One of the benefits of designing your own system -- as opposed to buying one -- is that you can include the features you want and exclude that which you do not. The system can be, and should be, customised to suit your personal preferences. This also helps you to stick to the plan in those inevitable periods of drawdown.
So from the word go, from when you begin the systems design process, you should already know what sort of system you want to design, and which features are more important to you than others.
If you like a smoother equity curve, then have a shorter holding period (higher stock turnover) for your positions. This can be accomplished by having a tighter trailing exit (probably at the expense of profits).
If you cannot handle long losing streaks, then a way around this is to design a system with an entry which will give you a higher win%.
I was looking at improving my win% yesterday by using different universes as I currently use the whole market. So I tried the all ordinaries, ASX100/200/300, small ordinaries, and dividend paying securities.
Testing was the first 3 quarters of this year, 01-01-2007 until 30-09-2007, in an attempt to minimise inclusion bias (indices couldn't have changed too much throughout the course of the year). Nevertheless, it's only for relative measure anyway.
Results were as follows:
All Ords (488 stocks)
#Trades: 32
MaxDD: 2.73%
Return: 17.83%
Win% = 59
R/R = 2.06
Small Ords (199)
#Trades: 26
MaxDD: 3.97
Return: 14.58%
Win% = 50
R/R = 2.60
ASX100
#Trades: 37
MaxDD: 17.33%
Return: -15.79%
Win% = 23
R/R = 0.63
ASX200
#Trades: 41
MaxDD: 9.87%
Return: 6.84%
Win% = 37
R/R = 2.58
ASX300
#Trades: 39
MaxDD: 3.39%
Return: 21.51%
Win% = 61
R/R = 2.30
Div paying (645)
#Trades: 37
MaxDD: 4.65%
Return: 23.13%
Win% = 54
R/R = 3.08
Whole market (1951)
#Trades: 43
MaxDD: 4.57%
Return: 23.75%
Win% = 63
R/R = 1.93
The best returns were from the whole market, dividend paying securities, and the ASX300.
So I did 1000 monte carlo runs of each of them with the results as follows:
The format for profit is max/average/min/std deviation/%of portfolios profitable.
And for maxDD max/average/min.
Whole market
Profit: 55/18/-7/9.68/97.8
MaxDD: 17/6/1
Dividend paying securities
Profit: 43/12/-11/8.78/93.5
MaxDD: 17/6/1
MaxDD: 17/6/1
ASX300
Profit: 43/12/-14/10.34/88.9MaxDD: 17/6/1
The last figure on the profit column is one I pay the most attention to. 100% would be ideal.
To give some sort of benchmark the XAO (All ordinaries) was up by 12.5% over this 9-month period.
So it seems from this test that we can't expect an improvement by trading an alternative universe, at least from the options tested.
I will be adding $20,000 of trading capital to the system this week just to make it compound quicker as we are approaching the month of November. November to April is traditionally a very strong period for equities.
From the weekend scan, the full list of possible candidates are as follows:
AQA, BEC, BLY, BKN, CPB, CEY, DWS, FLT, GMI, JBH, LEI, LGL, MSL, MIN, NCM, NOD, RCR, SMX, SDG, UGL, WTP, WOR, WTF.
I applied 3 filters. The chart has to be good. Clear uptrend and not in a trading range.
This narrowed it down to AQA, BEC, BLY, LGL, BKN, MIN, NOD, RCR, SMX.
Then I checked if any were under takeover. None were.
Then I picked the lower price securities or the ones I liked. LGL a gold play I just had to buy it.
MIN and BKN even if they didn't trigger, I could pyramid because i have reached 15% profit (that's what was backtested).
Since I could probably afford 3 more positions, I'm going with LGL, NOD, and BLY.
Here is a website I check to see Broker consensus. It's another filter I've added. I like to see a minimum rating of buy or strong buy.
Remember the system is robust and well tested so it doesn't matter HOW we pick the stocks from the candidates spat out by the scan. All the testing initially was done by picking random trades (monte carlo analysis).
Ranking by price and even pyramiding trades, only adds 1-2%p.a.
No comments:
Post a Comment