As European politicans cut short their grande vancances to attend to the latest crisis, it's worth having another look how the Halloween effect is once again playing out.
The theory is that markets perform best between November and May and this has certainly been the case going back to 1990 on the S&P 500.
The chart below shows three investment strategies:
Nov-May does give you more months in the market, but they do seem to be the right months. Food for thought...
After reading this post from dshort musing that stock markets are looking a bit rich, I thought I'd update on the performance of the PE10 + momentum.
You can view my last update here and my original post here.
The basic idea with this model is to use Robert Schiller's Cyclically Adjusted rolling 10 period PE ratio as the value measure. I've used a reading above 20 as the point of being overvalued. Below you can see that historically the S&P 500 has actually spent more time in with a PE 10 reading below 20.
The second part of the model looks beyond just looking at value. The idea being that markets can be expensive and carry on going higher for years. The best example is the UK Housing market which was expensive for years but kept on going up and up. Even as early as 2003 people were saying they had sold their house ahead of the coming crash, but that was too soon. Prices have dropped, but are still above 2003 levels, so selling based purely on something being 'expensive' is not enough. They might get back down there mind you, eventually in an inflation adjusted basis.
So you need a momentum component. For this I used the 10 month moving average of the market as suggested by Mebane Faber.
The rule is to buy if the market is above the 10 month MA and sell if the market is below it. The kicker is that this rule only kicks in at points when the market is over valued (PE10 over 20), otherwise you just buy and hold.
So after all that waffle, here's the updated performance for the longer and shorter term:
Right now the momentum is almost relentlessly pushing higher, while the valuation level makes it look expensive. The Value + momentum model would keep you in the market until there is a pull back.
[I wrote a draft of this piece the other night and with well timed irony, the S&P 500 has finally pulled back]
One thing that's puzzled me is what to do with trend following markets. There has been a lot of research on what to do with a mean reverting market using short term indicators such as the DVB or RSI (2) (2 period RSI) - I.e. buy over sold and sell over bought readings etc.
But what about markets which are not as prone to mean reversion? Should you just reverse the rules and buy on strong readings etc? Another options its to employ a relative strength rotation model which holds the best trending markets as with the Livermore Active Issues.
Many people have also be commenting on the recent rally on the S&P 500 and other markets. I can't remember the exact stats, but Bespoke pointed out that it had been some time since there was a 1% correction.
Anyone trading the basic short term mean reversion rules for indicators such as the RS2 or DVB will have had their patience tested in recent weeks as the market continued to push higher. The screen grab below shows the S&P 500 with the DVB indicator below it. The basic trade idea is to buy when the DVB drops below 50 and sell when it moves above 50. As you can see, the DVB has been above 50 for some time.
I wondered what would happen if you used the average of the DVB as a trading input, specifically to adapt any methods to situations like this.
I tested the 20 period average of the DVB on the S&P 500.
Interestingly, your average return per trade is increased by trading in line with the 20 period average of the DVB. For example, if the DVB is below 0.5 and the average DVB is above 0.5, no trade.
It's not a spectacular addition, but maybe worth some further investigation. Looking into the standard deviation of the DVB or RS2 also seems to be useful from my quick study. Essentially the returns from trading a short mean reversion strategy are lower when the standard deviation of the short term indicator is below average. This kinda makes sense as if you look at the chart above, you can see the DVB barely moving for weeks, which is what you get during a steady uptrend (bad for mean reversion).
Trend following markets
This got me thinking about trend following markets and the best approach to take. If a market is more prone to trend following then trading short term mean reversion will be ineffective and trading the opposite rules may actually work out better.
Foreign exchange (Forex) is certainly more prone to trend following and in my experience short term indicators such as RS2 don't work when applied in the usual way. They actually make a small profit when traded with the opposite rules (Buy on high readings etc).
I wondered what would happen if you applied the 20 period average of the DVB to EUR/ USD (Euro to US dollar spot rate). The red line is the 20 period moving average of the DVB.
The rules were as follow:
Buy the EUR/ USD when the 20 period average of the DVB is above 0.5
Sell(short) the EUR/ USD when the 20 period average of the DVB is below 0.5.
No commission spread etc. EUR/ USD spot since 1998.
The average margin per trade is small, with the average trade making 0.03% profit with a 51% strike rate. Too small in its own right to be a devastating strategy. However, it does lay out a good platform on what is usually a very tricky market to trade and I've found a few interesting angles already by delving deeper. I'll have to keep those under my hat for the moment though.
Some thoughts:
Appreciate any thoughts...
[Post Script] David Varadi's De trended indicator concept looks to be an elegant way to deal with trending markets. I just haven't had chance to analyse it yet.
