Perhaps you need to separate the execution application platform from the strategy that has your IP. In that way you just need a platform in which to implement your strategies as separate code base. Find someone you trust or otherwise learn some programming? A difficult and ambitious task. To me this area is always more difficult without programming experience as you will always rely on someone else for accuracy, debugging, and on-going maintenance.

Do any of the off-the-shelf applications provide a platform to work from and implement your strategies? For many these are not adequate.

Does your system need to interact with a broker or live data?

(Sorry not able to access your examples to see what information you are relaying there)

Nigel, I have not seen anything off shelf that even comes close to doing what I need. I do want my programs to be able to update with live data for updates prior to execution. I think I did not have my links set to be shared. I've corrected the links below.

https://drive.google.com/open?id=0B-U8FWMbrFnxS1hlOWI4OU9EcVU

https://drive.google.com/open?id=0B-U8FWMbrFnxLV83LVdGR0lRUm8

https://drive.google.com/open?id=0B-U8FWMbrFnxLWhoYnZMQV82MFk

Keith - send me a pm and I can point you in a couple of directions that may help. I am a quantitative algo developer who is lucky enough to have the programming skills I need to get my algos into production.

Recently I had a need to use Excel models to test out the efficacy of an algorithm prior to committing the time and effort to programming it at a production level. The best solution for me was to actually program an Excel interface into my existing code base but along the way, I also found other solutions that may suit your needs.

I would post the links to the products I came across while researching this issue but but as a moderator, I have had to caution plenty of posters about posting links to commercial products so I don't want to be accused of not abiding by the group's rules in that regard.

Keith I think you need to invest in buying a copy of Multi-Charts because Excel is very close to the same logic Easy Language. You will easily adapt your own Excel algo's to either TradeStation and or preferably Multi-Charts.

I would never consider letting any programmer have your secret sauce as you put it. Maybe you should have said awesome sauce, lol. There are some programmers who are so out there (good and trade on their own) they could care less about stealing your stuff. I do know one he is reasonable and the most honest programmer I have ever met. It has been years since I talked to him but if interested let me know, otherwise Multi-Charts.

Keith - Nigel makes a very valid point there are some third party products that will allow you to:

* use Excel with real time data and

* route automated trading electronically.

Kinda cludgy but it works, I ran a system once this way not so bad if you use big data like hourly and up.

Remember, all strategies work in a 6 year bullish market. Best advice is, next time you decide for a development system, make sure that it is future proof.

Seems like Linkedin now limits the maximum post length, so let me continue.

1. Even in case you hire anyone, you will be able to read and understand the code — very important as it allows you to stay in full control of the project.

2. Last, but not the lease, this is all for free.

I'd suggest to look at Python(x,y) if you work under Windows, or have a look at Enthought distributions which come with a lot of ready made useful packages.

If you have questions please feel free to ask.

Most of the HF's I work with aren't using apps like Tradestation etc. These are predominately seen as Retail style platforms and for most large Asset Managers are not nearly robust enough. I see FIX for communication, SQL for data collection and housing, Excel as a mid-tier calculation engine, often coupled too a c/c++ add in for complex algo's with a separate Excel WB as UI (dictator). This allows for a lot of flexibility and dynamic changes without having to recreate the wheel or rely on a team of programmers to develop logic from scratch. It scales easily and is broker/liquidity provider neutral. The only tweaks from broker to broker is the FIX protocol. I can tell you from experience it makes sense to do it right from the very beginning, switching platforms while trying to maintain trading/returns, compliance etc. really sucks. Did it in the mid 2k's and it was a massive headache.

I think maybe mistaken Keith wants to trade and not develop software. Python maybe great, FIX is what powers our proprietary platform also and is far beyond novice abilities. Python maybe easy I don't know I will have to look at it but Keith may need help along the way not for complete code but for snippets etc. For that how can you beat the retail type platforms that have been out for so long? There are several not just the ones I mentioned. But easy language is most like Excel and has a huge amount of community support.

Excel Environment

Pricing runs on MathFinance’s Excel Add-In library with fully documented pricing models comparable to those used in many banks. Add-In functions are also directly accessible in excel with detailed help files.https://mathfinance.com/expertise/products.html

Guys, Thanks. You gave a lot to look at. fyi my systems are all end of day no need for low latency. A lot of it is pairs trading. it did not see in super charts or trade staition the functionality needed. i assume multi charts is simular.

Go for python or matlab. Worth the time and effort

Keith, neither TS, not MultiCharts, nor (as far as I know) Metastock and many other software do have the functionality you need because they all live in a bar-based world with the "one chart — one strategy" paradigm. This basically means that it's impossible to simultaneously (and conveniently) send signals from one source to various instruments. So, for research purposes I believe what I suggested is the very way to go. On the other hand, if you want to avoid the hassle of writing your own trading plugin (if you want to automate your trading in the future) then maybe using one of these ready-made platforms could be reasonable, but only at later stages. Feel free to pm me, I will try to help you out with all that.

Python is the way to go, along with its great libraires, numpy (numerical computing), scipy (statistics, optimization and more), pandas (highly efficient data manipulation and analysis) and matplotlib (ploting). I suggest the Anaconda distribution (https://www.continuum.io/downloads), which already come with all these. There are several introdutory, intermediate and advanced courses in Coursera, Udemy and Udacity, and some nice books, like Python for Informatics - Esploring Information, from Charles Severance, Python for Data Analysis (pandas), from Wes Mckeney, and last but not least, Python for Finance, from Yves Hilpisch.

I would go with TT. https://www.tradingtechnologies.com/platforms/x_trader-platform/

TT's Algo Design Lab, ADL, is an excellent tool. I worked there for years and the support they have is top notch. They will help you convert what you have into the system and train you on the software.

This would be the best way to go so you never have to worry about keeping up with anything but your trading.

ADL is not a bad way to go and TT are releasing a new version of their platform in the new year and have been busy upgrading their exchange interfaces too.

TS and MS are chart driven environments, but you can display multiple time-series in a single chart window and call up to 99 time-series in a single strategy. TS also allows you to embed strategies within strategies and if you need GlobalVariables you can pick up an add-in to do that.

NinjaTrader is not chart driven (i.e. you can trade from a grid of market-strategy combinations), but has a flaw it its CEP architecture (still) so if you want to do anything other than enter or exit at the next day's open you'll need intraday data to backtest (pathetic).

That aside, you can pull data out their API and send orders to it, so you can use it like middleware to access your data and your broker. If you are going that way though, TT is probably a better solution, just more expensive.

@John Putnam II has my vote, but I would throw Mathematica into the mix replacing C++ (can always be converted back to C using an external development resource), Mathematica and the Mathematica workbench are easy to master, can be connected to Excel and are enterprise grade (a lot of dealing rooms use it), the standalone licence is cheap and it is very scalable, this would allow to write complex algos with a minimum effort, plug them into Excel ... et voilà. Also would replace SQL with Postgres.

Excel has certainly serious limitations and not suited for serious trading systems. But any other programming language or tool also has its specific limits. So there is no perfect solution. I've programmed about 300 trading systems for clients so far and my advice is: Select your tools dependent on what you want NOT to do with them.

Your trading systems are not very complex, they are just a hodgepodge of simple indicators? Then you can use put-together systems such as ADL or even stick with Excel.

You don't need processing power for large training and optimizing runs? Then a rather slow language, such as R, Mathlab, Mathematica, or Python is ok.

Your work is more research oriented and backtests need not be realistic? Then you need no trading platform at all, the rudimentary backtest functions of R or Mathlab are sufficient.

You don't use statistical or machine learning libraries? Then you can use a platform with a proprietary language, such as MQL4 or Easylanguage.

For my own system development I'm using the C-based Zorro platform and R, but for client's systems I've used many platforms and languages so far. Some considerations about language and platform choice: http://www.financial-hacker.com/hackers-tools-zorro-and-r/

And if you really don't want to code the systems yourself, here some suggestions about hiring a freelancer:

http://www.financial-hacker.com/i-hired-a-contract-coder/

@Keith and group. I have a long love / hate relationship with the TS platform and more of the latter in the last few years.

TS has a Excel interface to EasyLanguage.

The platform does have the ability to send information and / or trade signals across hundreds of instruments simultaneously.

As well, with using this ability you can have back test history across all of the instruments.

When you hire a freelancer, what you must realize is that you will be getting regular copies of the source code also. The programmers job isn't to trade. But to put your ideas and logic into workable code. You can keep the true settings to yourself. But yes, the programmer will need enough info to backtest / forward test the strategy to be able to prove it works according to your expectations. You might be surprised to learn that there is no secret sauce....or rather that the recipe has been known for quite some time....like KFC's "secret recipe"

Philip, yes, you can use multiple data streams in one chart in both TS and MC, but you can't send multiple orders to different markets from one chart. So, basically you need to have two charts to trade a pair — quite resource consuming. Global variables are nice, but it's more like programming in 1970's style, possible, but phenomenally inconvenient.

@Alex, Yes it is all very messy, but the tools do offer so much that can take ages to program as well.

Although I'm loathe to, I'd probably recommend Keith use NinjaTrader, but with Python wherein you write your strategy in Python and interface to NinjaTrader via ZeroMQ or Redis. If you're using ZeroMQ you could use kdb+ 32bit as your database.

This way, you just program your Strategy in Python and still have the features of NT available. Also, you don't have to write broker or data interfaces, you get access all that through the NT API. All Keith would need is the middleware which I'm happy to provide for free. The current version is NT8 so still a prototype, but hiring a programmer to modify that wouldn't not be much.

That is, you have a strategy in NinjaTrader which unlike TS and MC can run orders to multiple instruments. The limitation is that you can't see orders for the non-primary instruments, but you can see them in the broker interface. You have yet to check that you can get these fills through the API, but should be able to.

The ZeroMQ interface populates the database so you can have proper fail-over. The Python program reads the data and fills from the database and sends new orders to it which ZeroMQ fires into NT.

Sadly, the managed orders in NT are also, let say flaky, so I'd recommend using unmanaged orders. Then he should be away.

@Philip - I have been testing NT for a few months, and I have wanted to do something similar to what you are outlining here. Did you mention you might have some glue code you could share?

@Naren et. al. Sure, just connect and shoot me a request and I can send you the code.

Checkout this video.

http://www.bazaaranalysis.com/AthenaInAction.aspx

If you like this platform then we can talk further

@Alex and Group, please understand that I do not wish to be rude. I use the abilities that I have described on a daily bases.

I can easily send information or trade signals from a signal chart to hundreds of charts. (resource consuming)

I can easily send trade signals from a signal chart to hundreds of instruments using Radar Screen for trading.

With using Radar Screen there are No charts involved (very little resources used)

The typical Global Variable used in TS has no ability to save history (only from the start of the first live bar)

Yes, I have a Global Variable that can save history. ( but it can not be used during optimization)

No, I do not use any Global Variables in my Total Strategy setup.

In the TS library there is a example available where it was possible

to trade pairs on a single chart coded by David Cohn using data1 and data2 setup.

“Algo – Pairs Trading”

I hope this is helpful

А у меня один алгоритм и очень простой. Но главное всегда в плюсе.

Find a partner... for your project and explain what would be such a partnership

I am sorry but 10% ROI per year is nowhere close to be enough for algo-trading. Minor unexpected drawdowns will kill all your return in no time at all.

I would suggest to look at strategies that can offer you something closer to 40-50% return to cover for all eventualities.

@Germann Arlington ; 10% is good enough if can be consistently done.... 40-50% returns consistently would make just about every trader a multi-millionaire in a short period of time

Depends on 10% of what. When you mean the profit to capital ratio, where the required capital is maximum drawdown plus maximum margin, 10% is indeed not enough. Acceptable systems start at about 100%.

Когда с начала торговли три года в минусе, а на четвертый 10% плюс, это уже неплохо. Необходимо двигаться дальше. На пятый год торговли надеюсь выйти % на 50, с просадкой менее 10% от депозита. Поиск торговых систем окончен...

@Сергей Олешкевич: if you are in negative territory for 3 years than you are doing it wrong.

Due to unpredictability of market movements you should always aim at 40%+ as the unexpected drawdown can take it back to 20% in no time at all.

I started with algo trading in January 2009 and turned on the fully automated system in November that year.

In that time I experienced 5-10% drops in the market (in matter of hours) and if your algo is not ready for such an event it can hurt an overall returns very badly.

P.S. My worst loss (during one of the drops described above) was 1.5% of the total investment, over that month the total overall GAIN was about 5%.

I don't think you need a further programming skill sets to be a good trader. Excel is supported by most of brokers.