I recently came cross Dan Mirvish’s Hathaway Effect article. In it, the author presents his “discovery” of a hidden relationship between Anne Hathaway and the stock price of Warren Buffet’s Berkshire Hathaway:
When Anne Hathaway makes headlines, the stock for Warren Buffett’s Berkshire-Hathaway goes up.
Using no less than six instances (trading days when Berkshire Hathaway went up concurrently to Anne’s making headlines) as “strong evidence” for a correlation between the two events, Mirvish then takes a stab at a causation hypothesis:
My guess is that all those automated, robotic trading programming are picking up the same chatter on the internet about “Hathaway” as the IMDb’s StarMeter, and they’re applying it to the stock market.
I am not so sure the $125,000.00 BRK.A stock is subject to much high frequency trading from these robots – although some trading robots most likely scan some internet chatter for trading decisions, as discussed in the entertaining Nerds on Wall Street for example.
Actually I literally just came across a very real example, as discussed in that book, with another article relating how a hedge fund (Derwent Capital Markets) is launching today – on April Fool’s Day (!) – aiming to track the internet chatter (from Twitter actually) to identify the “online mood” to predict the stock market, based on this paper (PDF).
In any case, the evidence from Mirvish’s article is only anecdotal and can easily be dismissed as the probable results of random occurrences, or even worse, potentially biased data selection: Anne Hathaway has probably been making as big headlines (as the six examples used in the article) in other occasions. Why these are not included?
Of course, this sort of “bogus analysis” makes for good headlines and viral sharing – and I have to admit I am falling prey to it right now. I did find the mock film poster amusing and thought you might do too… But it does highlight a general lack of understanding of statistics and skepticism by public and journalists alike – an idea nicely discussed in Fooled by Randomness or The Drunkard’s Walk.
Another Correlation: A More Serious Study on Trend Following amongst CTAS
To end on a more serious note, there always seems to be some confusion between the terms Managed Futures, CTA and Trend Followers, which are often used interchangeably.
I recently came across a paper from Newedge (“Two benchmarks for momentum trading” – PDF) which establishes that a strong link exists between CTAs and Trend Followers, as the correlation between their respective returns at 0.97. Also an interesting comment on the concept of diversifying away from Trend Following with non-correlated strategies:
That is, trend following is a major force in this industry. The fortunes of trend followers and the industry tend to rise and fall together, although with one significant difference. While the two paths follow one another closely, the total index path exhibits much lower volatility than that exhibited by the trend following sub-index path. The lower volatility stems from the presence of low correlation returns, and the result is a higher return/ risk ratio for the industry than for the trend followers by themselves.
The paper also presents a new Trend Following index (Newedge Trend Indicator), based on a similar concept to the State of Trend Following Report or the benchmark created by Conquest Group (discussed here), using a mechanical system instead of tracking underlying CTAs. The paper discusses several points of consideration for this index such as market selection, contract rolls, liquidity or transaction costs. An interesting read.
You can download the paper from this page, which also contains an interview of the paper author (Galen Burghardt).