Guest Post: A “Braided Basket” Trade On The Apocalypse
Submitted by JM
A “Braided Basket” Trade on the Apocalypse
So we all know that gold prices and UST 10Y yields are as high, and low, respectively, as they have ever been. This is nothing more or less than human adrenaline overriding reason and logic, driving return expectations to the distribution of max entropy. It’ll pass.
Sometimes it makes sense to fight the crazy impulses of greed and fear. But often this gets you creamed in the center. Sometimes it doesn’t. For those times, the prices of straightforward hedges like 10Y Ts and gold make them very unhedge-worthy. There is no sense in jumping on trades that already have the risk premium baked in.
The alternative is to ride the apocalypse with an eye on the relative mispricing of extremal points. I’m creating what I call a braided basket to do this. I’ll take two pair trades and go short the rapier points of the apocalypse and long something correlated, but underperforming it. In this way I’ll catch some hedge on tail risks on my core book due to the darkening outlook. At the same time, I’ll catch some cover when people come back to their senses. Why braided? Check out the charts, and see how the pairs interweave.
Note that the point is close correlations. Actually, the point is coupling. Coupling is a technique where two dependent copies (Xn, Yn) of a random process where Xn is stationary and Yn starts from a fixed arbitrary state—and they have the property that they coincide with higher and higher probability as time evolves. So for example, the underlying random processes are negative real interest rates and credit risk spikes.
The endgame is minimizing the total variation distance between two probability measures p and q over all subsets A of G. This is what the coupling method does: it determines the time T is the smallest n such that Xn and Yn coincide. The total variation distance between the law p(n) of Yn and the stationary law q of Xn can be bounded by Sup|p(n) – q| ? Prob(T > n). The construction of a good coupling that estimates Prob(T > n) is the decisive issue for this technique. Construction of a good coupling means finding pairs that exploit relative mispricing (measured by the total variation distance) of two assets.
Remember that this is a hedge position on the irrationality that fear brings to the table, and the short is only there to cover pullbacks. Core holdings should always go to buying what is cheap. Here we can only say we are buying a relatively cheap alternative and shorting a way-too- expensive alternative. It would of course be far better to buy cheap “just in case” hedges rather that these “just in time” hedges that are necessarily more complicated, but it is way too late for that now. You need to balance it and not be equal weight short and equal weight long in either pair. Not sure that cointegration helps here to determine the weights, because illiquidity implies non-stationarity, at least in a local sense of the term.
Negative Real Interest Rates Leg: Gold is fighting a lot of history here.
Long Platinum, Short Gold
Credit Risk Explosion Leg: For this leg, it makes sense to go with more term risk, so I’m looking at 20Y nominal Ts and 20Y General Obligation Muni bonds. This is really a play on my vision that credit risk and interest rate risk will become an expression of the same phenomenon going into the future.
Short 20Y, long 20Y GO Munis
Of course this assumes that basic recurrence, which is a very general notion of mean reversion, holds. It may not, or mean reversion could be to zero, in the “naked I came into this world, naked I go out” kind of way.