No buys were triggered for next week.
I'm starting a bit of work on sector analysis, and relative strength between sectors. The purpose of this work is to see which sectors are outperforming the market.
This sort of analysis can be used instead of (or in addition to) an index on/off switch, because such a switch used alone may keep you completely out of the market for a prolonged time in a bearmarket. This may not necessarily be a bad thing but it may deprive us of trading opportunities.
Take a look at property trusts during 2000-2002. Nothing staggering but a fairly significant outperformance of other sectors and of the general market.
The problem with this sort of stuff is that it's hard to backtest anything to do with index switches or sectors because you cannot really purchase historical index/sector constituents from anybody, as far as I am aware.
And it would take a hell of a long time to construct them yourself.
Or maybe we can purchase them from Standard & Poors?
Any suggestions?
8 comments:
"This sort of analysis can be used instead of (or in addition to) an index on/off switch, because such a switch used alone may keep you completely out of the market for a prolonged time in a bearmarket. This may not necessarily be a bad thing but it may deprive us of trading opportunities."
Hi Niz,
This has been my observation. I will most likely not be using one in my trading.
ASX.G
Hi ASX.G,
Thanks for your comments.
Have you done any work on sector analysis?
If so, what were your conclusions?
Nizar.
I haven't. And I'm not sure I need to. I have found that rather than attempt to channel funds into performing sectors I would rather design a system that can be exposed to the entire market (all those stocks we can test on via Premium Data, under Equities). I've created more stringent filtering than say TechTrader uses, for example, to frame the precise kind of trade that I want. This way, I don't care which sector a stock is from, or whether a sector/index is in gear, if a stock matches the filters, it's a candidate.
ASX.G
ASX.G,
I used to think the same way for a very long time. Up until this weekend actually, when I started giving some thought to an index-based switch.
But now I'm thinking, from the 20,000 different portfolios that I usually backtest with (monte carlo analysis), it is possible that the very best of these are those that picked stocks from the strongest sectors.
Anyhow, I think its an idea worth testing. At the minute, I just have no idea how to.
Nizar.
Hi Nizar,
Why do you say that it is difficult to obtain the index related data. Data for XJO, XSO or whatever ASX based index can be downloaded exactly as any individual stock data, in my case, with Justdata.
Secondly, you should be able to backtest your system with and without the index switches. I have always found that sure the number of trades are reduced, but usually there is not a significant difference between the returns (within 10% of each other) when tested over long periods of time and the drawdowns of the system with the index switch is always lower than the system without the switch. In turbulent times like now, it is imperative to have a system switch in place, especially if your system is a long only system.
Bingk6
Bingk6,
Thanks for your comments.
Yes you are correct, it is easy to backtest with and without an index based switch.
But with sectors, I think there is a problem, and here's why.
Say, during a backtest, my system was looking at relative strength between sectors during 2001, and found that the property sector (XPJ) was the emerging sector. So buying is restricted only to that sector. Which stocks were in the XPJ in 2001? Remember these sectors/indices are updated quarterly.
ASX.G,
Just thinking about it some more. This is the way I see things.
The benefits of having a system switch is less trades and less drawdown.
The drawback is that you may be out of the market for a long time, thus decreased trading (and profit) opportunity.
There maybe sectors that outperform during this time (like property trusts in 2000-2003), but you will not trade them because you are out of the market.
The problem with NOT having a system switch is that if your system is being traded in conditions OUTSIDE those that were endured during testing, you may reach max.DD. So you stop trading. But when do you get back into the market?
I see sectors as a way of overcoming both these problems. But we have another problem - how to backtest using sectors.
Nizar.
Nizar,
Couple of quick comments
1) Sector selection - don't know much about them but for mine, that is not the approach, as you have indicated, there is no easy way to backtest it. A better question to ask yourself is whether the system that you are trading has been successful in getting you onto stocks that are running hot and therefore belonging to a hot "sector". If the answer is no, then I suggest that you find out WHY!!! A well designed system, IMO, should be generating signals for stocks that are in an uptrend and a hot sector simply contains lots of stocks that are in uptrends, and if you have not managed to secure any of them, you really should be re-examining your entries again. What I am saying is that a good entry will be looking for stocks that are in the hot sectors anyway, without you specifically looking for it.
Secondly, the question as to how how you turn the system back on is not difficult. You need to be running two versions of your system. The first version is run as if there are no index filters and all buys and sells are taken a normal. The second and real system will look at the performance of the first system, say on a nightly basis, and if the conditions meet up, the second system will take the trades that the first system takes. The second system could be looking at the MA of the equity curve of the first (as an example) to see when it should recommence the trades from the first system. Basically you need to run the two systems in parallel (moonitor the first and trade the second)
Bingk6
Historical ASX Index constituent list;
http://www.asx.com.au/about/pdf/CompanyIndex.pdf
In my experience, top down analysis ("hot stocks in hot sectors") produces much the same bottom line results as bottom up analysis.
I did find two weak correlations when doing relative strength sector analysis; Energy and Materials.
Eventually you realise that entry analysis, switches and the like are overrated and not where you should be concentrating your energy...but you have to find that out for yourself.
Post a Comment