|
  |
|
|
Are Automated Trading Systems Worth Anything?
By Jobo Smith
Ever since the internet sprung up in a big way, people have had access to stock market information like never before. Where 15-20 years ago it would be rare to have access to real time stock quotes and charting software unless you were a professional trader, it is now very common and such tools are often given away free as long as you open a trading account. With the advent of all this technology also comes the constant push to find the "tool to beat the market". It is this drive that keeps people programming computers and researching ideas. While the idea of a program that can not lose is complete absurdity, there are tools and trading systems out there that do work. The key is realizing that its a TOOL not an absolute solution to the hidden money tree. Assuming the tool or system is legit or not depends on how it was developed. The back test is really kind of meaningless unless the developer has gone out of their way to test on out of sample data repeatedly. What this does is immediately tell you if something is curve fit (meaning the rules are written to match the data given in the most profitable manner). After all, if you are thinking of following or trading a system with real money, that is not a back test - that is real world trading. If the system cannot perform on out of sample data (meaning you test it on 1 year of data, then put it on 6 months of data it has never seen), how would you expect to actually make anything but a loss in real life? After all, real live trading is definitely "out of sample". Assuming it does o.k. in the out of sample trading, you then need to assess the actual per trade net of the system. In general in real life, you can assume 1 tick slippage on entry, 1 tick on exit (for futures, for stocks you would want to give it 3-5c each way on most stocks), and then the commission. Whatever this total is, your system needs to have an average trade at least 175% of this number. So if a tick is 5 bucks, and commission round trip is 5 bucks to trade a futures contract, that is $15.00 in costs. Your average net per trade needs to be at least 1.75*15.00 or about 25.00-26.00 per trade to have a reasonable expectation of winning. The goal here is not to get bogged down in exact math, that is not important. What you are trying to do is avoid even looking at a strategy that cannot overcome this hurdle. Do not fall into the "limit order" myth - sure you can use limits, sure they do go off - but not always. The main issue is always the entries go off fine, its the exits where it just so happens to tick there then back off and you are not filled. I know this from personal experience. In general you cannot assume a fill, so you have to figure in you going in at market when a trading signal happens. If you don't slip or can do better, its all gravy. Another issue here to think about is getting out of sequence. What I mean by this is you have a limit order waiting that did not fill (you were in line, but either did not fill or got a partial fill) and now the market has moved away. The trading system wants to do another trade but you are still stuck in the last one - this causes you to be out of sequence with the trading system OR cause it to not do a trade it would normally do. No trading system I know can contemplate this event really, nor the expected impact. So the goal is to not have this happen - murphy always works here, odds are the missed trade will be a winner AND it will be in the opposite direction you are currently placed in, waiting for your limit fill. So, assuming it passes this test also, you have something that is probably worth trading. Most strategies are best used as tools not fully automated. The reason is the market dynamics shift from day to day, often in subtle ways a strategy cannot contemplate. So you have a few solutions here. Let it trade a simulated account, then you decide when to trade a signal with real money and put it in manually, or you have the system trade for you and then you confirm each entry and reject the ones that do not look good because of market conditions. Another thing to keep in mind on any strategy - usually it cannot contemplate news events - that is up to you. Do not let it trade on a fed day or at the very least through a fed decision - the crazy volatility is completely unpredictable most of the time and is more akin to pure gambling than actual trading. Make sure you have reasonable expectations no matter what the strategy has done in the past. Most strategies have a type of market that is ideal, and many that are sub ideal, but it does o.k. When its in the ideal market, the equity curve goes straight up. When its not ideal, its much choppier, bigger draw downs happen and it can even cause you to doubt it works. While choppy action is fine, most strategies should never break more than about 1.5* the maximum draw down over the test period, including out of sample. If it does, there is decent odds something is wrong. Of course you have to look at WHY and HOW it happened. If you held stocks overnight with a strategy and one of them blows earnings and is down 45%, that is an outlier event that really does not indicate anything about the merits of a strategy other than it picked a crappy loser that gaped down on news. You can never predict overnight news, so you have to filter those extremes out BOTH ways. You need to look at how a strategy makes money and loses money to tell anything. Using these guidelines can at least point you in the right direction when trying to assess if an automated trading system is worth anything. The open ended question is always "What is the best method to make money using it". That one is not easily answered, but in general if you can follow rules and limit risk and are disciplined, you can do better manually trading a lot of the signals. If you are a gambler and tend to ignore rules and let greed creep into your trading, then I would say the only method is automated trading, with no intervention other than in completely obvious times ( fed announcements etc). |
Day Trading
| Stock Day Trading Robot
PLEASE VISIT THE CONTRIBUTOR'S WEBSITE
No reactions yet.
Please login or sign up to rate this intel.
Please login or sign up to add a comment.
The copyright for this content entitled "Are Automated Trading Systems Worth Anything?" has been specified by the contributor as:
Creative Commons Attribution-No Derivative Works 3.0
Details 
This content may be copied and distributed (but not modified), as long as the original author is acknowledged with a link back to the content page.
If you use this content according to the license specified, you must link to the following URL:
http://mytradingrobot.qondio.com/
|
 |
May, 2012
2008
January, February, March, April, May, June, July, August, September, October, November, December
2009
January, February, March, April, May, June, July, August, September, October, November, December
2010
January, February, March, April, May, June, July, August, September, October, November, December
2011
January, February, March, April, May, June, July, August, September, October, November, December
2012
January, February, March, April, May
|
|
Not a member yet?
Qondio is a powerful network for making it online. If you have a website to
promote, we can help.
Sign up and get in on the action.
|
|
Welcome to Qondio! Discover the awesome power this network can deliver by going to our About page. Or you could skip straight to the Sign Up form.
|
|
