Jump to content

Building Custom Financial Apps for Alpha


schin
 Share

Recommended Posts

I was listening to this Malcolm Gladwell podcast and one tidbit peaked my interest.... The PGA database they were talking about and how if you overlay it with hours on the golf course relative to CEOs of companies -- More hours on the links might be a predictor of bad stock/business performance. Makes sense, but never would have crossed reference this because I am not a golf.

 

Is this some correlation that Renaissance HF would leverage? Is there value in create a system to actually do this? I have friends who creates government enterprise systems - they might be able to create these systems for us.

 

Are there any finance guys that want custom apps create for them? Maybe, parse SEC data accordingly. I know there are a lot of website for insider trading and 13/F holding docs... are there other needs that can generate alpha?

 

If so, I can pair you up with the right techie.

 

http://revisionisthistory.com/episodes/11-a-good-walk-spoiled

Link to comment
Share on other sites

Here's the (most) relevant part of the paper:

 

Operating performance appears to suffer when CEOs spend more time away from work, but this doesn’t necessarily mean that there will be a noticeable impact on market valuations. Since stock values reflect the marginal investor’s expectations of future cash flows, the relationship between CEO golfing levels and firm value should be stronger if CEO leisure is expected to persist. We investigate persistence for high quartile golfers (reported in Table 9) and find an autocorrelation coefficient of 0.68 (t-statistic < 0.001). Given that CEO golf is highly persistent, it is reasonable to expect that investors will assign lower market valuations to those fimrs where CEOs are expected to shirk their responsibilities in the future.

 

To evaluate the link between shirking and market value, we regress Tobin’s Q onto control variables and variables that capture the amount of golf a CEO plays in a manner identical to that reported in Table 4 (for ROA). We are also concerned about endogeneity in this context, so we conduct additional 2SLS regressions that instrument for the amount of golf CEOs play with the number of non-cloudy days.

 

The results presented in Table 7 Panel A support our hypothesis regarding the impact of shirking on firm value. We find Tobin’s Q is higher for firms that are smaller, younger, more profitable, have higher dividend payout ratios, more independent boards, and are members of the S&P 500. Using the continuous variable Number of Rounds in column 1, we find that the coefficient estimate is -0.00223 (p-value=0.048). Consistent with results in Table 4 for operating performance, we find that it is the most frequent golfers (Quartile 4) that are associated with lower Tobin’s Q. When we include only the indicator Quartile 4, we find a coefficient estimate of -0.109 (p-value=0.028), indicating that firms with CEOs that are the most active golfers are associated with a Tobin’s Q that is almost 10% lower than other firms in the sample.

 

Consistent with analyses presented in Tables 4 and 6, we implement an instrumental variable approach in order to account for the potentially endogenous relationship between CEO golf play and firm value (Tobin’s Q). Using the number of non-cloudy days to instrument for CEO golf play, we present the results of our 2SLS estimation in Panel B of Table 7. First stage regressions are presented in column 1 for the continuous golf variable Number of Rounds, and in column 3 for the discrete variable Quartile 4 golf. In both regressions we find that the number of non-cloudy days is significantly correlated with the amount of golf play, consistent with earlier reported results. Second stage regressions presented in columns 2 and 4 provide a reasonably clear picture that larger amounts of CEO leisure cause lower firm value. Using the fitted value for Number of Rounds we find a coefficient estimate of -0.064 (p-value=0.103). Alternatively, using the fitted value for Quartile 4 golf, we find a coefficient estimate of -1.725 (p-value=0.037). While the coefficient presented in column 2 is slightly outside of conventional levels of statistical significance (p-value=0.103), when evaluated together with the evidence using Quartile 4 golf, we believe that a very reasonable conclusion is that high levels of CEO golf do indeed lead to lower firm values. We also include AR statistics in columns 2 and 4, which are 11.40 (p-value<0.001), in order to alleviate concern over any bias that might be introduced in the second stage regression.

 

This doesn't feel like a super robust, alpha-generating finding. This does feel like the sort of thing that produces an abstract that gets Malcolm Gladwell excited, based on a methodology he probably doesn't remotely understand. Which isn't a dig at Gladwell; I don't think I understand exactly what's being said here. At least I know for a fact I have no way of mapping this onto any notion of how much of an edge this would imply. It seems like if it implied a substantial edge, business school/finance researchers would be motivated to actually present a backtest.

Link to comment
Share on other sites

There is already an industry around this, it's called "Alternative Data".  Check out a company like Quandl.

 

Hedge funds will pay a lot of money for these alternative sets.  They integrate them together and look for indicators.

 

The value isn't in building an app, it's in finding some sort of alternative data, building a structure out of it and selling it.

 

For example, if you could build a network of webcams that read license plates and then leased spots on busy roads or around busy banks you could then buy DMV data and track movement.  This would be valuable because you could track who comes and goes to certain destinations at what time and that could be correlated with other data.  That would be valuable.

 

Link to comment
Share on other sites

There is already an industry around this, it's called "Alternative Data".  Check out a company like Quandl.

 

Hedge funds will pay a lot of money for these alternative sets.  They integrate them together and look for indicators.

 

The value isn't in building an app, it's in finding some sort of alternative data, building a structure out of it and selling it.

 

For example, if you could build a network of webcams that read license plates and then leased spots on busy roads or around busy banks you could then buy DMV data and track movement.  This would be valuable because you could track who comes and goes to certain destinations at what time and that could be correlated with other data.  That would be valuable.

 

Now, that seems to be getting closing to the 'Billions' episode where they are using satelite images to detect inventory at car dealer lots... pretty interesting concept you wrote above.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...