Greetings all,
First, I am glad to see much ado here about X-DEV...hats off to Myst!
Second, I would like to discuss some recent tweaks to the system that I've made to see if anyone has been down this road and can advise of any pitfalls. Please keep in mind that my usual condition is to be hopelessly behind the times. If any of this has been preveiously discussed, just point me in the general direction of the post and I will follow from there.
What I have been tweaking about for the last day is a mod to the system which uses the typical buy points generated by X-DEV but then uses sell stops with predetermined limits for either a gain or loss. This version goes for the gusto, either 100% in or 100% out at all times. No partial vests.
The reasoning behind it is to pay close attention to the buy points and then let typical money management techniques take over. Sell at specific gain or loss percentages based on historical data.
The sheets that I've constructed can be modified for the SMA period for buys (sells based on SMA are no longer considered), and then I can vary specific buy and sell points based on the gain or loss of the most recent purchase.
When a buy point is hit, 100% of the stash is invested. Then the system waits for either a certain percentage gain or loss, depending on optimization for past data. When the numbers hit, it's 100% out.
What I've found is somewhat interesting. Cutting to the chase, here are the results for INTC and MSFT based on five years of data (1/2/97 - 5/6/02):
(I call this tweak, 'X-DEV Stop')
INTC
X-DEV
Buy limit 0.91
Sell limit 1.10
SMA Period 13
Performance +159%
X-DEV Stop
Buy limit 0.91
SMA Period 12
Stop Gain 10%
Stop Loss 10%
Performance +240%
AIM
Performance +50%
B/H
Performance +46%
MSFT
X-DEV
Buy limit 0.94
Sell limit 1.12
SMA Period 16
Performance +262%
X-DEV Stop
Buy limit 0.91
SMA Period 13
Stop Gain 27%
Stop Loss 9%
Performance +398%
AIM
Performance +45%
B/H
Performance +63%
To clarify, in the case of INTC, when the 12-day SMA hit 0.91 or less for the first time, the system invested all the cash at once. All shares were sold as soon as the price went up 10% or down 10% from the initial purchase point. The system then waited for the next buy signal.
In the case of MSFT, the system kicked in when the 13-day SMA hit 0.91 or less for the first time. Selling occured when the price went up 27% or down 9%.
Please keep in mind the following:
The basic X-DEV results are based on the system as of about 10 days ago. There could be some better tweaks that I am not aware of.
These results are based on 5 years of data. All settings are optimized for best results in that time frame.
These are very preliminary results, and could be subject to 'bad cell syndrome' which can plague early spreadsheets.
Some of the things that intrigue me:
The performance gain over the basic X-DEV system (although recent mods could already bypass all of this).
Performance is not based on the initial entry fee like AIM (and sometimes X-DEV) seems to be. The percentage change is the same for $1000 as it is for $30000 (excluding commissions).
Performance seems to hold up during up or down markets.
Some of the things that concern me:
I could be flat out wrong, 'bad cell syndrome', also known as not being able to see the spreadsheet for the cells.
I haven't worked the data over longer or shorter time frames.
The constant need to re-optimize for every stock.
That's about it really. If anyone has a take on this or any questions, let me know.
And keep up all of the good work!
stiv
