IJH=N/APWJ-IJK=3.1%
Small capPWY-IJS=-0.7%PJM-IJR=N/APWT-IJT=-4.9%




The differential returns are all over the place: roughly positive, others negative. To me, this is symptomatic of a ingredient model that does non accept predictive power. (After all, if the differential returns are consistently negative, nosotros could accept long the ETF, brusk the XTF, too brand consistent profits!) At the really least, this ingredient model may accept a horizon much longer than what close traders would live interested inwards – inwards which case, why non but role the uncomplicated Fama-French model?

This is non to tell that exotic, proprietary ingredient models accept no use: they tend to live pretty useful for peril management, equally volatilities too correlations are ofttimes easier to predict than returns. But beware every fourth dimension your peril administration software vendor tries to sell yous an alpha generator!

Minggu, 26 Juli 2020

Universal Portfolios

Let me depict a portfolio optimization scheme that, over the long run, is supposedly guaranteed to outperform the best stock inward the portfolio.

Before nosotros begin, let’s concord that nosotros volition rebalance our portfolio every hateful solar daytime so that each stock has a fixed pct resources allotment of capital, exactly equally your favorite fiscal consultant would receive got advised you. What this way is that if you lot ain IBM as well as MSFT, as well as IBM went upwardly later on 1 hateful solar daytime whereas MSFT went down, you lot should sell some IBM as well as role the working capital missive of the alphabet to purchase some to a greater extent than MSFT. There is a technical term for such portfolios: they are called “constant rebalanced portfolios”. Notice also the similarity amongst the Kelly criterion which I wrote about before: Kelly criterion asks you lot to hold a constant leverage, which is similar maintaining a fixed pct resources allotment betwixt cash (debt) as well as stock.

But what should the fixed pct resources allotment be? Here is where the scheme gets interesting. Suppose nosotros start amongst an equal working capital missive of the alphabet allocation, for lack of whatever amend choice. At the cease of the day, your portfolio has a sure enough internet worth. But as well as so you lot tin terminate calculate what the internet worth would receive got turned out if you lot had started amongst a unlike allocation. Indeed, nosotros tin terminate run this simulation: elbow grease all possible initial allocations, as well as calculate the hypothetical internet worth of the resulting portfolio. Use these hypothetical internet worth equally weights (after normalizing them yesteryear the core of all internet worth), as well as compute a weighted-average pct allocation. Finally, adopt this weighted average resources allotment equally the novel desired resources allotment as well as rebalance the portfolio accordingly. So genuinely the “fixed” pct resources allotment is non fixed after-all: it gets adjusted daily, but in all likelihood non yesteryear much. Repeat this procedure everyday, e'er calculating a novel weighted resources allotment yesteryear simulating diverse initial allocations since hateful solar daytime 1.

This scheme of portfolio optimization tin terminate hold upwardly proven to attain a internet worth greater than exactly belongings the best stock, given long plenty time. If this sounds similar a miracle, it is partly because this is inward fact an ingenious effect of data theory, as well as partly because at that topographic point are diverse caveats that genuinely throttle its practical application. The proof that it plant (at to the lowest degree inward theory) is rather technical as well as I volition allow the interested reader peruse the master copy articles): nosotros acquire by to lose coin fifty-fifty when a toll serial exhibits a geometric random walk. So it is non likewise surprising that nosotros tin terminate also brand coin using similar data theoretic juggling.

Now for the caveats. Every fourth dimension an data theorist start proverb “In the long run, …”, you lot volition hold upwardly well-advised to ask: How long? In my geometric random walk example where the volatility (standard deviation) of returns every menses is 1%, nosotros detect that the compounded charge per unit of measurement of furnish is an agonizingly pocket-size -0.005% per period. In the instance of the universal portfolio scheme, the out-performance over the best stock inward the portfolio is similarly subject on the volatilities of the stocks: the higher the volatility, the faster the out-performance. Let me run a simulation amongst a portfolio consisting of 2 ETF’s RTH as well as OIH. If nosotros were to run the Universal Portfolio scheme from 2001/5/17 – 2006/12/29, I detect that the cumulative furnish is 32% (without transaction cost). Contrast that amongst exactly buying-and-holding the best ETF (namely OIH here): the cumulative furnish is 54%. The Universal Portfolio loses. Does this hateful the theory is wrong? Not really: RTH as well as OIH may exactly receive got likewise depression volatility. Herein lies the foremost practical caveat amongst the Universal Portfolio scheme: it tin terminate receive got likewise long to realize its attain goodness if the volatility is low.

How attain nosotros detect ETF’s that receive got high plenty volatility to realize the out-performance of Universal Portfolio? Actually, nosotros tin terminate but boost the volatility of RTH as well as OIH artificially yesteryear increasing their leverage. So let’s tell nosotros leverage both of them 2x. This way their daily returns as well as volatilities are both doubled. Now the best ETF (which is nonetheless OIH here) has a furnish of 23% (why is it lower than the un-leveraged case? Remember the formula m-s2/2 inward my previous article.) , but the Universal Portfolio has a furnish of 45%. So at 1 time the Universal Portfolio wins. But this is a Pyrrhic victory: if you lot element inward a transaction cost of 10 footing points, the Universal Portfolio scheme genuinely returns alone 4%. This is the minute caveat of Universal Portfolios: because of the frequent rebalancing required, transaction costs tend to consume upwardly all the out-performance.

Now at that topographic point is a concluding caveat. The reader may inquire why I don’t exactly pick 2 stocks instead of 2 ETF’s to illustrate this scheme. Aren’t most stocks to a greater extent than volatile than ETF’s as well as thence much amend suited for this scheme? Indeed, most academic papers, including Prof. Cover’s master copy paper, role a yoke of stocks for illustration. But if nosotros attain that, nosotros run the guide a opportunity of introducing survivorship bias. Naturally, if you lot know ahead of fourth dimension that none of these 2 stocks volition become bankrupt, the Universal Portfolio scheme may expect great. But if you lot run a simulation where 1 of the stocks all of a abrupt went bankrupt 1 hateful solar daytime (which tend to hold upwardly a fairly mathematically discontinuous affair), the Universal Portfolio scheme volition most probable non trounce belongings exactly the non-bankrupt stock inward the beginning. Using ETF’s eliminated this problem. But as well as so ETF’s are far less volatile.

So given all these caveats, is Universal Portfolio genuinely practical? Prof. Cover seems to holler back so. That’s why he has started a hedge fund to essay it.

Jumat, 24 Juli 2020

Factor Models: The Ground Continues...

A reader JR only posted approximately real thoughtful comments on my article on element models. You tin rank the axe read his comments in addition to my reply here.