Tradestation 9.1: Portfolio Maestro

[Wide Tailz]

Quote from intradaybill:

Note that I did not include trend-following because it can be done with simple indicators. Trend-following success does not depend so much on method as much it depends on discipline and MM.

How would I know if you succeeded? You cannot trade stochastic processes with deterministic lagging indicators. The error keeps increasing until you blow up.

However, feel free to post the results of your success. But be sure about one thing: you will never succeed where others more capable than you failed. Judging from your posts your level of understanding of trading is very low. You should perceive this as advice, not as an offense.

So you either failed to make money at trend following or failed to make money when applying common indicators to your non trend following methods.

Judging from your posts, you don't read very many of mine. If you did, you would realize I am focused on using automated trend following methods combined with additional MM and portfolio composition rules.

I have no interest in intraday methods discovered through datamining, genetic optimization, nor do I have interest in fading a decaying system.

No disrespect intended.

 

[intradaybill]

Quote from Wide Tailz:

So you either failed to make money at trend following or failed to make money when applying common indicators to your non trend following methods.

Judging from your posts, you don't read very many of mine. If you did, you would realize I am focused on using automated trend following methods combined with additional MM and portfolio composition rules.

I have no interest in intraday methods discovered through datamining, genetic optimization, nor do I have interest in fading a decaying system.

No disrespect intended.

Also, I bet you are not capitalized properly to execute a trend following strategy efficiently.

 

[Stan_MultiCharts]

Quote from syswizard:

Stan - although I'm no fan of Tradestation, the problem was not endemic to their software per se, but of the granularity of the time series. The problem was dealing with time bars where buys and sells were triggered on the same bar. If you think about it, none of the data vendors will tell you if the HIGH of the bar was reached before the LOW or vice-versa. Because of this DATA ANOMOLY, backtesting results could look great, but the system could fail in real time testing.
This makes sense, right ?

Any backtesting is an assumption, a simulation that tries to approach reality as closely as possible. The goal of any software make is to minimize them as much as possible. In regular backtesting there is an assumption that if open > close, then price first went Open -> High -> Low -> Close. When open < close, then price went Open -> Low -> High -> Close. But this is relevant to analysis that is not required to be that accurate.

For traders that want maximum accuracy, at least in MultiCharts you can increase granularity of your backtesting (Bar Magnifier feature). You can specify how granular you want your data series to be. For example, if you have a daily chart, this feature can load 1 minute data behind the scenes and run your script with one minute intervals, which will give you a pretty accurate picture of price movements. If you need to be very specific, you can load 1 tick data behind the scenes, and you will essentially get the real price movements (depending on the quality of your tick data).

It used to be difficult to find software or computers that could work with years of tick data, but with the modern multi-core CPUs that have gigabytes of RAM, and the ability to download virtually limitless tick data online, you can be as specific as you can. You will need the 64-bit version of MC though to work with that much data, but now there are virtually no limitations to how specific your backtesting gets. Only the physical memory on your computer and the amount of tick data that you can obtain are limiting factors.

 

[syswizard]

Quote from intradaybill:

Also, I bet you are not capitalized properly to execute a trend following strategy efficiently.

Exactly. Another point: if you are trading short-term (i.e. day trading), the best strats are COUNTER-trend, not trend-following.

 

[intradaybill]

Quote from sbokov:

In regular backtesting there is an assumption that if open > close, then price first went Open -> High -> Low -> Close. When open < close, then price went Open -> Low -> High -> Close. But this is relevant to analysis that is not required to be that accurate.

No, this is a bad assumption. Never use it.

Quote from sbokov:

Only the physical memory on your computer and the amount of tick data that you can obtain are limiting factors.

Again no. The real issue here is cost of tick data. If I want to test an algorithm in all stocks in NASDAQ 100 I must spend a fortune in data.

Anyway, the issue with the TS2000i software AFAIK was that it did not execute stop-losses on entry bars. Also, it did not generate a valid entry at the open of a bar if a profitable trade was closed during the previous bar. Strange, but whoever wrote the backtesting algorithm had little relation to trading.

 

[syswizard]

Quote from sbokov:

It used to be difficult to find software or computers that could work with years of tick data, but with the modern multi-core CPUs that have gigabytes of RAM, and the ability to download virtually limitless tick data online, you can be as specific as you can. You will need the 64-bit version of MC though to work with that much data, but now there are virtually no limitations to how specific your backtesting gets. Only the physical memory on your computer and the amount of tick data that you can obtain are limiting factors.

Are both the 32 bit and 64 bit versions of Multicharts architected for multi-core processors ?
Or just the latter ?

 

[Wide Tailz]

Quote from intradaybill:

Also, I bet you are not capitalized properly to execute a trend following strategy efficiently.

I don't have any difficulty limiting risk to 2% per position while keeping slippage / comms to a tiny fraction of expected R.

But this thread isn't about how impossible trading is, it's about Portfolio Maestro and how it appears to allow multiple strategies to be backtested / WFOptimized together as a portfolio with MM tactics designed to give the ball to the best playerz on the field, and bench the losers before they damage the score............



Video tutorial is up and running now:

http://www.tradestation.com/educati...radestation-basics/quick-tips/market-analysis

 

[gmst]

I asked earlier if anyone knows the difference between portfolio backtesting capability between multicharts and tradestation. Even though noone has replied on it yet, but I have a specific follow up question and I will put it here.

If you want to test multiple strategies on same symbol, multicharts portfolio backtesting doesn't allow this. So, if you want to run Strat1 and Strat2 on AAPL, you are basically limited to setting up 2 charts and running the strategies separately. This is a big shortcoming in multicharts. Does TS Portfolio Maestro also suffer from this limitation?

 

[intradaybill]

Quote from Wide Tailz:

I don't have any difficulty limiting risk to 2% per position while keeping slippage / comms to a tiny fraction of expected R.

But this thread isn't about how impossible trading is, it's about Portfolio Maestro and how it appears to allow multiple strategies to be backtested / WFOptimized together as a portfolio with MM tactics designed to give the ball to the best playerz on the field, and bench the losers before they damage the score............



Video tutorial is up and running now:

http://www.tradestation.com/educati...radestation-basics/quick-tips/market-analysis

Tradestation is a tool for newbies. If you are an advanced trader who makes money you can afford to hire a programmer and develop custom code, backtesting, FIX, etc. Why in the world would anyone in the right state of mind implement an edge of any kind in a system out of his control? Which maybe can anticipate the orders of an automated system and frontrun them?

I am not saying TS or anyone other broker does that but an edge is an edge and you would never leave your wife in the house of the guy next door if you go to a business trip, especially if you have seen him around hitting on women hard.

So you spend your time please looking at these videos while we spend our time working with our programmer.

 

[intradaybill]

Quote from gmst:

If you want to test multiple strategies on same symbol, multicharts portfolio backtesting doesn't allow this. So, if you want to run Strat1 and Strat2 on AAPL, you are basically limited to setting up 2 charts and running the strategies separately. This is a big shortcoming in multicharts. Does TS Portfolio Maestro also suffer from this limitation?

You could just combine the strategies into one using OR statements and get over with it. Make sure you allow multiple positions.