there are two problems with this TA stuff 1) you mentioned the "double-bottom" pattern. Any idiot can look at a chart and tell you, hey thats a double bottom right there. But, How would you know at the time? Suppose you have the first bottom, the recovery, and then a second decline. You need to decide at this very point in time if this is the pattern you are looking for, and not 2 weeks later. 2) which brings us to the second point: in order to test it, in practice that means writing some computer code to quantify it precisely. Try describing precisely all of the logic needed to detect the doubleâbottom and you will see that it is not so easy. In any case when you run against real data you will discover that the computer "misidentifies" certain points on the charts as double-bottom which you as a human would not call that. Haha, now you are back to square one â what is exactly a doubleâbottom? Same goes for any other complicated patterns, this is why in practice people use things which are easy to code, like breakouts and EMA crossover. Those, as you are probably know, do work but not very well.
Backtesting via computer code is only suitable if someone plans on "auto trading" (computer does the trading) or someone that's not trading (academic research). Thus, if it's not auto trading, we're just discretionary traders. Therefore, discretionary traders need to "manual backtest" (no computer code). Yep, it takes longer and is much harder in comparison to a computer code. However, it's not as flawed as computer codes for backtesting by discretionary traders... This gets into the classic problem "my real trading results is not similar to my backtesting results". Mark
Thanks for the link mokwit, I read their statistics a while and it seems head and shoulder formation still ranked very high, while double bottom is not that impressive. Regarding the discretional trading, I try to avoid that as much as possible, or to say, systemize it as much as possible. Non systematic approach for me typically resulted losing confidence when it does not work. Only the systematic approach ( e.g. "I have done all the possible research and I know I will eventually win with this method) can help me to pass the time when the system not working well.
Personally I would rather stay out once my trades stopped working. One point I did not make is that I view chart patterns more as a defined risk:reward/clean entry into a trend/pending trend rather than something to be traded in their own right - in my view in anything other than a bull market the cost of failures can outweigh the textbook targets for patterns. Some patterns are better than others in this respect.
Personally I find chart patterns too subjective to use in automated trading. The way to go IMO is carefully selected price patterns that reflect directly PA and have exact R:R levels otherwise you cannot delevelop a sound strategy to automate. It is even better if you in addition automate the price pattern search and you have a two-level automation. That is what I am doing at this point. It is difficult to put everything together but I think it worths the effort.
The reason that I want to have a look at such kind of statistic is because: I have a friend who has been using 7-8 classic chart pattern setups to trade forex intraday (mainly breakout ) and gained 300% in 3 months without a more than 5% drawdown, typically 10-30 trades per day, seems very stable. Since he has done thousands of trade, statistically it has some sense. On the other hand, I still doubt if this is sustainable, I think it is because we just had some financial storm recently, and the breakout method happened works well in this climate.
So? There has been significant research on chart patterns and they are generally anemic. "I have a friend" is about as significant as trading based on the pain in a corn on your foot... Huge amounts of patterns get tested. Then knucklehead newbies think therefore that the best performing ones must be interesting. Someone who knows statistics understands that ANY tested population of something MUST have some on top. And it means absolutely nothing.
Hi, If it is not automation trading... It's discretionary trading regardless if it's systemized (rule based) or not (intuition). However, if your a trader that needs to do statistical work to better design your trading approach... A systemized (rule base) discretionary trading approach is the way to go or go the automation trading route. Here's an example of how the results can be different between traders using the exact same systemized (rule base) trading method... * Trader A trades a specific basket of stocks only. * Trader B trades one stock only. * Trader C trades Emini NQ only. * Trader D trades Oil Futures only. * Trader E trades only the first two hours of the trading day. * Trader F trades only the last 3 hours of the trading day. * Trader G trades from work whenever he can sneak on the computer for trading. * Trader H day trades only. * Trader I swing trades only. * Trader J simulate trades only for academic research. * Trader K trades Forex only. * Traders L, M, N, O and P - L saids It's a double bottom, M saids no...it's a rising wedge, N saids no...it's an inverted Head & Shoulders then O chimes in saids it's just price noise as the market searches for a bottom while P saids you all don't know what your talking about...classic TA is stochastics, macd, rsi et cetera. Ok...you should now get where I'm going with the above. I'm not talking about the debate about if TA is art or science nor if it's subjective or objective. I'm trying to show that there's a reason why people trading classic TA patterns have different results and that's why there will always be a debate about the reliablility of classic TA patterns... There just is no uniformity (not sure if that's a word) from one trader to the next trader or from one backtester to another backtester et cetera. We all have a different perspective and/or application. More importantly, our real money trading results is determined by many variables and not via a classic TA pattern alone. The above doesn't hold true for backtesting results because backtesting is intentionally setup to remove the influence of those other variables. ----------------------- That's why we often here newbie traders say the following... I just don't understand what went wrong with my live trading... It worked great when I backtested it. ----------------------- That's why some traders go the automation route...they want to remove the impact of those other variables. With that said, lets now say each one of the above types of traders publish an article with results about that exact same classic TA pattern they traded. I guarantee you the results will vary greatly from bad to good. My point with all of this... No matter who's article you read or trust... You will need to make a choice. Trade exactly like the way the other guy traded or tested... Trade via your own systemized rules. There lays the great TA debate... Some will say it works and some will say it doesn't work. Now, getting back to your original question... I've tested a few of those classic TA patterns via my own rule-based (systemized) trading approach via real money trades, simulator trades and backtesting only (no trading) by more than 1000 trades. I put more weight on the merits of real money results. However, I've met many (including here at ET) that ignore real money results and perfers to only look at backtesting results... While others look at real money, simulator and backtesting to determine the merits or reliability. There's more to the story now. Your looking for someone that's done thousands of real money trades via day trading Forex Currencies via classic TA patterns. Further, your looking for results much longer than a 3 month time period because you seem to have doubts that the current trading environment (financial and economy problems) won't continue. Note: That's not exactly what you said and I intentionally twisted the story a little. The reason why I twisted your story is that you made a direct correlation between your friends success over the past 3 months to what you called a financial storm. Therefore, we can assume that if the financial storm (economic problems) doesn't improve... Your friends results should continue. I strongly believe the current state of the financial sector or economics will continue for many years. Thus, based upon your correlation... Your friends success should continue for many years. By the way, I know several traders using a break out method (off the top of my head I know about 10 types of different break out methods)... They are not having any success over the past several months of trading due to this financial storm as you call it. However, they don't day trade forex. Last of all, here's the forex thread. http://www.elitetrader.com/vb/forumdisplay.php?s=&forumid=38 I have read a few message posts in the past within that thread above where some traders have tested a particular TA pattern of a specific forex trading instrument. You may find whatever your looking for in that thread. Regardless, eventually you will need to sit down and say the following... I'm going to use his/her systemized rules about his definition of a particular classic TA pattern or I'm going to use my own systemized rules about what I think is a specific classic TA pattern. P.S. I don't day trade Forex. Mark
Good point...selection bias...it leads to a disaster if you try to trade these patterns on their own right. But using patterns in combinations and with other indicators you can remove the selection bias.