The Best TA Book Ever Published!!
- Thread starter marketsurfer
- Start date
- Status
Rest assured that Wiley will make a profit, and that any price over $19.99 will satisfy that criterion. For example, if you get a book printed privately through Thomson-Shore, then a run of 5000 books will cost you about $4 per book, including dust jacket. Now, I'm sure that Wiley will claim that marketing, advertising, and distribution costs are significant costs, but actually they spend little on marketing niche books (see Niederhoffer's comments on his Practical Speculation book). The key is that they have the shelf space with the major chains. I don't know what your cut is David (I'm guessing 10-15% of cover price), but if you sold independently on Amazon.com and shipped your own books, then you would get 50% of cover. The technical bookselling business is vastly profitable, just not to the author, although Elder is probably an exception because of his proven titles (sales volume, not technical enlightenment).
And that's why their pricing is a "hazy mystery"
So, if I had a set of systems that I'd like to vet with a statistical framework, then Aronson's book would not be helpful in this regard?
I think you may have missed the point, HG. Marketsurfer has been born again and is now on the righteous path. He no longer trades using TA. Just ask him.
(Of course, anything other than the use of economic fundamentals is TA in one form or another, including statistically valid studies of price. But don't tell him that. It will spoil his lunch.)
He mentions certain biases which are inherent in plain everyday statistical analysis. If you don't know what those biases are then I guess that information could help you.
I would say the most important thing he mentions is to find a way to detrend the market before you do your testing so you can eliminate bias.
This book should not have been a book. It should have been an article on his research. The remainder of the book should have be entitled "The history of statistics". The most boring read I have ever encountered.
thunder, that is absolutely true, and a constant source of semantic wanking
a lot of people think that if they are not using chart formations, or indicators, they are not using TA
if u are using TAPE - it's TA
if u are looking at a price chart, and making any decisions based on support/resistance, oversold, overbought, or ANY other thing related to price (past or present) - it's TA
TA is the study of price/volume (volume is just price repeated)
all derivatives of price are still TA
a lot of people think that if they are not using chart formations, or indicators, they are not using TA
if u are using TAPE - it's TA
if u are looking at a price chart, and making any decisions based on support/resistance, oversold, overbought, or ANY other thing related to price (past or present) - it's TA
TA is the study of price/volume (volume is just price repeated)
all derivatives of price are still TA
Why do you wish you read this book 10 years ago? The only reason I wish I would have read this book 10 years ago would be that it would have been forgotten by now. Again, I ask WHAT HAVE YOU LEARNED AND APPLIED FROM THIS BOOK?
Every book I have ever read with the exception of this one, has at least given me one little tidbit of information I could use. Can you at least give us a little tidbit? I bet you can't. You mentioned earlier in the thread that you weren't good at math. How could this book possibly help you? It's all history and math.
Actually I discuss two biases. There is the long/short position bias which is dealt with by detrending the market data before measuring the rule's profitability.
The there is the data mining bias if the rule is discovered by testing many rules than then picking a best performer. In this case, the ordinary test of significance will have a higher, perhaps much higher rate of type one error ( false reject of the null hypothesis) than the significance level chosen for the test. This bias, which I refer to as the data mining bias, is dealt with by using either White Reality Check, or a new method never described before called the Monte Carlo Permutation method. Both tests are designed to give an accurate assessment the best rule's statistical significance.
By the way, despite the fact that the best rule found via data mining is has a positively biased performance, I advocate data mining, properly done, as the modern technician's best approach to discovering good trading systems.
If Holy Grail was already aware of these methods, he is no doubt a highly sophisticated researcher who reads the academic journals or is perhaps an academic. This explains why he found my discussion of statistics to be of little value - its all old hat to someone who has studied the field. What I tried to do in chapters 4, 5 and 6 was take someone with no knowledge of statistical inference all the way to the point of understanding why statistical inference is so crucial to technical analysis and how it is to be done.
I would still be grateful to Holy Grail for suggestions on how the book might be improved. Thanks
David Aronson
The there is the data mining bias if the rule is discovered by testing many rules than then picking a best performer. In this case, the ordinary test of significance will have a higher, perhaps much higher rate of type one error ( false reject of the null hypothesis) than the significance level chosen for the test. This bias, which I refer to as the data mining bias, is dealt with by using either White Reality Check, or a new method never described before called the Monte Carlo Permutation method. Both tests are designed to give an accurate assessment the best rule's statistical significance.
By the way, despite the fact that the best rule found via data mining is has a positively biased performance, I advocate data mining, properly done, as the modern technician's best approach to discovering good trading systems.
If Holy Grail was already aware of these methods, he is no doubt a highly sophisticated researcher who reads the academic journals or is perhaps an academic. This explains why he found my discussion of statistics to be of little value - its all old hat to someone who has studied the field. What I tried to do in chapters 4, 5 and 6 was take someone with no knowledge of statistical inference all the way to the point of understanding why statistical inference is so crucial to technical analysis and how it is to be done.
I would still be grateful to Holy Grail for suggestions on how the book might be improved. Thanks
David Aronson
Dr. Aronson:
Since no one seems willing to offer suggestions on your book, let me say a few things.
First, I like the content. As I have said before it is appropriate for this audience. If however you are going to offer a book with that title, you may find that those who actually buy it have a different set of expectations as to what they will find inside. Also I think that those involved in publishing books of this type feel they have to stay with a specific style when in fact they should look to the bottom line. How many sales can we expect if we market the book using this title, this jacket, etc.
I would have used a different title like "Statistical Research for Beginners" or "Statistical Research for Daytraders"
Also I would have modified the approach to include a step by step account of how you did your research. For instance I would have started with the stat package, how you obtained your sample data, how you designed the study, everything from A to Z.
I realize this is a far cry from what an academic author is used to doing, and it smacks of "Statistics for Dummies" but I believe it would reach a more receptive audience and sell significantly more copies.
I hope this helps.
Steve
Since no one seems willing to offer suggestions on your book, let me say a few things.
First, I like the content. As I have said before it is appropriate for this audience. If however you are going to offer a book with that title, you may find that those who actually buy it have a different set of expectations as to what they will find inside. Also I think that those involved in publishing books of this type feel they have to stay with a specific style when in fact they should look to the bottom line. How many sales can we expect if we market the book using this title, this jacket, etc.
I would have used a different title like "Statistical Research for Beginners" or "Statistical Research for Daytraders"
Also I would have modified the approach to include a step by step account of how you did your research. For instance I would have started with the stat package, how you obtained your sample data, how you designed the study, everything from A to Z.
I realize this is a far cry from what an academic author is used to doing, and it smacks of "Statistics for Dummies" but I believe it would reach a more receptive audience and sell significantly more copies.
I hope this helps.
Steve
I have been ignoring your question because you are a very polite and sincere person, and I don't want to tell you how I really feel. I will not go on amazon and write a bad review on the book, but I think anyone who is thinking about purchasing this book should first find it in a bookstore and read a portion of it. Should it meet their requirements then they should certainly buy it. if not look elsewhere. If I would have done that I would not have purchased the book. I blame myself for not knowing better than to listen to the recommendation of a person who has very little knowledge of the markets.
- Status