An Educational Stat Arb trade

Here is the full GDX/GLD trade even though I don’t find it attractive anymore. We had bad luck with the 5% move in GDX that happened before we could start this trade (it would have been wonderful if we were on it before that). The idea behind the trade is the strong correlation between GLD and GDX which is around 93% long term. However the correlation has been very inconsistent of lately, around 30% in the last 90 days then going up to 87% for the last 60% days and it is now at 70% for the past month.

The idea of the trade is that the correlation remains high or improves in the next 50 days. Also another important aspect of the trade can be seen on this table:

  Symbol Expiration   VIX    HV    VRP      IVF      HVF
1     GDX 2017-04-28 41.38 21.05  96.57 2.996379 2.491124
2     GDX 2017-05-05 34.97 23.77  47.14 2.740596 2.673791
3     GDX 2017-05-12 33.63 26.51  26.86 2.701205 2.586341
4     GDX 2017-05-19 33.95 23.51  44.41 2.787356 2.668558
5     GDX 2017-05-26 33.65 24.68  36.35 2.762726 2.876457
6     GDX 2017-06-02 32.89 23.08  42.51 2.718182 2.835381
7     GDX 2017-06-16 36.17 31.30  15.56 2.888978 3.287815
8     GDX 2017-07-21 35.93 32.02  12.24 2.811424 3.331946
9     GDX 2017-09-15 36.13 35.91   0.61 2.664454 3.436364
10    GDX 2017-12-15 36.25 41.41 -12.48 2.519110 3.473993
11    GDX 2018-01-19 37.00 40.89  -9.50 2.532512 3.500856
12    GDX 2019-01-18 34.31 48.14 -28.72 2.031380 3.272604

The HVF column represents the realized volatility of GDX divided by the realized volatility of GLD. You can see that GDX tends to be more volatile than GDX usually above 3.0X however the IVF column shows a that the implied vol of GDX is only about 2.7X the one of GLD. So there is a small edge there. The sad part is that the edge was good enough to have captured most of the 5% recent move in GDX, we missed the boat there.

The trade consists of buying a number of straddles in GDX and selling a number of straddles in GLD. Now because I don’t like unlimited risk, instead of selling straddles we’ll get a butterfly in GLD (which is basically an straddle hedged by two OTM wings).

The ratio of the trade is computed by this:

1 GLD fly = ( GLD_price/GDX_price / HVF) GDX_straddles

1 GLD fly = (120.15/22.11/3.3) GDX_straddles = 1.65 GDX straddles.

So for each GLD butterfly we open we need to open 1.65 GDX straddles. In round numbers we could use 5 GLD flys and 8 GDX straddles.

The GLD fly is the following

BUY 5 GLD Jun16 110 CALL / SELL 10 GLD Jun16 120 CALL / BUY 5 GLD Jun16 130 CALL

Mid for the fly is : 6.11

The GDX straddle is the following:

BUY 8 GDX Jun16 22 CALL/ BUY 8 GDX Jun16 22 PUT

Mid for the straddle is: 2.21

So the trade costs 5*6.11+8*2.21 = 48.23 to put on in total (x100)

It is not a cheap trade. Now if GDX and GLD remain within the limits of their implied volatilities then the trade makes $1.59 by default (until expiration) which is like 3% of the total investment. This is the worst case profit scenario, now because we know that GDX will move at least 3X the equivalent amount than GLD (we’ll that is the thesis) then the profit will start to grow dramatically. If we had been on this trade before the 5% move in GDX, we could have made about $15 in profits just that day alone.

The main risk of this trade is that GLD actually moves more (in percentage terms) than GDX, if that is the case we can accumulate losses. The worst loss possible, is if GLD moves 10 points during that period and GDX doesn’t move at all (it remains the same), in that case we lose $30.5, an scenario like that while not impossible is highly unlikely but I leave it here to get a sense of the max theoretical loss of this trade.

Running Montecarlo simulations on this we can find the current edge of the trade and it is very small right now. We have about 32% probability of being wrong with the main thesis which although not that bad in terms of odds, doesn’t make the trade that attractive in terms of risk reward for me. Because of this I’m putting this trade as an educational post :)

We can paper trade this just out of curiosity and track it until expiration. I almost forgot to mention that this trade needs to run all the way until the last day, unless of course we had a day like that 5% move day in GDX :)

Leo Valencia hosts the Gamma Optimizer options service at