Programming Your System

[brownsfan019 17]

So I've been thinking more and more about possibly programming my trading methodology. I have no idea where to start since I am not a programmer. I'm not even sure if what I am looking for can be programmed... It's not as simple as when X indicator and Y indicator do this, buy...

 

Just looking for some ideas since we have some programming guys here and maybe some are on full/partial automation...

 

Thanks!

[Blu-Ray 10]

If you want any help programming or want me to program something, just give me a shout and I'll do my best.

 

With regards to partial/full automation, I have programmed various codes to fire off signals, but they're not black box systems, they all require discretion, and as you say with your methodology is not a X/Y then Buy......

 

So I think it all comes down to what you're expecting to come of it.

 

Cheers

 

Blu-Ray

[MC 10]

Same here...I am leaning towards ninja and want to automate but don't have a clue on programming. What does it cost to have something coded I've wondered and how can you be sure your edge won't be compromised as well?

 

Also can code or indicators be locked down so they can be sold without fear of the actual backing code being given away? Of course with the talent here we've had almost all hot indicators duplicated from looks alone, so that's never gonna go away 100%.

[brownsfan019 17]

If I did it, I'd really want a full blown system where it does it all.

[Blu-Ray 10]

Same here...I am leaning towards ninja and want to automate but don't have a clue on programming. What does it cost to have something coded I've wondered and how can you be sure your edge won't be compromised as well?

 

Also can code or indicators be locked down so they can be sold without fear of the actual backing code being given away? Of course with the talent here we've had almost all hot indicators duplicated from looks alone, so that's never gonna go away 100%.

 

mcichocki

 

I'm unsure about Ninja, but you can with Tradestation.

 

Cheers

 

Blu-Ray

[Blu-Ray 10]

If I did it, I'd really want a full blown system where it does it all.

 

Thats not a problem and it is possible, but you would need a full list of all the criteria required for both entry and exit.

[Sparrow 10]

It's possible for NT to write properitary indicators and sell them, you can choose between giving someone your indicators as plain code or a dll. I am sure there are other more professional options available to lock down your code.

As for having someone code your strategy, the only thing you can do is have them sign a contract which would make it impossible for them to legally use or pass it on your system.

 

If someone is really determined and they've got the skills and access they'll crack your system/indicators, nothing is really 100% secure.

[BlowFish 10]

Personally I'd stay away from the obvious charting packages. They really are not robust enough unless you are going to sit and watch which kinda defeats the object. The obvious place to run this stuff is at your brokers where they have (if they are any good) suitable redundant infrastructure. I have never used them but maybe check out strategy runner. They also have in house programmers to help customers with automation. Quite a few brokers offer to run SR strategies. Step one is writing a detailed enough spec that people can quote on it, you can have placeholders for key elements if you are worried about plagiarism.

 

I know a couple of people that have done it themselves but they have either been pretty hot Java or C guys or had access to pretty hot Java or C guys (friends or colleagues). It is possibly a greater undertaking than you might imagine. Long story short you aren't going to want ninja, ts, amibroker, ensign, or whatever running on some virtual machine in a data centre. Of course if you are going to run it on a PC in your office while you watch TV with one eye and make sure everything is OK with the other that's a far more trivial proposition.

 

Cheers.

[Hlm 10]

It is possibly a greater undertaking than you might imagine.

This is such a TRUE statement. However, I would remove the word "possibly" . In my experience (and others I have talked to) automating a real trading edge correctly is extremely difficult. It can be a real eye opener to the way you actually trade. The big danger when automating is getting off track. When trying to simplify your trading to fit nicely into signals you will discover that your profit will drop dramatically. This will create a tendency to curve fit areas without realizing it which is very dangerous. Over the last six months or so I have carefully traveled down this road of automation (as many of you know). So far my total research, development, and programming has totaled several hundred hours. It has been a very interesting experience to say the least. If you want to bounce some ideas off me feel free to send a message my way. I wish you luck. Edited March 27, 2008 by Hlm

[BlowFish 10]

Many moons ago I used to work as an IT consultant and whilst I always called the truth as I saw it I got into the habit of using a qualifier. My girlfriend was always telling me off for prefacing everything with 'I suspect...' Old habits die hard I guess.

[MC 10]

Many moons ago I used to work as an IT consultant and whilst I always called the truth as I saw it I got into the habit of using a qualifier. My girlfriend was always telling me off for prefacing everything with 'I suspect...' Old habits die hard I guess.

 

Great input from you all.

 

This post was funny BF, I work in a helpdesk/consultant role over the phone but the challenges are much the same as IT really. I often, even before I started watching the market would always throw a slight disclaimer in there to cover myself on calls. In the technology industry you kind of have to since you often have only a partial picture from the customer.

 

Good thing too, it carries over well to the market since we often have a bias that can be upset by one sideline participant coming in and changing the game for us.

[darthtrader 10]

Personally I'd stay away from the obvious charting packages. They really are not robust enough unless you are going to sit and watch which kinda defeats the object.

 

Cheers.

 

I don't know if you should lump Ninja in there though with the others. Honestly, I think this is exactly why in 5 years Ninja will have blown everything else away. Its really nothing more than a C# compiler, if C# is robust enough for the project then so is ninja if you have the programming knowledge to do it.

Something I would think that could be interesting is to just hire a computer science student to work for you part time. Could probly find some wiz kids who are going for masters in CS on the cheap and who would be enthralled with the project idea.

[mezameo 10]

I don't know if you should lump Ninja in there though with the others. Honestly, I think this is exactly why in 5 years Ninja will have blown everything else away. Its really nothing more than a C# compiler, if C# is robust enough for the project then so is ninja if you have the programming knowledge to do it.

Something I would think that could be interesting is to just hire a computer science student to work for you part time. Could probly find some wiz kids who are going for masters in CS on the cheap and who would be enthralled with the project idea.

 

Great idea except the language we take for granted as traders doesn't translate very clearly to a non-trading programmer. A true test of the robustness of ones system is if can pass the einstein test. Can a kid understand it?

[gooni 10]

Throwing some weight behing NeoTicker, which is miles ahead of most of the retail competition. I think OpenQuant also has a powerful product.

 

The problem is that without a decent backtesting platform, you will have no idea whether or not your system will actually become profitable in the end. For me, NeoTicker is the only product I saw that has the flexibility to code anything that could possibly come up. Languages of choice would be Delphi or C#.

 

Send me a PM if you want more info, contacts for programming (skilled traders and programmers, not just programmers!) or more info re: features or anything. Note, I don't have any interest in NeoTicker other than the fact that I use it and would like to see more people reaping the benefits.

[darthtrader 10]

gooni, I wonder if you could give a run down on how you use neoticker as far as what you do with it that you wouldn't be able to do with other software?

While it seems like there is alot of amazing stuff for it(if Ninja can't copy the custom TICK16 stuff in 7 i'm going to look into Neo just for that) the whole program feels so clunky and spaghetti like to me. My biggest issue from trying the demo is that without a massive rewrite that they might eventually get dusted by Ninja or Openquant if Openquant ever decides to hire some writers to actually document the damn program or Ninja gets a database up to snuff with openquant/neo.

[gooni 10]

Valid concerns darthtrader. It took me a few months to get 'efficient' with NeoTicker, and now I can test things faster than any other platform. But here's where I find the most benefit:

 

- The option of true time-driven charting (as opposed to bar-driven), allowing combinations of many different kinds of series on a single chart. Imagine taking signals off a 15-minute chart and managing the trade using a 100-tick chart?

 

- Creating managed time- or bar- series to support decisions, easily plottable. In fact, and this might be the same as NinjaTrader, a system is just a special form of indicator. Nearly everything is series and can be used by anything to else to make on the fly decisions.

 

- Superposition plugin, allowing us to create custom series. I haven't used this as much yet, but have some ideas regarding volatility based bar series.

 

- NeoBreadth indicators (like TICK16, but you can do practically anything really). Anyone who follows breadth in general should look at this alone as a reason to come to NeoTicker.

 

- Tick Replay, allowing tick-based indicators to be used in system development and while looking at historical data. This is usually things to do with bid/ask analysis that require every tick to be replayed.

 

- Simulation Server, that can use the disk cache to replay the whole platform (not just a chart). The Simulation Server requires a restart of NeoTicker and the platform pretty much treats it as just another data server (like eSignal, IQFeed etc). I'm sure you can imagine how useful that is. You can even replay trading systems at say 5x speed and get realistic fills from the trade simulator.

 

- An OLE layer that allows automation of the entire platform.

 

- Grid Optimizer product for optimisation (haven't used yet, heard very good things).

 

- Being outside the US and trading many markets, NeoTicker actually handles timezones properly.

 

- I think it's probably more efficient than NinjaTrader if you want to have 50 charts open, I never got very far with NinjaTrader.

 

- A proper pattern scan architecture (but I don't use it, no stocks for me!).

 

Of course, there are some downsides. It is, at the moment, not great for Forex. They are having some stability issues with some of the order managers in the latest build, and the support forum can take over a day to get back to you. Certainly, the 'community' is smaller than NinjaTrader. But NeoTicker do make sure to tell you that their institutional customers far outweigh retail. Take that how you will.

 

One final caveat - I'd be prepared to switch in an instance if something better came along. Rewriting strategies is not a mammoth task, and I'd rather retain the flexibility of being able to choose at all times. Hence, anything major I do in NeoTicker (ie, beyond prototyping), I do in C#.

[darthtrader 10]

great rundown, the create custom series stuff sounds really amazing.

In another thread i've mentioned i've been looking for a tick database solution but it really seems stupid to not go with this or open quant.

I guess I'm really going to have to just take the plunge and buy this program. The biggest selling point to me is I've never once read about anyone who has taken the plunge and regretted it. Its always the same story like this...

I know deep down part of my problem with it is that its not using .net's beautifull anti aliased charts like ninja but even there I would imagine there could be a work around. Not to mention thats an extremely irrational reason to not use something.

 

thanks a bunch

[gooni 10]

Eh, I prefer how the NT charts look to be honest (now that I've got the colours all the way I like them). All that said, there's a 30 day trial and a monthly lease option. Just be aware that 30 days is the minimum time you will need to 'get' NeoTicker, and that's using it often.