I am going to hazard a wild guess: Since he told us he changed his name and what the original name was he is unlikely to be hiding the seminal work he did on behalf of the The Reich studying the effect of torture on infants. Q) Who gives a fuck that the guy changed his name? A) People on the anonymous chat board -- lol. As Johnny Carson used to say: "You can't make this stuff up."
Ahh.. The old fart changed his name because a bunch of old ladies couldn't pronounce his name? Sounds like a stupid reason to change his name. Maybe he has something to hide. How about you? Have you changed your name because your next door neighbor can't speak English? Duh. What a moron. Are you still watching old videos of Johnny Carson? How old are you? 75?
The speculation on Brooks himself is a red herring. The reasons why he changed his name, his ability to hire a competent editor, his reasons for writing a book--these are all immaterial compared to the most important question: are his ideas on daytrading the ES worthwhile? He does not provide any evidence that they are, a valid criticism of the book, so traders/readers must determine for themselves.
An example of how I look at some of these ideas from a systematic/objective perspective. At the very open he makes an assertion to go long on trending (large white body) and short on trending (large red body). Likewise, dojis are considered trading ranges. To make it somewhat objective, in this example a measure of the candle body is set as a percentage of the total daily range. The body percentage is then swept from 0 to 95% and a buy is executed from the day's close to next day's close, when certain criteria are met. The results show that bodies > 95% outperform all other, BUT have a very low frequency of occurrence-- i.e. you will get just 3.7% of these in the entire span of the S&P500 history. Conversely, if you only took trades that had bodies <50%, which occur in much greater frequency, you would have done much better than taking the trades with bodies in excess of 50%, but less than 95%. This type of evidence is completely counter to the type of assertion that he is making, and a reason why I insist on backtests before listening to the wisdom of TA pundits. Also, keep in mind returns are stated as $dollar values only, and do not take into account percentage gains, nor slippage/comm. Obviously another great factor in which, possibly only the 10 and 95% thresholds might be worthy (ironically both doji and trend candlesticks implying to take a long position at the close). P.S. Thank Swan Noir for giving me the incentive to post this, as he's been a pleasant poster to chat with. It is a very crude example, but hopefully opens doors to further investigation when reading objective TA books. P.S.1 Granted he uses 5 min bars vs. daily in this example, but you are welcome to run the same test on 5min bars. If someone wants to attach 5min bars data for whatever instrument, I'll re-run here.
Why would I give a flying fuck why the man changed his name? Smart reason, stupid reason -- it's his name!
dt ... In the early 90's a friend of mine had a Dot Machine (Direct Order Transfer) put in his study by Bear Sterns. He was obligated to average 100,000 shares a day after the first 60 days. His commission rate was stupendous for that era (the 1/8th was still King ... no penny spreads) but by today's standards very high. Early on he ws placing both bids and offers on IBM according to a fairly simplistic formula he developed and it started off with a bang -- 92 of his first 110 trades were profitable. He told me he was going to rework it to bring down his winning percentage just a bit. It took me a few seconds to catch on but I then realized he wanted many more trades and was willing to give up a few points on the win stat to get them. His guess was that while what he had was good, fewer wins (on a percentage basis, of course) might be the actual sweet spot as opposed to what I have come to call the belt and suspenders sweet spot. It can be awfully costly to be too risk averse. I'll take a look at your numbers when I have a few minutes and see if they point out something similar. My current problem is that while I have become convinced that even small levels of back testing can pay wonderful dividends for a discretionary trader -- impressions masquerading as reality can be very expensive -- my computer skills are non-existent. I know I must bite the bullet and develop an ability, even a simplistic one, to test. Any suggestions where to start?
Hi Black Swan, I think there is some truth in the statement you posited, regarding your friend. There already exist subject areas that address this issue, commonly known as optimization. To make it very brief, you generally want to operate on the flatter (I call it mesa) portion of a curve, rather than the peak. The reason is that it allows more leeway with respect to reality. The flat portion has some room to maneuver to the left and right, while sacrificing the best spot. Whereas, most likely the best spot, will shift to the left or right in future out of sample performance. Engineers often use a similar concept called 'corners.' There are many ideas and ways to approach expectation and optimizing. I've been reading a book called "Risk Trading," by a former risk analyst for Paul Tudor Jones. One comment he observed, based on experience, was that many of the successful traders he knew, made 90% of their profits on 10% of trades. This type of 90/10 or 80/20 principle often goes against the grain of parametric or normal statistics, as the fat tails are the bulk of the profit sources, not the averages. So even statistical backtesting needs to be evaluated with an open mind. I'm surprised no one caught that I accidentally ran the analysis on monthly trades (because I pulled up the wrong data source). That being said, the daily results are also telling. They are attached. If you are interested in learning backtesting objectively, I hope I've shown you can use something as simple as excel to validate ideas. In fact, I often use it as the fastest canvas to run an idea. However, there are plenty of trading backtest platforms that you can search around here or ask others that will translate and test your ideas for you (I generally write my own). Basically, I read bar by bar (ok only few opening chapters, then didn't feel the need to continue) and dozens of others like it , until I eventually realized they were often cherry picking and making ideas sound good with one or two examples, and no out of sample test performance (IBD is amongst the most egregious offenders of these types of issues, but I digress), nor objective validations. At some point, you want to try to objectively find a way to validate some of the many subjective claims put forth in dozens of garden variety TA books: I've said it a million times, but trading in reality can be a bit sobering when viewed under a calibrated microscope. Cheers, dt98
And here is a man who says he doesn't care why Al would change his name yet he is getting quite excited that someone else would ask a question about it.
dt, Your post makes sense to me -- particularly the 90/10 or 80/20 principal that is true in many more things than most realize. The two sections I've pulled allow me to give you a better sense of my current frustration. It is clear to me that I MUST -- as you say -- find a way to validate (or rule out) subjective claims and, I would add, any observations I think warrant an objective look. To disagree with that (and in a moment you'll understand why I would LOVE to) would be folly. If, as I believe, this is a tough game to beat with a foundation, without one it becomes more akin to a gamble than a business. It is easy to tell that "whipping up" a quick analysis is, for you, no more difficult than for a Neopolitan to toss a mound of dough up in the air and twelve minutes later serve up a first class pizza. Not a seven course French feast but damn good grub that does the job. Although I have played some interesting and profitable hands over the years each was approached in a very "old school" way. In one instance demographics played a key part in the business and I hired, for a half day, a demographer who lived in Cambridge and had gone to MIT, spoke to him, and then ordered census data and hired a computer guy from an ad in The New York Times; it was pre-web and that's how you hired. Although this may seem strange to you I have never so much as added two plus two in a spread sheet. My total experience with Excel is seeing my staff's output and using that as a decision making tool. Often the primary tool and at times the only tool I needed. But, again, I was looking at the numbers that had been "crunched" and never crunching them. I look at your output and it occurs to me that I don't have the slightest idea where you obtained the data not to mention what you did with it after you got it. It would seem that the obvious thing would be to have others crunch after I have explained what concept I want tested. Obvious ... yet I think incorrect. I need to work on the numbers at a simplistic level -- maybe rudimentary is a better word. I sense that in this venture moving closer to the process -- again in a rudimentary way -- will result in serendipity that may turn into an edge. And for serendipity to have a chance one must be able to wander around a subject like one wander through a second hand book shop. Not knowing what your looking for but confident you'll know it when you see it. I don't want to be a consummate tester but rather to think of it as a pair of reading glasses that makes the page legible. I want to at least be able to "whip up" a simple cut that allows me to articulate what a particular trade is and then see if it tests favorably. For without an edge that has some statistical basis the odds of being a punching bag simply escalate to an unacceptable level. I'm not trying to develop mechanical systems or think that someday I will "master" complex testing. I want to continue to view futures markets -- indexes, currencies, crude and metals -- as a discretionary trader yet I want to know that 40 - 70% is the range to look at and that after looking I want to know that it does not even come close to covering a $3 commission never mind the more realistic number of $5 per. I've looked at Amazon's offerings and none seem to be a starting point. Any ideas how I get to "Two plus Two" in a reasonable time frame? Remembering that really do need to start at the beginning. When teaching an old dog a new trick it is very important that the trick be simple ... even if he's a pretty bright dog.