nukethewhales31
Registered: Sep 2008
Posts: 738 |
01-15-09 05:25 PM
Quote from bilbod:
Hello NTW,
I did a preliminary statistical study of your trading method. It appears that you have found a tradeable edge. I don't have time to do a complete analysis (maybe later in the year) to thoroughly test it and possibly improve it but I can pass on some ideas based on my observations that you may want to research.
The reason I have an interest in your method is because years ago we traded an 'opening range break out' method on daily bars (stock index futures) that was based on a similar statistical concept.
1. Choosing the open of a bar to start with is arbitrary (not statistically determined to be the best place to start). It is the price at some fixed point in time like 10AM. What is special about 10AM? Why not use 10:05AM or 9:55AM? When we traded from daily bars, there was no over night or electronic trading so using the open made sense, it doesn't for intraday bars.
2. A 21 bar look back period is arbitrary (not statistically determined to be the best length). We used a statistically determined look back period to create the BO range, it was based on volatility. I don't remember the exact look back period but it was less than 10 (possibly less than 5). Observation: if you look at a lot of charts, narrow range bars tend to be followed be wider range bars and vice versa. That looks like some thing that could be statistically exploited.
3. Using a spreadsheet to do your analysis is inefficient and prone to errors. Most charting programs have scripting capability and you can use that to do your statistical analysis, then display the results on your chart. It will be less time consuming, eliminate a source of errors and make it easy for you to monitor multiple markets to find the best trading opportunities.
A possible solution to the arbitrary time dilemma (#1): Since we are trading price, instead of using a time point to start, it makes more sense to use a price point. A price swing high or low would be a reasonable choice. What I am suggesting is statistically determining swing points, e.g., when price reverses, what is the probability that if it moves x points in the new direction that it will continue in that direction for y more points (x will become the BO size and y the profit target). Also, since time plays no role in statistical price trading, it makes more sense to use range bars instead of price bars.
You may find this book interesting:
“Day Trading with Short-term Price Patterns and Opening Range Breakout” by Tony Crabel
Bill
gheeze im not trying to give away all my trading secrets just helping people become to the idea that they can increase what they already do with stats tho it will take some time and thought a research...
if you give a man a fish he eats for a day but you teach him to fish he eats for a lifetime right?
anyways im not that great of a programmer i have a book by my trade desk C++ for dummies but ive only gotten through about 40 pages.
|
