Renaissance / Simons and Net-Nets

[snailslug]

I've been running into 13Gs and 4's from James Simons' Renaissance fund a lot recently in microcaps that are net-nets.  Anyone else experience this phenomenon?

 

Maybe the fancy computer algorithms and Graham's ideas about net-nets are really not that far apart!  Different path, but same outcome!

 

I know this is probably one of only many strategies at Renaissance but just thought it was interesting.

[Guest kawikaho]

Where do you get those filings?

[Cardboard]

I wish we could know more about this man. His performance is simply amazing and I thought that he would get hit last year, like so many quant funds, while he delivered over 80% to his investors in his Medaillon fund. And if I am not mistaken that is after a 50% management fee!!! How can someone with such a large fund switch so fast into a short mode? What kind of instruments did he use to get this kind of return for his overall fund?

 

On the net-net thing, there is one small cap that I own since early last year which he owns too: QLT Inc. It is a biotech trading around net cash generating enough profits to support its current research on a promising drug delivery system. It was a liquidation play and kind of still is. The problem is that they are now defending a new lawsuit which could eat a fair bit of that cash. What surprised me is that Simmons owned it before me and he is still an owner. So it means that he is not rotating all his holdings that quickly. Kind of surprised also that the news of the lawsuit did not force him to kick it out. His computer would tell him that this increases risk and could reduce returns.

 

Cardboard

[snailslug]

Where do you get those filings?

 

http://www.sec.gov/edgar/searchedgar/companysearch.html

 

Put in the stock symbol and look at the 4 or SC 13D / SC 13G forms

[snailslug]

I wish we could know more about this man. His performance is simply amazing and I thought that he would get hit last year, like so many quant funds, while he delivered over 80% to his investors in his Medaillon fund. And if I am not mistaken that is after a 50% management fee!!! How can someone with such a large fund switch so fast into a short mode? What kind of instruments did he use to get this kind of return for his overall fund?

 

I think, as value investors, it's not really that important for us to understand how something like Simons' fund or another successful quant guy, Ray Dalio's Bridgewater, works, because unless you have a PhD in Math like these guys, you probably can't replicate it.  However, as a rational person, a la Munger, this is indeed a Lollapalooza in the making, so I think it's of interest just simply from a rational, philosophical point of view.

 

My understanding of these kinds of funds is that they traffic in all asset classes, from commodities, to equities, futures, anything that's liquid and easily tradeable.  I read an interview with Simons where he said Buffett's approach is making a bunch of money on a concentrated bet, while his is like picking up pennies.  I believe what they do is hire Math and Physics PhDs, many with no finance background, who comb through historical securities trading data looking for strong statistical correlations between price movement and certain variables.  The reason they probably hire the PhDs is that finding statistical correlations is similar to finding correlations in science experiments.  Also, note that they do not generally hire MBAs, they hire people who probably actually understand what sigma means, which probably reduces their blow-up risk.

 

For example, with the net-nets, they probably found a correlation between stocks selling for less than cash and their prices going up over time, which is exactly what Graham found, but he found that using reasoning rather than statistical analysis.  But, microcaps??  In many of the cases I've run into, they file a 4 because they own 10% or something, and the entire cap is like $30mn, so it's a $3m position in a $4bn fund (for Medallion)...talk about picking up pennies!  I also read that they lever up about 3-4X. 

 

So, spread out a whole LOT of small sized, positive expected value risk-adjusted bets that are market neutral that is agnostic towards asset class, and have the computer trade, eliminates the problem of diworseification.  Appears to be a rational approach to me, but difficult to implement for mere mortals.  Makes sense that they would do well in an environment when there is a lot of volatility though if this is really the approach. 

 

Final point, note how different this approach is from other "quant" hedge funds that blow up.  For example, the Bear Stearns funds that blew up were, buy a whole bunch of one thing that you think is very "safe" due to your "formula" (CDOs) and lever up like mad.  I believe (but am not sure) that LTCM suffered from a similar idea, buy a bunch of trades on interest rate spreads only and then get margin called.

 

Anyways, like I said, I don't think it's of much practical use, to me anyways, but I just find it fascinating.

[NumquamPerdo]

I also find this fascinating and , to be honest, too good to be true.

[nodnub]

....while his is like picking up pennies.  I believe what they do is hire Math and Physics PhDs, many with no finance background, who comb through historical securities trading data looking for strong statistical correlations between price movement and certain variables.  The reason they probably hire the PhDs is that finding statistical correlations is similar to finding correlations in science experiments.  Also, note that they do not generally hire MBAs, they hire people who probably actually understand what sigma means, which probably reduces their blow-up risk.

 

Does this remind anyone else of what worked so well for LTCM?  I am sure they thought all their Nobel Laureates understood sigma too. But they didn't realize that their models relied on unrealistic assumptions.

 

[watsa_is_a_randian_hero]

Neural Networks are used:

http://en.wikipedia.org/wiki/Neural_network

Real life applications

 

The tasks to which artificial neural networks are applied tend to fall within the following broad categories:

 

    * Function approximation, or regression analysis, including time series prediction and modelling.

    * Classification, including pattern and sequence recognition, novelty detection and sequential decision making.

    * Data processing, including filtering, clustering, blind signal separation and compression.

 

Application areas include system identification and control (vehicle control, process control), game-playing and decision making (backgammon, chess, racing), pattern recognition (radar systems, face identification, object recognition, etc.), sequence recognition (gesture, speech, handwritten text recognition), medical diagnosis, financial applications, data mining (or knowledge discovery in databases, "KDD"), visualization and e-mail spam filtering.

 

Look familiar?  I think I read somewhere Simons has hired almost all of IBM's speech recognition department away.  He claimed "speech recognition is very similar to finance.  In both fields you are trying to predict what happens next."

 

A friend explained neural networks to to me, as if you are at the top of a mountain, and drop a bucket of water, creating a formula that predicts the most likely path of the water down the mountain.

[ExpectedValue]

``Every little stalk of wheat was not doing so great, but most of them were, so you're working on statistics,'' Simons says.

 

By contrast, he says, the traditional focused investing practiced by Warren Buffett is akin to intensive farming, in which each individual plant really counts. ``It's two completely different ends of the spectrum,'' Simons says.

 

http://www.bloomberg.com/apps/news?pid=20601213&sid=aq33M3X795vQ

 

My guess is that at Renaissance they use a variety of strategies, investing in net-nets is just one of them. Like others have stated, they've probably seen a trend in net-nets being out performers in the market, just like Graham.

[snailslug]

Does this remind anyone else of what worked so well for LTCM?  I am sure they thought all their Nobel Laureates understood sigma too. But they didn't realize that their models relied on unrealistic assumptions.

 

But who ran LTCM?  John Meriwether:

John Meriwether earned an undergraduate degree from Northwestern University and an MBA degree from the University of Chicago Graduate School of Business.

 

Also, there is a big difference between PhDs in Economics (which the LTCM folks were) and PhDs in Math and Science.

[nodnub]

 

Also, there is a big difference between PhDs in Economics (which the LTCM folks were) and PhDs in Math and Science.

 

true