On the Drawing Board (Revisited)
Time to (re)focus.
This is the updated short list of things on the drawing board I’m most excited about. Some are carryovers from my previous list (humor me, this is more for me than anyone else), some new, and some done.
Disaster Overlay [New]
I’ve mentioned this before on the blog – the concept is to use options to add a simple “disaster overlay” to each of our portfolios to protect against big market moves over short periods of time (a’la Oct. 1987 or Sep. 2001).
After not giving the concept the time it deserves for way too long, I’ve done the only sensible thing and turned it all over to another developer I know and trust (and you probably do too) to design and implement. I think it’s going to be feasible to roll this out to both subscribers and managed accounts. Expect more this year.
Taking the Pulse of Short-Term Mean-Reversion [Done]
We issued the first monthly report of what I expect to be an ongoing series. The format isn’t set in stone and feedback is welcomed.
RH Modification [Done]
July made painfully clear that, group-think be damned, some of the concepts I talk about like the abnormal market filter needed to be applied to Timer Seed RH. I’ve worked with the developer and we have a solution that’ll be rolled out with Monday’s trade. I’ll be describing the mod in more detail in the monthly performance update.
Combining Strategies [From Previous List]
So given a group of successful individual strategies from many different developers, how do we intelligently dynamically allocate funds between them? This will of course come into the play with the performance-based product being launched. I’m working on a framework that goes beyond the traditional MPT approach to allocation, and should have a very high-level description out shortly.
Timing the Strategy [From Previous List]
This is the concept that’s been getting a lot of blogosphere chatter recently re: taking an effective strategy and then timing that strategy itself (level 2 timing or whatever we want to call it). I had considered dropping this one from my list because I don’t think there’s a one-size-fits-all answer, but I think it at least deserves an example. Whatever that example is will of course respect this critique.
Interesting Strategy Ideas
Two that have the wheels turning: (a) predicting growth vs value outperformance in order to trade the broader market, see What Makes the Market Go Zoom, Growth or Value?, and (b) using either just the overnight market or just the daytime market as an input into a trading strategy, see RE: Why Daytrading Stocks in the U.S. is an Increasingly Limited Proposition.
Back to work.
Happy Trading,
ms
To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed, or following us on Twitter.
Filed under: Random Stuff | 24 Comments
From “Combining Strategies:”
“This will of course come into the play with the performance-based product being launched.”
Are you planning to introduce a new investment program that combines some of your strategies?
RE to Corby: we’ve just briefly touched on this on the blog before. Yes, we’re involved in a project to combine the MarketSci programs, Timer Seeds, and other developers in some “smart way”. Can’t give too many details at the moment because it’s still WIP. More to follow. michael
Glad to hear of your decision regarding the development path for the Disaster Overlay. I’m assuming you are turning it over to someone who a) is maybe more familiar with all the details surrounding options (knows his gammas, vegas, and thetas), and b) has the time get this puppy done. If so, I’m starting to get excited about it. (Maybe I’ve been reading too much Nicholas Taleb!)
All the other stuff sounds pretty good, too. I’m looking forward to the next few months.
Michael, I am greatly looking forward to what you have to say about combining strategies – it’s a topic very close to my heart!
Glad to hear of these projects. It seems to me that besides risk mitigation you can use options in two additional ways.
1. Increased leverage. You trade triple leveraged funds for leverage. Options can provide even more leverage. Of course increased leverage has its own inherent risk. But that’s not unique to options. Other than the risk inherent in increased leverage the other disadvantage of options is that they are subject to much more friction than what you do with index funds. There is both a bid-asked spread loss and a commission loss. (By the way, how do ProFunds, etc. make money from accounts that use your strategies? Do they charge a management fee for investment in their funds?)
2. Options can provide two additional edges — although both are relatively small.
2.a One can sell instead of buying options thereby earning day-to-day time premium. Selling naked options of course is very risky. But selling instead of buying in the sorts of programs you offer carries more or less the same risk.
2.b One can buy or sell options so that the market move is on your side. That is, as the market moves, the leverage of a position changes. Depending on the strike price of an option its value changes by a given amount for every unit change in the price of the underlying. (That’s called Delta.) If you divide Delta by the option price you get the percentage gain or loss. Here are some examples using Sept Calls for SPY, the ETF for the S&P 500. (These figures were computed using the data at http://quote.morningstar.com/option/options.aspx?ticker=spy and taking the mid price as the option price.)
Strike Percentage gain or loss in option value per unit change in SPY
100 0.160465116
101 0.173553719
102 0.191275168
103 0.215189873
104 0.23655914
The value of a Sept 100 call for SPY would increase or decrease by about 16% as SPY increased or decreased by one point. The 101 calls would increase or decrease by about 17%. Etc.
The point is that the more SPY increases the smaller the percentage increase in the Calls for every unit increase in SPY. (That’s because as SPY increases the option that you own moves “up” in this table. In other words, a 101 call when SPY is at 102 is like a 100 call when SPY is at 103.)
Similarly the more SPY decreases the faster the option value decreases with every decrease in SPY.
So what this says is that if you are expecting SPY to decrease, sell Calls. That way as the market moves in your favor, your leverage will increase. As the market moves against you, your leverage will decrease. If you are expecting SPY to increase sell Puts.
RE to Russ (blue): hello sir – just a couple of general thoughts (not on the specific options strategies):
First, re: how ProFunds, etc. make money. They charge an expense ratio just like any other mutual fund (i.e. it’s spread over the year and accounted for in the NAV). The expense ratio runs a lot higher than a normal index fund because investors are able to move in and out of them so quickly.
Second, I’m already at the edge of my comfort level in terms of leverage given how our strategies trade. Some folks have asked about more (ex. 3x ETFs) but I don’t have the stomach for it.
Last, my intention is to stay away from intraday vehicles as long as we possibly can. You’re right, part of it is trading frictions. Part of it is also the consistency of results across all managed account sizes. And part of it is that the MFs exhibit such low tracking error to the underlying index (as compared to ETFs, options, etc). In other words, everything here: http://marketsci.wordpress.com/2009/03/09/faq-why-i-trade-leveraged-mutual-funds/
If we did break away from using MFs, it would be b/c of scalability – at some point Rydex/ProFunds wouldn’t be able to support our weight. Based on conversations we’ve had with the trading desks at both companies, I have a number where I think that becomes a factor (we’re not close yet). If it does become a factor, that’s a pretty good problem to have and we’d find a way to work around it.
P.S. as for the specific options strategies, you know I can’t intelligently comment (not my forte) but hope other readers have gotten value out of it.
Michael
Hi Michael,
First of all, let me say that I’m a great admirer of you and your work. So anything else I say should be taken in that context.
With respect to leverage, options give you choices; they don’t commit you to high overall leverage. Deep-in-the-money options are relatively low leverage. Far out of the money options are high leverage. Intermediate (close to the money) options are intermediate in leverage.
But even more importantly, one can choose how much to invest in an option position. Buying a highly leveraged position with a very small percentage of one’s account creates relatively low leverage with respect to the account as a whole. At the same time it leaves the remainder of the account available for other purposes.
If I had time to investigate it (unfortunately I don’t) I would test to see whether your program would do better investing a small amount in high leveraged positions or a larger amount in lower leveraged positions. I suspect it would come down to whether YK (or one of the other strategies) makes most of its money by being right on lots of small day or on being right on the important big days. Or perhaps neither predominates.
That is, how would the results compare if after the fact one kept only the consequences of using YK (or some other strategy) on days where the market moved more (less) than 0.05%? (If you already have that information, I’d love to know.) I think that one could implement something like such a strategy by using options — without knowing in advance whether the coming day was going to be a big or small move day. It would be worth testing to find out anyway. (Unfortunately, historical option prices are available only for a fee from the CBOE.)
RE to Blue Russ: no need to provide context for your comments – you’ll always be a friend at the blog and one of my favorite commenters.
Last I looked at the numbers, YK was running a tad over 50% win % with a W/L ratio of about 1.6:1, so definitely a matter of increasing position sizing on right days and reducing on wrong days than getting most days right. So given that, what would the options strategy you had in mind? michael
I did an example. I requires more assumptions than I had at first anticipated. But here it is anyway.
According to Morningstar these are the bid/mid/asked prices for the Sep SPY calls at several strike prices. (I hope this comes out readable.)
Strike Bid Mid Asked
108 0.53 0.56 0.58
109 0.37 0.39 0.41
110 0.25 0.27 0.28
111 0.16 0.19 0.21
112 0.1 0.12 0.13
That means (approximately) that a 110 call (0.27) will be worth approximately what a 109 call is worth now, or 0.39 (an increase of more than 0.12) if SPY goes up a point and what a 111 call is worth now or 0.19 (a decrease of only 0.08) if SPY goes down a point.
Of course bid-asked frictions will make those numbers virtually impossible to realize or it would always make sense to buy at the close every day and sell at the close the next day.
Even this makes sense only if no matter what happens on any one day an equal amount is invested the following day. Therefore the gain/loss is calculated in absolute rather than percentage terms. Admittedly this is something like doubling down but not as extreme. That’s a practical strategy since we are assuming that options will be a relatively small percentage of one’s account balance.
Also since in this case we are buying options there is also the loss due to time value.
All that aside, on very big days, the frictions will be significantly outweighed by the gain (or loss).
As I said, this may not be feasible, but it’s an idea.
Michael,
First, let me say that since I trade ES futures with YK, I only track it’s performance versus SPX/GSPC, and ignore the Russell component. With that said, since 11/3/08 I show YK as having the day’s direction correct 52.5% of the time, and a W/L point ratio of 1.56.
So that seems in general agreement with your numbers, which I assume are based on the combined SPX/R2K performance. In fact, my gut feeling is that the SPX-only strategy does slightly better than the combined strategy, but I have no data to back that up. (and without data, that thought isn’t worth much)
RE to Carlos: good observation. I don’t think I’ve mentioned it on the blog before, but yes, since inception the strategy has actually performed better trading just the S&P 500 leg rather than the S&P 500/R2000 mix (returns relative to volatility similar, but drawdowns significantly improved). I keep trading the mix b/c I think over the long-term this will balance out (or at least not be significantly worse). michael
One further thought. As equity prices increase implied volatility generally (although not always) declines, which means that options prices also decline. Implied volatility generally increases as equity prices decline. So if you are expecting a decrease it’s best to be long and if you are expecting a decrease it’s best to be short.
This is only partially consistent with the previous edge strategy. It says be short puts (as in the previous comment) if you are expecting the market to go up. But if you are expecting the market to go down, one might be long puts to take advantage of the possible volatility increase or short calls to take advantage of the leverage effect.
I just got a $50 gift cert to Amazon, anyone have any good book ideas on system development? Thanks!
Here’s some book suggestions:
“Design, Testing, and Optimization of Trading Systems” by Robert Pardo. This may seem ancient, since it came out in 1992, but a lot of the basic principles it contains just don’t change with time.
“Quantitative Trading Systems” by Howard B. Bandy. Good book, although Dr. Bandy leans a little too heavily on AmiBroker to suit me. But if you like AB, this should be your first choice.
“Way of the Turtle”, by Curtis M. Faith. Written by one of the original turtles, this book covers their methods in detail. But the real value is in the more general chapters, which have a wealth of info for aspiring trading system designers. And any book that has a chapter titled “Lies, Damn Lies, and Backtests” has to be OK by me!
And while not specifically about system development in general, I would highly recommend that all developers read David Aronson’s “Evidence-Based Technical Analyis”. Is data mining good or bad? I think you’ll be surprised at the answer.
As one of your customers as well as one of your blog readers I’m happy that you use the abnormal market filter. My question is how the filter is calculated. As you may know Condor Options publishes a weekly Volatility report. This week (http://www.condoroptions.com/wp-content/uploads/2009/08/09_08_23_volatility_tracker.pdf) Jared noted that “At 16%, the 21-day realized volatility of the S&P 500 is as low as it’s been since the financial crisis began.” Since the actual volatility of the S&P is so low, why is the abnormal market filter around 40%?
RE to Russ Blue: hello sir. I don’t provide the specific calcs, but the general idea can be found at: http://marketsci.wordpress.com/2008/10/15/a-new-approach-for-coping-with-abnormal-markets-shades-of-grey/
Condor’s vol tracker is looking at volatility in and of itself, but vol absent other factors isn’t what worries me. The AMF is looking at markets that have become “stretched” beyond what would be normally expected based on the idea that a very stretched market is prone to do unpredictable things. Volatility is a part of that calculation (i.e. the faster the market moves in a specific direction, the faster the AMF ratchets up) but the AMF is also looking at the direction of the move and whether there are the pullbacks and exhaustion rallies you would normally expect to see. Hope that helps. michael
As far as I can tell, the AMF posting you point to doesn’t mention anything other than volatility. That’s why I asked.
Here’s an interesting thought, though. Quantivity (http://quantivity.wordpress.com/2009/08/23/market-regime-dashboard/) is starting to investigate what he calls “market regimes.” Seems like a good idea and a nice extension of the basic notion of degree of abnormality. One might be able to adjust position size with more subtlety based on the extent to which the market is in one regime or another. Of course, the regime may also be used as part of the basic position calculation.
RE to Blue Russ: thanks for the quantivity link – i’m a little behind on my reading. RE: the AMF, volatility will determine the size of the bands, but market moving in a particular direction with strength is required to cross over the bands. michael
Michael,
While you’re catching up on your reading, David Varadi has a nice piece on the relationship between volatility and MR that would seem to be of interest from a market abnormality viewpoint.
RE to Jim: yep, I’m very interested in that series re: factors that moderate short-term MR – it’s a very interesting idea to me – rather than saying the market does this when condition X exists, saying ST MR does this when condition X exists. On the short list to put through the paces. michael
MichaeI,
I saw the same thing, and said “wow”! The only reason I didn’t post about it was because the reason I saw it to begin with was because it was on your “recently read” list, so I knew you were on top of it. But it sure does sound interesting — very much looking forward to part 3 of the series.
Did you go on strike? :)
=)
No, just going to extremes…
Trying to get more playtime in (lots of time hiking with the new puppy lately) and spending more time on for-profit stuff, leaving all the stuff in the middle (like this blog) a little short.
Back to the geekery soon, but thanks for checking in.
michael