Adventures in Automation

[fan27]

Was I right

I answer that when the position is closed.

 

[fan27]

Long 2 ES at 2837.75 for 15 min bar strategy.
Trailing Stop at 2822
Limit at 2885.75
Exit after 10 bars

Exit 2 ES contracts at 2868
+$3016.80

 

[fan27]

Was I right

Now it is safe to say your were right

 

[fan27]

Deployed a new strategy portfolio this morning (92 Long strategies and 4 Short strategies). I will continue adding new short strategies to the mix.

 

[HobbyTrading]

What kind of financial instruments are you using in these strategies?

 

[fan27]

What kind of financial instruments are you using in these strategies?

Equity Index futures (ES, NQ and YM).

 

[Daniel.a]

A few notes on my approach. I found there are many strategies with equity indexes on the 15 min, 30 min and 1 hour time frames where the likely hood of an up move is greater than a down move and it can be taken advantage of. The thing is the trade occurrence for these strategies is not that frequent. For example, a strategy may only have 100 trades for the past 15 years. For those that say that is not a big enough sample size, I have created tools which can empirically indicate how often the strategy beats random entry (i.e. 97% of time for 5000 random trade backtest runs). The key in my approach is to have many strategies that might not trade often but together give enough trades to make it worth while. All strategies have a combination of the following exit criteria with all strategies having a stop.

Limit order based on a multiple of the ATR
Fixed or trailing stop based on a multiple of the ATR
Exit after X number of bars.

A few examples:
limit = entry + (5 * ATR(14))
fixed stop = entry - (3 * ATR(14))
Exit after 20 bars

trailing stop = entry - (3 * ATR(14))
Exit after 8 bars

Also, all strategies use at least one higher timeframe with all strategies incorporating the daily timeframe. In the video posted above there is an example strategy shown in the Code.txt file about halfway through the video.

@fan27 thanks for sharing your work... i am facing some issues sometimes where sample size of trades are to few, same as you mentioning, around 100 trades .. so i am looking to create a tool to further test my strategies.. you are saying that you have created a tool to see how often your strategy beats random entry.... would it be possible for you to expand on your approach on this? i am looking into creating something similar ..... are you just doing this trough a monte carlo run end then randomise the entry for each variation ?

 

[fan27]

@fan27 thanks for sharing your work... i am facing some issues sometimes where sample size of trades are to few, same as you mentioning, around 100 trades .. so i am looking to create a tool to further test my strategies.. you are saying that you have created a tool to see how often your strategy beats random entry.... would it be possible for you to expand on your approach on this? i am looking into creating something similar ..... are you just doing this trough a monte carlo run end then randomise the entry for each variation ?

What platform are you currently using for backtesting? I will try and give you some ideas based on the answer.

 

[Daniel.a]

Hi @fan27 i am Using Multicharts for building, testing and execution of my algos, towards IB, then using MSA from adaptrade for evaluating my collected portfolios, Multicharts lacks lots in portfolio evaluation... i am looking into expanding my tests to include analysis of randomination of entries and possible exit with Montecarlo... i am however interested in understanding how you have included your tool fo revaluation randon entry... Multicharts can now also be connected to python so my options are quiet open...

 

[fan27]

Hi @fan27 i am Using Multicharts for building, testing and execution of my algos, towards IB, then using MSA from adaptrade for evaluating my collected portfolios, Multicharts lacks lots in portfolio evaluation... i am looking into expanding my tests to include analysis of randomination of entries and possible exit with Montecarlo... i am however interested in understanding how you have included your tool fo revaluation randon entry... Multicharts can now also be connected to python so my options are quiet open...

With FasterQuant, it is trivial to execute a backtest X number of times and my backtest component can take a collection array indexes which map to a point in time in the data series where each index will be used as an entry signal. This makes it very easy to randomize the entry indexes and test random entry against specific exit criteria. I have no idea how you could accomplish something similar with MultiCharts as you limited to what they offer via the platform.