Papers & Statistics

As part of my role at a hedge fund we were constantly testing and searching for new trading strategies. The best resource I ever found, besides my past experience and monitoring real time trading, was BehavioralFinance.net . The site is absolutely loaded with academic research on the ways humans make errors in the market. As such, I was forced to read and understand academic research papers and PhD thesis. Now, I am not an academic, my highest degree is a BS in Mechanical Engineering. Some of the mathematics and presentations in these papers are quite rigorous and way over my head. However here are some things I found helpful:

1) Read the Abstract and Conclusion first. These papers are usually 40-50 pages long and no need to read it all until you establish some cause.
2) If the abstract/conclusion sound interesting, look for any mention of commission/slippage in their return estimates. They should specifically mention somewhere if they do. If not be very, very critical of the results.
3) Read the data section next. Good data is very expensive and hard to manage. Fundamental data especially (revisions, estimates, etc make it a nightmare) If you can’t get the data needed to replicate the strategy then why proceed?
4) Most strategies are based on an X period holding. You, my friends, are open to a whole host of other methods of testing. Most academic papers give you the basic edge on entry and it is up to you to develop exit, position sizing, and risk management. This is where your intelligence and creativity eventually payoff because unfortunately the academics spend little to no time on that. Their entire focus is entry and as alluded to in #2, very little with realistic trading.
5) Just be aware of the time period studied and the overall market regime during that period. The tech bubble or 2008 crash are large outlier events that can skew data.
6) Be aware of the universe of stocks utilized for testing, mega cap stocks trade very differently then small cap stocks.
7) Know your own size. Academics will poo poo an idea if the capacity of the trading strategy is too small. If your book is < $1MM this can leave a ton of space for you to generate a return as the big boys completely ignore it.

Don't believe everything you read and every headline. There is a lot of good information out there in the world of academia, but one needs to develop a practitioners "eye" to spot inconsistencies or opportunities in the published work. It is possible and in fact becomes quite easy with practice. Despite my handicap (being not so bright) I tend to think I can sum up a trading a strategy at one glance and can tell if it has merit and worth further investigation or needs work. This just comes with time and practice.

Here's an example of my thought process:

Bloomberg Article: Short the Rumors Pays 14% on Rumors that Don’t Work

“The surest way to profit from takeover speculation in the stock market is to bet it’s wrong.
Electronic news services, brokerages and newspapers reported at least 1,875 rumors about potential buyouts of 717 companies between 2005 and 2010, according to data compiled by Bloomberg. A total of 104, or 14.5 percent, were acquired, the data show. While stocks that were the subject of takeover speculation initially jumped 2.9 percent, betting on declines yielded average profits of 1.2 percent in the next month, an annualized gain of 14 percent.

Stocks tracked by Bloomberg fell 0.2 percent, 0.6 percent and 1.2 percent on average in the day, week and month following a rumor report, Bloomberg data show. “

Sounds pretty good? If you have access to some of the retail news wires, your familiar with takeover speculation on a daily basis almost. Here’s my thoughts as I read this:

1) The title of the report is confusing, “….on Takeovers Tales that Don’t Come True”. Possibly I’m reading too much into it but is it saying there is a 14% return when looking at ONLY stocks that aren’t taken over? Or does the return also include the few outliers that actually WERE taken over as well? Is this report the result of huge survivor ship bias? In other words, 100 stocks have takeover rumors, 2 were taken out, the 14% is only calculated from the 98 that were not taken out? Hopefully the question makes sense.

2) The day, week, month decline stated, is that measured from the highest point after the announcement or some set time period. For example, the stock is shorted 1 day after the announcement at the open and held for 1, 7, 30 days. This is huge because in real life there is no way to catch the top tick or even the highest close.

3) Commission and Slippage included? Depending on the frequency, this could be the difference between +14 and -14.

4) Is short cost or rebate also included in the calculation?

All in all I think it’s a fantastic statistic and a great place to start doing some research. Empirically I was looking at doing something similar as I noticed most traders (except the fastest to the news) tended to lose money chasing these headlines. There were some rare large winners and losers but mostly small losses. I contacted the reporter, Tara LaChapelle for more information regarding the data and still awaiting a response. I’m not sure if she’s allowed to but if so I will share here as well.

Advertisements

~ by largecaptrader on January 16, 2011.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

 
%d bloggers like this: