Candlestick Automated trading

Discussion in 'Automated Trading' started by GG1972, Jan 22, 2010.

  1. GG1972


    Is it possible to scan for candle sticks or develop an automated system that can detect candle stick patterns. I was told earlier that Tradeideas could do that and after tweaking, re-doing and spending countless hours with tech support I am still unable to fix the problem.

    Mainly looking for patterns like hammers, shooting stars, gravestone doji etc. Trade ideas will pick up maybe 20 stocks everytime and 70-80% is junk and by the time you find one you could ve traded it s already moved away.
  2. Chabah


    The book "Professional Stock Trading System Design and Automation" by Mark Conway and Aaron Behle cover candlestick pattern trading in detail. I have this book and typed in every single function (available for purchase from the authors through their web site) and tested it and such. I didn't trade their strategies, though I did trade a similar strategy (rectangle breakout) I designed myself.

    The scanning and trading works. Whether the strategies are actually profitable is a different question. But this book and the software (typed on the page) do automate this sort or scanning and trading. I wouldn't be very comfortable leaving it to trade on its own though - I would want to monitor it.

    Note that the code in the book, and most of the treatment, uses EasyLanguage/TradeStation. Also this book is from around 2002 so things have changed a bit (some of the code is out of date). I believe this book is out of print, and I bought mine used for about $80. Probably is more now, but I think the author charges less than Amazon used booksellers if you go through their website.


  3. GG1972


    Thanks a lot.

    Just the scanning if its perfect or near perfect would be awesome. It would at least be a start in right direction. I wasted 3 months tweaking and re-tweaking trade-ideas but I had my apprehensions right from beginning but didn't want to give up easily. I thought I might be doing something wrong and finally after spending a lot of hours with their support have given up on TI. It's awesome for some strategies but not for picking up candlesticks.
  4. GG1972


    just googled it and doh!!

    Massachusetts district court has sentenced former hedge fund manager Mark R. Conway to seven years in prison and three years of supervised release for defrauding investors out of $20 million.

    The court also ordered Conway to pay a $1,300 fine and $20 million in restitution. On October 23, 2006, Conway pled guilty to 13 counts of wire and mail fraud in connection with his scam. According to the SEC, the District Attorney's office has taken $15 million from Conway's bank accounts to be returned to the roughly 50 investors who put money into his bogus Groundswell Partners and Groundswell Capital investments.

    The Commission based a large portion of its case against Conway on a tape-recorded phone conversation with Conway's Groundswell partner, Aaron Behle. During the October 26, 2005 conversation, Conway admitted to Behle that he had used the fund to defraud investors from roughly 2000 or 2001 to October 2005.
  5. There are many auto scan software for Japanese Candlestick patterns. Some are stand alone while others are "within" the vendor's website. However, I've not seen one that accurately scans patterns consistently let alone be profitable. Most of them scan via some generic/canned code and a few allow customization.

    Also there's a problem where most traders or investors using automated scans tends to not even understand the price action of the candlestick pattern. Another problem is not one of these auto scans are able to read the price action to recommend good trade management after entry.

    Simply, entry signals are only small piece of the puzzle and these auto scans, codes or whatever can't even get that right on a consistent basis or ignore the trade management (you're on your own).

    Here's my recommendation...trash the generic/canned codes or auto scans and hopefully you didn't spend too much time going that route. Instead, learn Japanese Candlestick patterns the old fashion way via first learning the price action of a few trading instruments prior to looking for Japanese Candlestick patterns without any codes.

    Yet, I do understand that many are trying to make their trading, testing as easy as possible via codes or automation but this is one of those things (Japanese Candlestick patterns) you got to use such manually.

    By the way, you can still be using rule-base (objective) method for Japanese Candlestick patterns but it's best to use such manually and don't trade via these patterns all by themselves. Simply, you should already understand the price action instead of letting a candlestick pattern define the price action for you. Also, they should be merged with methods involving either supply/demand analysis, volatility anlaysis, volume analysis, traditional chart patterns, market seasonal tendencies or fundamental/economic events.

    * Trading Hammers (revisited) @

  6. GG1972


    Thanks Mark

    I ve tried to watch charts for about 30-40 stocks at a time and it becomes almost impossible to keep up ( I like to know how they traded the day before for my early morning trades), how the market is trading overall etc etc Thats why lately I have exclusively traded the Q's and P's and done well. Sometimes they dont offer as many or as good opportunities as individual stocks do. Hence the quest continues for finding some way to scan them for these candlesticks. I do have other parameters that will need filling in based on actual trading. As far as price action is concerned the trading isnt going to be automated. Only the scanning part. Trade ideas came close but still too far in picking up stocks that I wanted to see. Maybe one day.
  7. Candlesticks are micro-patterns made up of a few bars and their majority looks for simple relations between the O,H,L,C. These type of micro-patterns are for the most part a subset of a larger class of patterns called by some authors "price patterns". Why looking for candlestick patterns only and not for the general class they belong to?

    There are several programs that can scan for candle patterns but most of them do not calculate their historical performance. I have found out that many candle formations are not historically profitable although some refer to them as "bullish" or "bearish".

    Nison has a program to scan for candle patterns. I have not tried it. Link

    This is a search/scan program for micro-patterns in general. Many of the pattetrs it finds look like known chart patterns or candles. Link
  8. I have written scans and auto-strats for some ohlc patterns like popular candlestick patterns in TS. TS does offer some off the self scans as well. I run a scan on 75 stocks that detects patterns in Line Break charts and works very well for my day trading. My personal opinion is that ohlc patterns are more reliable the longer the time frame gets (minutes to years) and they must be coupled other indicators to increase reliability. Also they must be paired with the proper investment vehicle (stock, future,...) as in the case of stocks, many are horrible ohlc plays even on larger time frames because of how they trade (tight ranges, gappy, wide whippy ranges, so on) and some stocks are "spot on" and will produce great results on daily, weekly charts.