Recorded Client Webinars went out today. Newer clients who have started paper trading are finding good opportunities. Strong start to the New Year.
Some select on-the-run inter market positive correlations for your consideration: Canadian Dollar vs. Nymex WTI Crude 1 Month 0.869 3 Months 0.690 6 Months 0.780 1 Year 0.944 2 Years 0.785 Comex Gold vs. Comex Silver 1 Month 0.518 3 Months 0.820 6 Months 0.888 1 Year 0.943 2 Years 0.905 FTSE 100 vs DJ EuroStoxx 30 1 Month 0.866 3 Months 0.735 6 Months 0.912 1 Year 0.800 2 Years 0.723 ES vs China A50 Index (SGX) 1 Month 0.786 3 Months 0.264 6 Months 0.429 1 Year 0.812 2 Years 0.452 NOTE: Daily Close-to-Close
Some more on-the-run inter market correlations based on daily settlements. I blend the results from two different statistical studies, and most importantly, I use the results as a simple SCREENING method to see if a combination warrants further investigation. FDAX vs CAC 40 1 Month 0.887 3 Months 0.968 6 Months 0.904 1 Year 0.900 2 Years 0.941 5 Years 0.970 Copper vs Canadian Dollar 1 Month 0.901 3 Months 0.533 6 Months 0.839 1 Year 0.942 2 Years 0.962 5 Years 0.947 10 Year Swiss Note vs 5 year German Bobl 1 Month 0.943 3 Months 0.697 6 Months -0.085 1 Year -0.075 2 Years 0.805 5 Years 0.859 ICE Gas/Oil vs 4 Year Russian Government Bond 1 Month 0.918 3 Months 0.836 6 Months 0.536 1 Year 0.188 2 Years 0.229 Indian Rupee vs Coriander (NCDEX- Food&Fiber) 1 Month 0.729 3 Months 0.728 6 Months 0.677 1 Year 0.114
Remote electronic trading as an independent can be a very isolating endeavor. I am truly grateful for all of the close friends I have made over the years with by client business and appreciate the fact that I can exchange Skype messages and emails from clients going several years back.
Got it. Of the nine pairs you listed, can you give us your interpretation of one of the pairs and how it fits into your trade analysis and selection? Without revealing any secret sauce, of course.
I've always found inter market correlations interesting and relevant - even when I started out scalping in the Bond Pit at the CBOT we would not only watch the cash Treasury markets but we would keep on eye on currencies and oil and gold and the stock markets as well. I'm not necessarily saying that there's anything super special or relevant about any one of these particular correlations that I have listed here - I'm just trying to stimulate some thought and reflection and discussion.
One observation I'll share is that your comparison of correlations over multiple lookback periods shows how consistent the correlation is or isn't over time. For example, the FDAX and CAC40 pair has a fairly stable two-year correlation of +0.94 which is in the neighborhood of its correlations for shorter lookbacks. By comparison, take the Swiss Note vs Bobl. It has a two-year correlation of +0.80 which sounds moderately correlated. But correlations for shorter lookbacks range between -0.08 and +0.94. It's all over the board. The takeaway is that correlations over longer timeframes cannot be interpreted in a useful way because this single statistic does not convey any information about the variability within the series. (The exception to this would be a very low/high value since the range is bounded at -1/+1.) An alternative approach is to make the length of the correlation proportional to the holding period of your trade, maybe 3X. So if your avg trade holding period is 15 days, chart a moving 45 day correlation over the prior two years. The chart will give you a visual grasp of how stable the correlation is for your needs by showing you both levels and variance. This charting approach probably doesn't sound helpful when you are trying to evaluate a large number of pairs because it's tedious to setup and review so many charts. Perhaps calculating the mean and variance of the moving correlation series would be the next best thing to give you a couple statistics to rank. I don't have a better solution to offer and welcome comments from others.
Not exactly true. If there were such strict 1:1 correlation even at shorter time periods (vs longer ones) then there'd probably be less opportunity to make any money on movement of the spread itself. Obviously if you have something with a 90% correlation over 2 years but with wildly varying correlation at the month level, it's probably not going to be the most enjoyable position to hold. But the point that not all correlations are equal is valid and I do agree with that. I would say though that targeting correlation to your holding period, or some multiple of it, could have some issues in itself. You may end up targeting coincidental correlation that has no guarantee of staying that way and you'd also have to align your entries and exits with the cycles of this correlation (which might not even show up again) which would be pretty difficult to predict. In essence your trade ends up being based off of "well this is what it's been doing lately" which as we all know has a tendency to become "it's not doing that anymore" - particularly in the kind of markets we have these days.