I spent quite a bit of time over there at MRCI this morning. Looks like a first class operation. I didn't see any seasonal weather charts over there, but I'm not a subscriber yet. I suppose a trader could find those about anywhere on the web. But the whole thing gets deep pretty fast. Like in beans. Avg rainfall in Argentina vs Rainfall in Iowa vs CL vs USD vs S&P vs beans...etc If you like ES, I would think ES vs QM would be fun. But like you said, The calendar tuition in corn is about $320.
There are a few new spread ETF's by FactorShares. They are pretty thin volume so far, but hopefully that will improve with time. Here's the website: http://www.factorshares.com/ *Edit: It looks like they have a deal with Interactive Brokers to make them commission free at IB
So using your chart to time or size your trades, you are effectively using the equity curve as a mean reversion indicator. Interesting approach which I've come across before though not often. I wonder if the fact that these trades are seasonally based makes mean reversion more likely for the equity curve than any other fundamental or technical basis. Hmmm, I'll have to give that some more thought.
ok, but while you're thinking about it, just remember that this debate about whether spreads trend or revert back to the mean based on seasonals or anything else has been going on since before I was born. I will agree with you that it's one of the most, if not the only interesting thing to think about.
Actually, there is a nuance in equity curve trading that separates it from the basic concepts of trending/ranging markets. The equity curve is simply the cumulative performance of a trading system's P&L. You can use drawdowns in the equity curve to place trades or increase trade size if you assume that the system will recover to avg performance. Or, you can use runups in the curve to stand aside or reduce position size if you assume that wins are followed by losses. These assumptions are based on mean reversion of the equity curve (not the underlying markets/spreads). I was thinking about whether seasonal spreads are more inclined to have mean-reverting equity curves than other trading approaches. I have concluded that any system with approx 50/50 wins/losses that repeat with regular frequency would exhibit this tendency. Seasonal's may fit this profile over some time periods, but I can't think of any special attribute that would make the seasonal equity curve more or less mean-reverting than other approaches.
Another way to think about it is that if you assume my 2.5 years of data I used to generate the performance graph is representative of seasonal spreads in general, then the graph is proof that they are cyclical. There are of course assumptions that I made and MRCI makes implicit in the process. But in general it is nothing but a graph of one slice of human behavior that I think we all have experienced. We feel euphoric when our accounts run up fast and daydream of opening a hedge fund. Your account starts to move into a drawdown and you are too slow to realize it or you think it just a short pause, then you feel like a loser because you realize you're not good enough to run a hedge fund anymore! So when entire groups of people are doing this in sync we get cycles in the market because not everybody is euphoric and they take advantage. Maybe MRCI is trying to include psychology in their recommendations, but generally they appear to be methodically generated. If that's the case it is up to me to control my emotions to improve my performance and also to try to measure/determine what cycles are in the market and try to take advantage of them. I can't speak to whether or not seasonals spreads are more apt to revert to mean than other markets, but they appear to be there. As to mean reversion for the equity curve more/less likely than fundamentals or technicals, I would say the equity curve is simply another technical measurement that displays the cumulative effect of the trading system that generates it and the algorithms/methods within the system. The system could be fundamental or technical or both. One might also argue that every market and/or trading system is mean reverting otherwise we would find markets that don't revert back to mean and be profitable to everyone. Perhaps the exception to this would be shorting stocks of poorly run companies....
From my understanding, MRCI have a whole bunch of spreads, but only post the 14-15 each month that have the highest track record over the past 15 or so years. So could it be that the spreads they select are naturally at the top of the cycle - or mean reverting, which is why their results seem to be 50/50? Maybe we should ask for those that have the worst 15 year track record, and filter through fundamentals?
1. Mean Reversion has typically been the traditional methodology to trade spreads - the effectiveness for the strategy is still OK for Flys and Condors, but for futures pairs generally speaking they have been trending much more frequently than 'mean reverting' for the past few years. Alot of the pain felt in the Chicago prop scene is reflective of that proposition - look at the yield curve, grains, or energy spreads using continuous data in the Weekly and Daily timeframes. Certain spreads like WTI vs. Brent and TY vs. Bund diverge (trend) like crazy. In the early 2000's the Bund vs. TY had well above a 90 % positive correlation - now, it's in the mid 60 percentile range. 2. The one seasonal study that I really like is the Bloomberg forward seasonal curve. Very nice. Other more traditional seasonal approaches I believe to be marginally useful for the retail speculator as compared to a good price action model designed for spreads. My belief is that you'll take more trades and have a significantly larger pool of possible spread trades in more market sectors with the price action model as compared to the seasonal approach. My typical client in the live markets have been seeing Win to Loss Ratios in the mid-60 to mid-70 percentile range. YTD annualized Sharpe Ratio about 8.00, monthly max drawdowns I am seeing about 20 % of the monthly average returns.
That's an interesting observation. However, I think that the this thread is mixing two different but related ideas. There is a difference between SPREADS mean-reverting and EQUITY CURVES mean-reverting. So, Bone, if you took your or a typical client's equity curve and de-trended it, would it have a regular cycle? Would you take more trades after five straight losses? Given how diversified you are, probably not - drawdowns are probably random. What about within a complex (energy, grains, etc.)? If 5 grain calendar spreads stop you out, would you get more aggressive on your sixth? If you hit home runs on 4 out of 5 crack spread trades, would you dial back your sixth? THAT is what this thread is trying to dig into. From what I have seen, there may be some benefit to getting more aggressive after a drawdown, but only with proven, robust systems. (The most well-known case is waiting to invest in a CTA with a great track record only AFTER s/he has a negative year.) There was a pretty good discussion of this topic on the tradingblox forums a while back.