TAA Model for November, 2010
This is a new monthly feature at the MarketSci Blog.
Our Tactical Asset Allocation (TAA) model selects up to four assets from a diversified basket of asset classes on the final trading day of every month. Below is our position for next month. Click to read more about the TAA model.
The model performed well in October, our first month of trading, returning +1.9% versus +1.6% for the benchmark (as of yesterday, 10/28/2010).
Note that our first trade only included three positions. That was an early version of the model and after some additional testing I’ve settled on a four position model moving forward. Also note that I’ll be using a 50%/50% split between the S&P 500 (SPY) and 10-year US Treasury (IEF), rebalanced monthly, as the model’s benchmark.
I eat my own cooking, so I’ve devoted a fair amount of my own net worth to this new TAA model (read why). On the last trading day of each month I’ll be sharing our new allocation (see above) and tracking real-time performance (see below), excluding trading frictions and taxes.
. . . . .
[linearly-scaled, growth of $10,000]
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.
Filed under: Tactical Asset Allocation | 15 Comments
michael, so you’re going to use next to last day of month to measure performance and generate trades?
RE to Steve: no, everything is based on the last day of the month. I can’t show performance through the end of the month in these emails because they are issued before the day is over. michael
I’d be curious to see this strategy tested out using futures contracts instead of ETFs. CME products are diverse enough, in my opinion, to potentially work for TAA model(s).
It would be interesting to see how your backtest returns faired with the IEF/SPY and IEF/SPY/GLD/VNQ benchmarks (or equivalents).
Hello Ron: from 1970 a SPY/IEF benchmark would have returned about 0.5%/year less than SPY alone with about 40% less annualized volatility (but still more than the TAA model), and drawdowns about half SPY alone (but still twice as bad as the TAA model). Those are backtested returns so take with a normal dash of reality. michael
michael, can you provide an equity curve (or just monthly results) for 2010 so far as you did for 08 and 09? thx
steve
Hello Steve – these are of course backtested. Monthly 2010 results up to 09/2010, excluding transaction costs, slippage, and taxes:
-3.9%, 2.5%, 2.8%, 3.7%, -4.0%, 0.8%, -1.2%, 1.4%, 0.2%
michael
Michael,
in your backtesting how are you dealling with the issue that the three steps described take time and you can not buy the chosen assets on the same day as the signal is received, so with a lag?
RE to Eve: good question. There is no lag. I can calculate everything I need to calculate on the fly based on where I expect each asset class to close on the last day of the month.
Because the system is using monthly data (as opposed to daily, etc.) the calculations are not sensitive enough to that little bit of difference between where I expect the asset to close and where it actually closes to matter much.
michael
Michael,
how do you calculate return on bonds having in mind data available for backtesting provides yields, not prices?
RE to Eve: there is historical total return data available for Treasuries. Not sure of a free source, but it is available for $$$. Alternatively, you can approximate total return from the yield data (try a google search). Hope that helps. michael
nice blog michael.
FYI, we created a site to track total return strategies here — as a CFA charterholder and long-time money manager, I got fed-up with having to download and run dividend and distribution adjustments all the time. it sucked. so we took matters into our own hands. check it out:
http://www.etfreplay.com/backtest.aspx