Quantopian – an over complex approach?

I don’t get the (possibly curve fit) obsession at Quantopian for tiny low volatility returns on vastly complex long/short algorithms for the sake of then gearing up. Sector neutral, dollar neutral, factor neutral, everything neutral

I spent some time working through the Quantopian algorithms and code but at the end decided the complexity was (for my taste and personal investment) unnecessary and possibly dangerous. Possibly dangerous because at the end of it all you really can’t tell where the returns are coming from.  Or at least I couldn’t.

But of course that is merely my own personal and unsophisticated view.  The analysts at Quantopian, with infinitely greater mathematical and statistical skills, may well fully understand where their returns will come from and be comfortable with their approach.

By contrast, I have been back testing simple “All Weather” strategies using mildly leveraged bonds (5 year treasuries and below) + stocks. Something like 125% bonds and 40 to 50% stocks.

Using futures you are leveraging positive carry, courtesy of the yield curve. Or on occasions of course suffering from negative carry. But positive on balance over the long term, since longer maturities generally pay higher interest rates to compensate investors for the greater interest rate risk assumed.

Using futures, you can also reinvest most of your capital in T-Bills since very little is needed for margin on a low leveraged approach. I looked right back to the 1960s on bonds (and further) so I could see in practice the effect of a rising rate environment on bond/bond futures prices. I noticed the scary Volker period in the early 1980’s when rates rose rapidly and sharply within a short period of time.

Rising rates eventually mean bonds benefit, provided the rise is not too swift and severe.  Leaving aside inflationary implications.

Perhaps all the manic activity at Quantopian on long / short stocks is sufficiently different to attract their main backer Steve Cohen and other punters?

Perhaps Cohen and others would feel cheated by a simple risk parity approach?
For me the simpler the better. I’m not comfortable with a vastly complex long short approach. Way too many variables for me and way too difficult to see what is driving the P& L.

Bonds go up over time. Unless they go bust. And high grade sovereign bonds at the shorter end of the yield curve are usually low volatility.

Stocks go up over time – provided you use some sort of filter like a stock index.

Provided you are modest with leverage, you are getting similar return on a risk parity approach to stocks but for 1/3 of the DD and Vol.

You are investing in two asset classes which should go up over time. You can see very clearly where your growth and income streams are coming from. You are not investing in a potential powder keg of code and complex relationships between stocks and classes of stocks and factors/ investment strategies which may not last.

You could of course rightly argue that stocks may not always go up. You could also argue that the fortunate and usual relationship between stocks and bonds may not last – both may crash at the same time.

However perhaps these bets are simpler than the jungle of code and tangled web of relationships provided by the Q Long / Short Strategy with complex optimization subject to the lord only knows how many constraints.

I could well be accused of being naive, simple, ill educated and plain wrong. I probably am. For what it is worth however I prefer something I can fully understand and which is simple. Risk parity may not always produce the returns in the future it has in the past. And the risk reward ratio may not always be so attractive.

But if it does go wrong at least I will be easy to understand why.

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 )

Facebook photo

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

Connecting to %s