Jump to content

Why does Renaissance continues to (hugely) outperform?


Jurgis

Recommended Posts

I don't know if crowd here has insights into this. Possibly wrong crowd to ask. :)

 

Context:

https://en.wikipedia.org/wiki/Renaissance_Technologies

 

Recent performance is unknown, but known outperformance to 2013 is huge.

 

Why is there no competition that have ground the quant analysis advantage to dust?

 

OK, so Renaissance has a big advantage that they have been collecting data (and algorithms) for ages and they have size advantage to purchase huge data centers to crunch the data and squeeze actionable strategies. They also possibly hire the best and the brightest.

 

However, all of these advantages are also disadvantages. Huge outperformance should have attracted copycats. Although nobody knows Renaissance algorithms, you'd think other shops would have tried to replicate or do something similar. Cost of computers is not very huge nowadays (you can pop analysis into AWS even). They are not doing HFT, so you don't need millisecond decisions (but even HFT is full of competitive shops). It's always possible to find young bright kids who want to do a killing instead of going to "stodgy" Renaissance. And smaller shops could outperform in less liquid and less efficient markets, where Renaissance can't hunt.

 

Is James Simons just so much better? Do they really have so much first-mover advantage? Any other ideas?

 

So why this hasn't been competed into oblivion?

 

Sorry, this is not an actionable topic and possibly way OT in value investing forum.

Just another technologists wondering rant.  8)

Link to comment
Share on other sites

If you haven't seen this interview yet, Simmons drops some nuggets of wisdom towards the end:

Among other things, I think he cites the data sets and testing/training platform they developed, along with algorithms for understanding market impact of their trades.

 

His best performing fund, Medallion, is for employees only and that has been closed to the public for a very long time (and this is the fund you usually hear the crazy performance numbers from).

 

I sat down in a cap intro event for one of his funds that was taking on money back in 2011 and returns were not that impressive and definitely lagged Medallion.

 

Link to comment
Share on other sites

If you haven't seen this interview yet, Simmons drops some nuggets of wisdom towards the end:

Among other things, I think he cites the data sets and testing/training platform they developed, along with algorithms for understanding market impact of their trades.

 

His best performing fund, Medallion, is for employees only and that has been closed to the public for a very long time (and this is the fund you usually hear the crazy performance numbers from).

 

I sat down in a cap intro event for one of his funds that was taking on money back in 2011 and returns were not that impressive and definitely lagged Medallion.

 

I've seen the interview. It does not answer any of my questions. It pretty much says what I said in the beginning of this thread: we were first and we crunch a lot of data and test a lot of algos. Ah, yes he also says that he hires the best (and possibly at some time in the past nobody else did - but now that's definitely not true). IMHO none of these are insurmountable for competition.

 

I did not know that other funds had much lower returns than Medallion. I wonder why.

 

I guess we won't know. :)

Link to comment
Share on other sites

I believe it is because their strategies are more similar to those of a high frequency/market making, not investing. 

 

People should think of marketing making more like a business.  It requires investment in systems, algorithms, relationships, etc. 

 

Do you think of a car dealer as an investor in cars, only worth liquidation value at any given time?

 

the returns are huge because they are measured with a denominator of liquidation value.  Market makers make up for this by charging huge fees.  I believe Renaissance's medallion fund has a 50% performance fee, on top of a large management fee.

Link to comment
Share on other sites

I believe it is because their strategies are more similar to those of a high frequency/market making, not investing. 

 

People should think of marketing making more like a business.  It requires investment in systems, algorithms, relationships, etc. 

 

Do you think of a car dealer as an investor in cars, only worth liquidation value at any given time?

 

the returns are huge because they are measured with a denominator of liquidation value.  Market makers make up for this by charging huge fees.  I believe Renaissance's medallion fund has a 50% performance fee, on top of a large management fee.

 

I can't say that I understand your comparison to car dealer.  :-\

 

Let me try to comment on things I do understand:

 

I don't think Medallion did HFT. However, I might be wrong.

Algorithmic trading is not the same as HFT.

Question still remains why others can't repeat what they did? IMHO there are no supergeniuses that create algorithms that others can't replicate. I somewhat know the machine learning community. There are bright people there but mostly not 70%+ for 20 years outperformance level brighter than others. ;)

 

50% performance fee and large management fee do not matter for 70%+ annual return. Sure, for outside investors, this makes Medallion crappy investment (mostly). Simons actually instituted the outrageous fees as a way to close Medallion implicitly before he closed it explicitly. But this even more underscores that someone else with comparable algorithms could steal most of Ren market and possibly remove the inefficiencies that lead to outperformance.

 

Edit: I think you are also saying that returns are not measured based on assets invested in the fund. I am not sure this is true. I am pretty sure their returns are calculated the same way other funds do it. I'd have to dig though to see if that's right.

 

Thanks for commenting though.

Link to comment
Share on other sites

If I am not wrong, what he is trying to say is that they're doing market making like trading in a bank.

The appropriate returns calculated should include i.e. costs of the machines, hiring people, etc.

It is like reporting revenue numbers and not "net earnings"? Traders in a bank also have very low down days like 1 out of 30 days anyway

Link to comment
Share on other sites

If I am not wrong, what he is trying to say is that they're doing market making like trading in a bank.

The appropriate returns calculated should include i.e. costs of the machines, hiring people, etc.

It is like reporting revenue numbers and not "net earnings"? Traders in a bank also have very low down days like 1 out of 30 days anyway

 

Hmm, yeah, something like "market making" might explain it. I am not sure if that's what they do and if it really explains it. I'm not an expert in "market making", so I'll let it go at that. :)

 

I don't think the "costs of the machines, hiring people" are really relevant. You don't include them into the returns of regular fund either. I also doubt that it would subtract more than 1% of total per year. So 69% vs. 70%? Even if it's 5% (which I really really doubt is the cost) does it matter?

 

Edit: Reported AUM is 65B. Reported employees are 290. 1% of 65B*.70 is 455M. Assuming $1M per employee, this still leaves 165M for machines per year. I doubt they spend that, but I might be wrong.

I did not account for taxes/RE/etc, but if we go with 2%, that should cover everything. ;)

Link to comment
Share on other sites

Put it this way.. if a asset management firm makes 30% profit margin a year.. is it a fund that makes 30%? or is it a business?

 

No. This http://www.bloomberg.com/news/articles/2015-06-16/how-an-exclusive-hedge-fund-turbocharged-retirement-plan claims it's the fund that had 71% annual return. Not the asset management firm. (I don't know where they got the numbers from though).

Link to comment
Share on other sites

some reasons:

 

a) Medallion cherry picks strategies. If a strategy does well, it gets in Medallion. If it doesn't do well, it doesn't. So Medallion results don't factor in losses from failed strategies. Those losses may be borne by other investors/funds. So Medallion is the "Admirals account"

 

b) as was well-documented in a Senate report, Rentech was essentially running a market making operation, using Deutsche Bank and Barclays as the fronts. Rentech would use 20x leverage when legal limit was 7.6x. Rentech had none of the costs of running a broker dealer, however. It was a form of regulatory arbitrage.

 

c) there are no examples I'm aware of in which Rentech employees started successful new funds. That suggests that the strategies were less important than the business model.

 

d) where are the Rentech billionaires outside of Simons? Is he the only one? Again, that suggests the strategies were secondary to the fraudulent business model.

 

e) if Simons and his geniuses were so brilliant, why did they need 20x leverage?

 

f) why does Simons always get such good press? Does noone wonder how a man got a $15B fortune? Did no reporter actually read the Senate report, which laid out all of his tax fraud? Read Matt Levine of Bloomberg for a sad example of a reporter rolling over and licking Simons ass

Link to comment
Share on other sites

some reasons:

 

a) Medallion cherry picks strategies. If a strategy does well, it gets in Medallion. If it doesn't do well, it doesn't. So Medallion results don't factor in losses from failed strategies. Those losses may be borne by other investors/funds. So Medallion is the "Admirals account"

 

b) as was well-documented in a Senate report, Rentech was essentially running a market making operation, using Deutsche Bank and Barclays as the fronts. Rentech would use 20x leverage when legal limit was 7.6x. Rentech had none of the costs of running a broker dealer, however. It was a form of regulatory arbitrage.

 

c) there are no examples I'm aware of in which Rentech employees started new funds. That suggests that the strategies were less important than the business model.

 

OK, I think b) answers the question.

 

I have quibbles regarding a) and c), but it doesn't matter if b) is right.

 

Thanks.

Link to comment
Share on other sites

The book "more money than god" goes into some detail into Rentech's strategies and why they prospered - if the reason is solely the 20x leverage, why hasn't it below up yet like LCTM? Since its entirely owned by Simons and his employees, would they rationally go too far into the risk spectrum? I think its competitive advantage is akin to Google's, it requires both large infrastructure and the human resource investments to pull it off. Each of its strategies have infinite capacity. Rentech generally looks for strategies that do not have intuitive explanation why they work. The book talked about how a few Russian employees defected to a competing fund. The defected employees had to work on both the infrastructures and the ever evolving algos while facing aggressive lawsuits from Rentech. In the end, they failed to compete away Rentech's return because Rentech was pulling ahead so quickly.

Link to comment
Share on other sites

Wonder if this is similar to Bill Gross.  What you think is happening isn't what's truly happening.

 

There was an article in the CFA magazine years ago where Gross spilled too many details on how he runs his fund.  While everyone thought the guy was a bond guru he was really something different.  If I remember correctly he was operating more like a bank.  He sold these custom designed short term options on bonds he held and then reinvested the proceeds in long dated stuff.  He was able to gain leverage from this and there was some aspect of the options that made it exceedingly unlikely that they'd ever be called.  If you're in the bond world and can safely leverage yourself some you'll massively outperform.  I don't think this is how he started the fund, but it seemed like this is the approach he took once he got much bigger and it was harder to pick individual bonds in size to outperform.  The guy was essentially running a backdoor banking operation.

 

The Rentech articles I've seen are always along the lines of "the company has super secret algos that beat the market" with no discussion beyond that.  Someone sent me an email years ago saying they knew Rentech was doing some massive quant investing in net-nets worldwide and that helped them.  It wouldn't surprise me if they were doing market making type operations as well.  My guess is it was some combination of those two with some leverage.  The LTCM guys blew up because they hit a black swan type situation.  If Rentech hasn't hit anything like that they can keep printing money.

 

I think the mystique of these guys is they hired super smart people who invested some magic formula that prints them money.  The strategy is infinitely scalable and no one can figure it out.  I doubt that's true, but I also doubt they want the world knowing how they are truly making money.  It's probably a lot more pedestrian and boring.

Link to comment
Share on other sites

Wonder if this is similar to Bill Gross. 

 

OT: Some time ago I looked at the composition of Gross's fund, saw all these weirdo options/derivatives/etc. threw up my arms and said "WTF". I got tons on flak on this from someone in the industry who was helping to choose a bond fund. All the "but this is a great fund, it's Bill Gross, you don't understand". I just walked away from the discussion. Did not make any friends either.  :-X

Link to comment
Share on other sites

Like PatientCheetah sayed "more money than god" has some insights. They also mentioned in the Keynotes that "Henry Laufer", with his studies of short-term patterns, contributed most to the success of Medallion. And "Robert Frey" sayed that the pattern recognition hasn´t anything to do with reversion-to-the-mean or trendfollowing, instead that there would be reactions to shocks which cause numerous market movements, which they seem to exploit in some way.

Edit: In 2007/2008 the returns where pretty high so there might be some truth to the statement.

Link to comment
Share on other sites

It is a great question Jurgis.

 

About 8 years ago there were 2 particular big funds that I did not understand.

 

1. SAC - we now know more about this one.

 

2. Medallion - I still don' t understand this. 

The returns should just be competed away with competition.  Maybe it is luck and leverage at this point. 

Link to comment
Share on other sites

It is a great question Jurgis.

 

About 8 years ago there were 2 particular big funds that I did not understand.

 

1. SAC - we now know more about this one.

 

2. Medallion - I still don' t understand this. 

The returns should just be competed away with competition.  Maybe it is luck and leverage at this point.

 

How about Och Ziff?

Link to comment
Share on other sites

What I read somewhere is they focus on front running large institutional orders.  If that is true it would be a largely market independent strategy assuming they don't maintain a large inventory that would be shock susceptible (they've been around long enough it's unlikely they're doing stupid things like carry trades/ short vega).  Since fundamental analysis is really just a more sophisticated version of the same thing I don't think the strategy would be self-defeating assuming the algos were continuously updated.  The lunch is not free but that doesn't mean there is no lunch.

Link to comment
Share on other sites

  • 1 month later...

I have a long background in quant funds.  My thoughts:

 

1) Simons properly solved the research part of running a quant fund.  Most quant funds do not setup research as a collaborative organization; therefore, the knowledge base of the whole organization is lower.  This is much more important than non-researchers recognize.

 

2) Simons could have open research by pairing it with a never-work-in-finance-again non-compete.

 

3) Medallion strategies can only handle a limited amount of capital.  I'm surprised it is working with the amount of capital currently invested in it.

 

4) If they are running 20x leverage, they are doing a good job.  He mentioned in one interview that the risk management was the mathematically most complex aspect of their strategy.

 

5) I suspect the strategy has a base in hidden Markov models.

 

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
×
×
  • Create New...