Licking Our Wounds
Monday, June 13, 2011 at 7:39AMWe decided to unwind our short $GLD-$GDX spread this week. We first went short on the spread in April and have averaged down a couple of times since as the spread continued to climb. The spread seemed to finally turn around at the end of May before turning sharply higher again the past week. What do you do in a situation like this? Add more? Unwind all or part? Buy more $GDX?
We’re not sure there is one right answer, but we can tell you why we chose to unwind. We felt that something has obviously changed in the market sentiment regarding the mining stocks since we put on the spread. Gold has held its own the past two months as other commodities sold off, but the gold mining stocks haven’t performed well. This doesn’t feel like a short-term liquidity gap. Also, overall stock market sentiment seems bearish, something that won’t help the mining stocks. Lastly, although $GLD-$GDX is trading at the highest level since the financial crisis, there is no reason why it won’t continue to grow higher to crisis levels. And that would mean additional losses. So we unwind now, lick our wounds, and perhaps return later in a stronger position.
Written by Norm Winer. Follow me on Twitter and StockTwits.
Reader Comments (8)
Why are you trading like a human still? I also watch this spread and think it will come back at some point. However, I think when you see a situation like this you should be deciding which algorithm you want to go with...then put it on and come back in a year and see how it did. (well maybe watch more than that)...but you are getting out for an easily quantifiable reason... because the spread is going against you for x amount of time...if we had an algorithm watching this we might have gotten out much more cheaply and still had tolerance to re-enter when the situation looks less risky.
rule1: When the X day MA of the spread is moving against the idea & S&P 500 is hitting new Z(3?) month lows then exit trade.
rule2: when the spread makes a new high and then doesn't exceed the old high for Y days then you enter trade again
How about we find three or more historical examples of similar spread trades and see what set of X, Y and Z worked best for minimizing losses and transaction costs while waiting for the spread to stop moving against us while maximizing gains by getting in on the spread re-convergence when it eventually comes back in our direction.
look at the 2nd widest spread divergence over the last 5 years for $GLD-$GDX
look at XOI-CL biggest spread over the last 5 years
look at (Corn-NG) - AGU biggest spread over last 5 years
back test the three biggest spread scenarios for the three trades and see what X,Y and Z gave you the best return. When this is done it is subjective yet very important to determine when the test algorithms should be put in place....how about do it when the old spread record had been exceeded by some percentage(50%)...so this could vary between 1 month and two years of when the new spread peak was hit. If we find optimal X, Y, Z for these scenarios and then the $GLD-$GDX spread takes 3 more years to peak out then we will likely lose anyway...but it it peaks in 6 more months then our chosen strategy will probably make a lot of money and we don't have to watch it every day for the right opportunity....we can focus on looking for other opportunities.
btw:
$GLD-$GDX is getting hurt by the fear of ending QE2.
When google trends indicates "deflation" is trending up strongly and possibly "Scott Sumner"(one of the leading intellectual proponents of QE3) then we might look into using that as a component of a trigger for re-entering the $GLD-$GDX convergence trade....because then there will be enough political cover for the Fed to engage in more unconventional bubble blowing or whatever it is called.
Very cool thinking Gabe. Digesting all of this.
Thanks Gabe. Actually most of our spread trading is statistical based and guided by hard and fast rules. When trading GLD-GDX we look at stats, but we also make a lot of "qualitative" decisions based on the unique factors surrounding this spread, some of which you mentioned (QE, gold as a currency, Euro problems, etc.) Still thinking about your comments.
Boy look at how GDX is outperforming GLD the last two days. I'm sure you already saw that. Now your trend followers and your reversion-to-the-mean algorithms should be on board.
I didn't notice till the market closed yesterday so I wasn't on board either, but it seems if you had a long term "looking for a reversion-to-mean" trigger switched "on" and had your computer waiting for the short term trend following conditions to be met...then you would have made a lot of money between yesterday and today on the GLD GDX spread....without having to sit through all that pain the last couple months. maybe transaction costs would have gotten you though...stil seems worth backesting.
well deflation trade is on again today...so this isn't working so well.
Just very volatile. The spread is not cointegrated, so if we trade we have to make a decision call.