I have written a lot of posts on mean reversion in the past. Here is a post from September that is a good summary post including links. Summary:
In one study I examined asset class performance after a really bad month.
The take-aways from this study were:
- It does not pay to buy an asset class after a really bad month for the following 1 month.
- 12 Months later the return is not much different than average.
- 3 and 6 month returns, however, are stronger. You pick up on average about 3-4% abnormal returns buying after a terrible month.
A simple strategy would be:
After an asset class has a terrible month, wait a month then take a 2 month position.
It looks like a good “trigger” for equity like asset classes is around -8%, and for 10 yr bonds around -3%.
Note that this does not work for the commodity index, and one could speculate that is due to differing risk premiums and sources of return to that asset class.
How did this perform in 2008 and 2009?
Instances when the five asset classes had <-8% returns (-3% for bonds).
The below will show when the system would have triggered and the resulting return for sitting out a month, then long for a two month hold.
I’m not really sure what sort of benchmark to compare this system to, other than buy and hold of the asset class over the period. This system may prove to be a nice complement to the GTAA timing model. The stats show the date the system triggered, then the resulting "wait a month then invest for two months performance".
S&P500 triggered on, performance:
6/30/08: -7.6%
9/30/08: -6.2%
10/30/08: -7.6%
1/30/09: 19.17%
2/30/09: TBD (9.57% so far)
Average return excluding 2/30 is -0.55%. [Beating buy and hold by about 25% July/08 through April.]
10 Yr Bonds triggered on, performance:
1/31/09: -0.59%
[Beating buy and hold by about 1% Feb through April.]
MSCI EAFE triggered on, performance:
1/31/08: 4.5%
6/30/08: -17.87%
9/30/08: 0.34%
10/30/08: -4.37%
1/30/09: 20.18%
2/30/09: TBD (12.96% so far)
Average return excluding 2/30 is 0.56%. [Beating buy and hold by about 38% Feb/08 through April.]
NAREIT triggered on, performance:
6/30/08: 1.67%
10/30/08: -3.24%
11/30/08: -32.77%
1/30/09: 33.61%
2/30/09: TBD (27.97% so far)
Average return excluding 2/30 is 3.89%. [Beating buy and hold by about 28% July/08 through April.]
I just read the GTAA timing model and this article, as well as many of your very credible links.
As everything is trending up, it may take a while for a set up.
Is there any credibility in part of the explanation as to why this works is because money flows take so long for institutions?
KBW the Bank Index seems to be one that fit the criteria well. January was a bad month. February was worse. Entering at the end of either month should have outperformed. A year from that initial entry would be a good time to be long gone though.
Sixty days is a nice time frame to watch and plot – I will follow this idea.
I used to see an annuity product decades ago via The Travelers. It did not do very well. LOL.
Wonder wether same logic would apply for months with best monthly return? Any results?
Charts and Coffee Weekly Market Forecast – http://chartsandcoffee.blogspot.com/2009/05/sun...
Good comparison galore. At least finding how to correct bad results is now comparatively easy.
[...] Below I update the post from May on mean reversion. [...]
A one year back test on this mean reversion idea does not seem to be sufficient and since most of the market tanked and recovered together, the fact that it applied across asset classes may mean it only applies to 2008/2009. I'd be very cautious on this one.
The backtest goes back to 72, this is just the update…
A one year back test on this mean reversion idea does not seem to be sufficient and since most of the market tanked and recovered together, the fact that it applied across asset classes may mean it only applies to 2008/2009. I'd be very cautious on this one.
The backtest goes back to 72, this is just the update…
[...] Mean Reversion After Really Bad Months [...]