I am new to this forum and have some questions if that's OK.
I haven't started trading live yet and am really in the initial system development stage of my trading career, which I have been doing for about seven months. I have been using NinjaTrader with the Continuum data feed and Shark Indicators Bloodhound plugin to build strategies.
One of the systems I have built I have tested over eight years of data and it generates about one trade each week and on the historical testing generates on average five ticks per trade and average trade length is 2.6 days. It also has two losing years out of the eight so has gone as long as about 500 days in drawdown. The issue that I have is when I rework the parameters to increase the trade frequency the performance falls off dramatically.
So question 1: I wondered of the more seasoned traders here, especially on the ES, what is a realistic expectation of the balance between number of trades per week, average trade length, average ticks per trade and length of drawdown?
I just feel that I should be shooting for better statistics than I currently have.
Also this is using fixed parameters over the whole 8 years. I am wondering if this is part of the issue.
So question 2: Of the traders who use a strict set of rules do you keep them fixed regardless or are you constantly re-optimizing them as market conditions change. If so how often to you go through a optimization process?
Many Thanks, Alan.
Last edited by alanmillbrow; February 3rd, 2014 at 04:28 AM.
My first guess would be that you have almost certainly overfit (Curve fit) to the historical data.
You can quickly verify this a couple of ways:
a) Whatever time frame you are using, slightly change it. For example if using 5 minute bars change it to 3 minute bars and re-run the test.
b) Switch to a highly correlated instrument. For example if trading ES then switch to YM or NQ and re-run the test.
In both cases, your final results should be highly correlated with the originals (all the metrics such as win/loss rate, win/loss ratio, time in trade, number of trades, etc). If they aren't then likely curve fitted to specific data.
Due to time constraints, please do not PM me if your question can be resolved or answered on the forum.
Need help? 1) Stop changing things. No new indicators, charts, or methods. Be consistent with what is in front of you first. 2) Start a journal and post to it daily with the trades you made to show your strengths and weaknesses. 3) Set goals for yourself to reach daily. Make them about how you trade, not how much money you make. 4) Accept responsibility for your actions. Stop looking elsewhere to explain away poor performance. 5) Where to start as a trader? Watch this webinar and read this thread for hundreds of questions and answers. 6) Help using the forum? Watch this video to learn general tips on using the site.
If you want to support our community, become an Elite Member.
The following 3 users say Thank You to Big Mike for this post:
1) You didn't tell your data frequency. If you are using daily data i think that one trade per week is enough. You could use higher frequency if you want to place your positions more often. 2 years drawdown is bad, and i think you should try to figure what happened at the market that time and avoid trading in that king of situations. Setting stops can also help. Other metrics to measure the predictability skill of my strategies are (monthly) sharpe ratio (should be better than S&P500 B&H sharpe) and comparing your annual (or monthly) returns against S&P500 B&H strategy. This is one way to measure if your strategy is better than simply buy&hold strategy. Of course you can use more technical ways like bootstrapping. By comparing your strategy's profits against the underlying assets buy&Hold strategy you will see does your strategy provide any edge against B&H strategy. This is of course only for entries and exits. Position sizing is something that you should figure out careful. Expected value for your trading system is ok if it's from testing period.
2) Not sure are you using separated testing and validation period. If not you should. I dont like to re-optimize my systems too often because its hard work and by that you increase the probability that your strategy is curve fitted. Also market turns can happen pretty quick so you cant always configure your system to new environment fast enough. The way I use to build my strategies, so they work in all kind of market environments, is to choose your validation and testing periods in the way that both of those samples are facing some bull, bear and non-trending phases. This is how you can avoid validation and testing sample biases. Example about this bias is that if you are at strong bull market and you flip a coin and take long positions with tails and exit your long positions after couple of days you are going to make money. After the market turns to bear you start losing money. That is something that you should really be concerned. Let your strategy make your profits, not the market you are trading.
The following user says Thank You to jjsshh for this post:
Thanks jjsshh I really appreciate the feedback. The strategy is actually based on the hourly chart. So I think very low frequency of trades considering that. I have been reading David Aronson - Evidence Based Technical Analysis which also talks through some of the concepts you mention and more. Seems to be a great book. I'll be taking those into account going forward. Thanks again.