Hello Blaine,
Please forgive for my going out of sequence but this one will take a little longer than the one to Richard.
First let me say that Richard is absolutely correct in that you must know "what kind of trader you are". This means that you must ask a few questions of yourself. This is going to be pretty superficial but hopefully some of the older more experienced guys will add to what I put down.
1.Do you mind if your money stays in a stock for a long time or do you need to trade?
2.What kind of drawdown can you live with? 1, 5, 10, or 20%.
3.Will you sell a winner????
4.Will you sell a looser??? My greatest failing!
Personal observation: Before I realized #4 about myself, I sometimes found myself in a looser. My problem was, I would not sell. I would sit on it till it either came back to break even or went belly up. I'm sure this is not uncommon but many might not admit to it.
Since this is my greatest weakness it is the most important thing about any system that I use. Any system that I use has to get me out with no more than a 10% loss. I always set my stops according to the price, volatility, and the S/R of a particular issue but it is never more than 10%. This is rule #1 for David. This is what is meant by "know what kind of trader you are". Build your systems to guard against your weaknesses and utilize your strengths.
Now about testing: I hope you get to come to NJ. because this is what I did my report on and it is much more comprehensive than what I can put here. Also these are my ideas and others will have different ones I'm sure.
Testing is a two part exercise. There is testing correctly and then there is the interpretation of your results.
When you test the following is a good basic set of rules:
1.Set up your indicator to act only on the price of stocks that you would actually buy in the market.
2.Set up your indicator to act only on the volume of stocks that you would actually buy in the market.
3.Always include commisions in your p/l calculations.
4.Some will disagree with this one but include your stops in your tests if you are trying to get a feel for what kind of money you can expect from the system.
5.Always have your p/l numbers in percents rather than points.
6.When comparing two systems you must make sure they both have the same amount of trading time. For example if you were testing the two systems below:
#1: mov(c,5,s) > mov(c,21,s)
#2: mov(c,5,s) > mov(c,55,s)
You should run one set of tests with a certain start date for the 1st system and then a second set with an earlier start date for the second system. The second ones' start date should be 34 days prior to the 1st systems start date.
7.)Use as many yrs data as you can get your hands on.
8.You should test your system in all kinds of markets:
a)uptrending to see if you like how it handles the ride up the hill
b)downtrending to see if it gets you out and keeps you out of the downtrend.
c)flat or ranging to see if it whipsaws or not.
9.Test with the types of stocks that you will actually buy.
A couple of things to remember about interpretation:
1.How long were you in the trade? If system A makes 50% in 100 days and system B makes 25% in 50 days they are exactly equal.
2.If you didn't include stops in your testing did you go through alot of drawdowns in your trades?
3.What was the average win value?
4.What was the average loss value?
5.Did you get mostly winners or did you get mostly loosers with a few really big winners?
The point is you must trust the system that you are going to trade with. For you to trust it you must have tested it thoroughly, interpreted the results correctly, and be able to obey it. To obey it, it must fit your personality and keep you out of situations that you can't handle ie:never let David get into a bad looser cause he'd rather eat dirt than sell it!<ggg>
The more I write on this the more obvious it becomes to me that I don't express myself well in writting. Actually, I probably do worse in person. This is a very broad field and many, many pages could be written. This is my meager attempt at a start. I hope some will pick up on this and add what I've left out because I feel that this a topic that we don't do justice too. Testing thoroughly is very important.
David |
