I discovered these forums this morning and spent about an hour lurking on my schools computers. During that time I came across a recommendation for an ebook that I looked at and it appeared to be very informative. Unfortunately I was too dumb to write down the name thinking I would be able to find the thread again. Now I am back home and I can't. It started with a P and was either 5 or 6 letters long. Any help would be appreciated.
Never mind, found it. I was looking for Design, testing, and optimization of trading systems by Robert Pardo.
Well, it turns out I got a little confused. I found a preview of the book that only has about 1/3 of the actual pages and none of the important information. When I first looked at I just reviewed the table of contents and assumed the entire book was there. Here is a link to what I found. http://books.google.com/books?id=LT...oi=print&ct=title&cad=one-book-with-thumbnail
i dont own any pardo books "yet" but i read his articles in s&c and he and perry kaufman are the kinda guys you want their take on things, as well as dennis myers. and of course toby crabel only has one book but he's the real deal. actually toby did two books but one money cant buy and the other is around 700 buck's right now. mb
I am glad to hear that you feel he has relevant/important things to say. Thanks also for the other names, I think I will buy a book by perry kaufman as well as Pardo. I wouldn't mind Toby Crabel's book but do not think I can muster up $700 at the moment since pretty much all my spending money is going to pay my programmer for coding my system.
FYI, Kaufman's book was written in 1978. It is now on the 4th edition but isn't dramatically different from the 29 year old original. Here's a little gem that Kaufman hasn't changed one little bit since the first edition; it can be found on page 133 of the C-2005 4th edition: It sounds as though Kaufman believes the dice have memory. Pardo wrote his book in 1992. It is a product of its times and assumes 1992 era retail-grade testing software (Tradestation and his own (failed) software product, Advanced Trader). So he limits himself to tests that trade one instrument at a time, with fixed position size: one futures contract. Today, fifteen years later, the retail software market offers products that can test very large portfolios of instruments, traded simultaneously out of the same account, using dynamic positionsizing. Pardo's book offers literally no ideas how to design or test such systems. Yet today you get two such systems in every $4.95 issue of Active Trader magazine, with full system explanation including software source code, plus portfolio level test results. The world has changed since Pardo put pen to paper. Pardo's book is a good general introduction to some basic testing ideas, but it covers only the rudimentary beginnings. I would look for books written (not re-issued, written) after 2005, for discussions of trading systems in today's world.
Does this statement mean you believe the markets are random or that the markets (not the dice) don't have memory? "Do you really think that's air you're breathing? Hmm..."
i belive in reversion to the mean yes the dice have memory. if being up to date means knowing less i will take the oldies. http://www.purebytes.com/archives/omega/2001/msg10719.html
Two things. One, do any of you have any recommendations for a book that would guide me in the process of backtesting an algorithmic trading system better than Pardo's? Secondly, I am curious about the "dice having memory"; how do you propose that the third roll of a dice or flip of a coin has any possible probability difference from the first. I understand that stocks do since each outcome is dependent on a persons choice to buy or sell and that person will have the knowledge (theoretically) of all the previous sales. But dice seem random. Sure the probability of three straight coins flips being heads is .5*.5*.5 so yes if you have to place your bet before any flips the probability of the third flip being the same as the first two is 12.5% but once the first two flips are completed the probability of a heads or tails should still be .5. The probability of a coin flipping out to be head-heads-tails is also 12.5% when looked at from the beginning. Also, doesn't this contradict any type of trend trading? When you assume a reversion to the mean are you not saying that stocks do not trend?