The power of automated trading [Archive] - NinjaTrader Support Forum

Aggie549

12-06-2004, 03:09 AM

I am using Esignal. The problem is we are using the alternate chart type's. Renko, Kagi, pointbreak, etc. There excellent for trend trading. Problem with the stop's is that there bars don't print until completion. At this point, all the profit is given back. We are trying to get stop's activated off of the previous bar. Since I am not a programmer, getting the ideas thru to a programmer are difficult. All the test come out to good to give up. So, we will keep trying and just trade manually for now.

Travis

10-10-2005, 07:47 AM

Hi Everyone,

Does the DLL of the NT V5 works for both TS 8 and TS2000i? Any usefull info regarding writing script for automated trading?

Thanks

Travis

Fecdzo

05-21-2008, 11:36 AM

Is here anyone who managed to fully automate a strategy with Ninjatrader and using it an a regular basis? On 1 or 5 min charts or weekly or something?I really would have some questions to him/her.

I am looking for topics where people discuss their problems with automation. It is very hard work to find a strategy that really works. And then program it.

I am starting to think that it has almost no use at all. ATM is very very useful, but when it comes to programming and backtesting different strategies with crossovers etc., I get not so good results.

I wonder if some one has a strategy that performs as good as in a backtest.

mgbloomfield

05-21-2008, 12:40 PM

I spent my long nights developing the trading system in C#.

The strategy has been running live for 3 weeks now.

The strategy uses 1 min charts and executes long and short market orders for the emini S&P 500 instrument.

I intended for my strategy to run 24/7 but apparently my data feed (Patsystems) stops at 5:35 pm EST and this causes the strategy to stop and disappear.

Therefore, I have to restart the strategy every morning. Like some volume indicators that Josh, etc have created, my volume indicator is forward-looking only so that strategy is not able to compute the current position when I start the strategy. As a result, the strategy position is empty but my brokerage account is either long or short. So, I have to hack it a little to get it in sync.

I'm not superstitious but I'm not sharing the actual trading strategy. At a high level, it follows the "smart money" volume to initiate trades. I completely ignore price action.

NinjaTrader is on the right track. It is based entirely on the .NET Frameworks. There are unlimited possibilities for integration with my C# trading code. My #1 feature request is for the strategy position to match the brokerage position in real-time.

The best part is I can work during the day while my strategy makes a little money on the side. I get email alerts whenever there is a trade. That's the whole reason for doing automated trading in the first place.

Nick88

05-21-2008, 01:11 PM

I'm using NT to automate an intraday trading strategy which I apply daily on several futures. I developed the strategy on another piece of software which I am using for some years now.

To develop a strategy you need some years of intraday data. You can buy this data and import it in NT. Unfortunately NT uses a uncommon dataformat, so you have to manipulate the data you bought into the NT format. That's why I keep using my old software to develop and test strategies and use NT to trade them automatically.

It's not that hard to program a simple strategy in NT. Using the wizard makes at even simpler. I experienced some small problems with a few trades that were generated unexpectedly. I wonder if there is a problem with the code I programmed or if there is some 'bug' in NT. Most probably it's my fault. Tens of other of trades are being generated and executed without a problem, so I'm very satisfied with NT.

I used to sit in front of my screen the whole day waiting for and entering trades. Now NT does it for me. This means I can do other things if I like. Also, automation keeps mistakes in order entry to a minimum.

Elliott Wave

05-21-2008, 04:18 PM

My #1 feature request is for the strategy position to match the brokerage position in real-time...

I brought up the same point recently and Ray explained the reasons this feature isn't implemented.

The thread is here if you're interested...

http://www.ninjatrader-support.com/vb/showthread.php?t=7650

daven

05-21-2008, 05:47 PM

I understand your view, however I am also in search of the "Holy Grail" of automated trading. I have been learning C# in order to make better use of the NT platform. I have been discretionary trading, using the ATM to manage my trades, but I find that I miss a lot of setups for a variety of reasons, (boredom, breaks, phone calls, door knocks, etc.) I am in the process of implementing my trading strategy and I chose to use the strategy/ATM hybrid approach. You are correct it is difficult to test but here's what I do. You can run daily tests of ATM strategies by running the market replay function. I've been recording the full day of ES, NQ, ER2, and EMD for the last several weeks. When I have something running and ready to test, I run the replay with the ATM properly named and sitting in the DOM. I usually run at 50x so I can see how the strategy performs. All I care about is market entry since the ATM manages the position once it is entered. I've been real-time sim testing and tweaking the strategy for the past two weeks. I expect to do that for about two more weeks then I plan on going live.
I spent a lot of time on TradeStation and EasyLanguage only to discover that the brokerage and order management side of TS pretty much sucks. NT is so much more flexible, powerful, and robust. I will agree however, the learning curve of learning C# is pretty steep, particularly if you aren't a programmer by nature. I took an online course from the local community college, and I've found that it has removed a lot of the mystery of C# for me.

Fecdzo

05-22-2008, 03:02 AM

So you guys say that it is possible to program a strategy and let Ninjatrader earn the money for you while I am working?

How many people are doing like this approx,? I wonder...
This is why I started to learn Ninjatrader to automate strategies.

Is here anyone who developed a fully working strategy only with the wizard?

So far one thing I have realized: If I am trying to create a strategy that follows trends I only get about 30% hit ratio. That is clear because 70% of the time the markets are not following trends. I am having a very hard time to describe the definition of the trend to Ninjatrader, or break out.

Is here anyone who has some idea what indicator combination could do the work for describing break outs, and trends? I think this is the key and the order management, but first I would like to get to a certain point where I can continue.

Elliott Wave

05-22-2008, 03:16 AM

Is here anyone who has some idea what indicator combination could do the work for describing break outs, and trends? I think this is the key and the order management, but first I would like to get to a certain point where I can continue.

Here's a few ideas of indicators to try:
DM, ADX, ZeroLagEMA, RAVI, MASlopePlot, Candlestick Pattern, BreakoutRecognition, ModHeikenAshi etc. (if you don't find any of these in the file sharing section try searching the forum).

There are litterally countless combinations which is what makes things so interesting. :D

Fecdzo

05-22-2008, 04:15 AM

Thank you!

Do you have a good combination? Are you automating a strategy live?

daven

05-22-2008, 06:35 PM

I'm a little late getting back to you. Sorry for the delay. I do have a strategy that I built using the strategy wizard, but I think the real power of NT is the ability to tune an ATM to manage the trade exactly the way you wish. For that reason I used the Sample ATM strategy to guide me to write a strategy that passes control to an ATM to manage the trade after entry. It is difficult to test this, as I noted before, but it is a powerful way to initiate and manage a trade. Unfortunately it can't be done with a wizard alone so learning NinjaScript becomes a necessity.
I am not trying to create a perfectly autonomous piece of software that will trade on its own day or night. For many reasons, (stability of the internet connection, problems with trades not executed, a market condition not anticipated by the code....) I don't think it is reasonable to pursue the autonomous program route and unfortunately the problems, when they occur, do not have to be huge to cause a major hit to your account. I am simply trying to lift some of the drudgery of trading and replace it with a program that sees a trade opportunity the way I see it, enters the trade, then manages the trade. I personally think a good trader can always do better trading with a discretionary mindset (I underline the word "good" multiple times) and for that reason a purely mechanical system is not likely to achieve that level of success. I do think it can get close, particularly with entries, and that in itself would be a huge advantage.

daven

05-22-2008, 06:40 PM

So you guys say that it is possible to program a strategy and let Ninjatrader earn the money for you while I am working?
.

To amplify my previous message, I think it might be possible to program a strategy and let it make money, but I am too conservative to leave a computer running unattended trading my money. Other's might feel differently or have more money to throw away should a bad event occur, but I wouldnt do it, and I personally don't think it is a reasonable goal to try for someone who isn't a master of NinjaScript, (and I am not).

Elliott Wave

05-23-2008, 03:17 PM

For many reasons, (stability of the internet connection, problems with trades not executed, a market condition not anticipated by the code....) I don't think it is reasonable to pursue the autonomous program route and unfortunately the problems, when they occur, do not have to be huge to cause a major hit to your account...

I had my first near miss myself today...

I woke up this morning and noticed my laptop was dead. The power cord got unplugged somehow and I had an automated forex strategy running with simulated stops. :eek:

Luckily the market went my way and the last trade remained profitable, but it could have went either way.

mgbloomfield

05-27-2008, 03:34 PM

This is what is possible with 100% automated trading after 3 days of trading (including today). Sorry, I don't have snapshots of previous sessions.

Realized PnL: $3988
Unrealized PnL: $450

Thursday
http://i25.tinypic.com/23ts9l5.jpg

Friday
http://i31.tinypic.com/10r4y2p.jpg

Tuesday
http://i26.tinypic.com/35d7qmo.jpg

NinjaTrader_Ray

05-27-2008, 08:11 PM

Nice three day run!

mgbloomfield

05-28-2008, 04:07 AM

Nice three day run!

Exactly! :-)

For these 3 days, I only traded 1 lots. Pretty soon, I will kick it up to 2 contracts.

Fecdzo

05-28-2008, 07:33 AM

mgbloomfield

05-28-2008, 10:05 AM

The secret sauce is "Volume precedes Price".

Trevor.Richter

11-03-2008, 12:44 PM

This is what is possible with 100% automated trading after 3 days of trading (including today). Sorry, I don't have snapshots of previous sessions.

Realized PnL: $3988
Unrealized PnL: $450

Very nice! I have been trading live with my automated system for 5 weeks now. Similar to you, I trade S&P eminis and use a fairly system based on trend-based indicators. I have a positive PnL, which means my system is valid -- in 3 weeks time, if I am still at PnL > 0, I will begin scaling up my position sizing to reach my ultimate target size.

I took many months to develop an automated system before going live. I found that the toughest part was getting an indicator(s) that actually worked (when combined with appropriate management strategies such as stops) in the backtesting phase...the implementation of automation of the system via NT was at times frustrating but ultimately achievable.

PS. Just calculated my actual PnL for last 4 weeks: $4662.50

mktrend

01-05-2009, 03:00 AM

Exactly! :-)

For these 3 days, I only traded 1 lots. Pretty soon, I will kick it up to 2 contracts.
Any updates please?

Profm

02-22-2009, 02:44 PM

Very nice! I have been trading live with my automated system for 5 weeks now. Similar to you, I trade S&P eminis and use a fairly system based on trend-based indicators. I have a positive PnL, which means my system is valid -- in 3 weeks time, if I am still at PnL > 0, I will begin scaling up my position sizing to reach my ultimate target size.

I took many months to develop an automated system before going live. I found that the toughest part was getting an indicator(s) that actually worked (when combined with appropriate management strategies such as stops) in the backtesting phase...the implementation of automation of the system via NT was at times frustrating but ultimately achievable.

PS. Just calculated my actual PnL for last 4 weeks: $4662.50

Hello Trevor,
I have been looking for a system like yours. Could you share your current results ?

Trevor.Richter

04-03-2009, 01:38 PM

Hello Trevor,
I have been looking for a system like yours. Could you share your current results ?

Results for the period Nov 13, 2008 - Mar 20, 2009 are as follows:

Model*: $5,167 (+79.49% ROI)
Actual**: $1,375 (+21.15% ROI)

*This is the performance of the system performance when backtested over the same period.
**This refers to actual performance in the live market. The difference in ROI is due to 'human factors' such as (1) being out of town and not able to run the system (missed gain of ~$1500); (2) losing network connection with an open order and not being able to close correctly, e.g. because held overnight (caused loss of ~$2K).

Note: Trading is still set at 1 SP500 emini contract. Trading period limited to 9 am-4 pm weekdays only.

Note 2: I am still running this system live in beta using 1 contract. I have refined the system and will look to replace it after beta-testing the new system simultaneously also on 1 contract.

I don't think I will ever replicate the model's performance because there are always going to be issues that create less-than-ideal conditions. However, I started to live-test based on a predicted annualized ROI of ~80% assuming that getting even 50% of that would be a good 'real life' performance (i.e. expected annualized ROI = ~40%).

Would love to hear about others' experiences in 'the real world'!

Trevor