If the strategy is relatively simple in that it depends on data that can be derived directly from tick and book data, something like NinjaTrader is a good choice because the strategies are written in C#. There are also many platforms that use proprietary languages that will work but will result in a much less portable solution.

Finally, there is the option of writing a platform from scratch if they have the time, money and fortitude. I can tell you for personal experience, going that route takes a lot of all three.

Feel free to send me a pm if you need more information.

Fix api are the fastest but harder to manage. Most of the strategy does not require speed, so metatrader, strategy runner, interactive brokers or others could be considered.

The answer is not simple. It depends on many factors. One must take into account the type of implemented strategy (eg market, frequency). Is the strategy relatively simple or not uses some statistical methods, machine learning etc, or just technical indicators . What assets are used - whether it is just futures or e.g. options. Usually, retail platforms do not allow trading options.

What brokers use client (do they have own APIs or FIX ). For simple applications, retail platforms like Ninja Trader, Tradestation, AmiBroker etc. are often quite sufficient.

However, we also need to consider factors such as support, which on retail platforms is often too weak for a professional user.You also need to consider whether the customer has their own IT resources, developers,infrastructure.

Well, the big part is the budget. Retail platforms are relatively inexpensive, professional platforms aren't.

The answer is difficult and requires detailed analysis of customer needs.

One more thing. Choosing a platform is a very important decision. The later change of platform you are using is very expensive.

I think there are a number of retail platforms suitable for researching new trading ideas.

However trading a strategy, or better a portfolio of strategies professionally is best done with own software or software developed in cooperation with a 3th party that has experience in this field.

The reason is that you need the flexibility and control over every aspect of the trading process which can only be obtained by developing the automated trading infrastructure (hardware+software) yourself or with specialized partners.

i.e. What if suddenly the connection to your broker back-end stops working after they upgraded their API? If this issue isn’t addressed properly by the retail platform provider you might have to wait for weeks until it’s fixed.

Also imagine the situation where for whatever reason (connection failures, data stream failures, etc.) you need to restart your portfolio of trading strategies. How do you make sure that the strategies pick up where they left when the problem occurred? In other words do the strategies keep their internal state or can they be brought back into the correct state before continuing trading? Can you correct the internal state by configuration in case you had to manually intervene (for example when your system missed a buy/sell signal because of downtime).

In many trading platforms you are forced to close the open positions and start from scratch in case of restarts. This is unwanted of course.

Personally I also like to be flexible on the broker’s back-end as well.

In our trading solutions this is implemented by having an abstraction layer between the strategy logic and the actual broker implementation for price data and order execution so that different brokers can be supported easily by configuration. The API of the broker or FIX protocol is totally hidden by use of such an abstraction layer. But to be flexible you must be willing to support multiple broker API’s as well as FIX protocol implementation and it’s nuances.

As indicated by Scott and Andrzej this mostly depends on budget. Retail platforms are very affordable to make a start, but having control over every aspect of the trading process can be a big money safer in the long run.

Thank you all for your input.. as I was afraid, such a decision is not so black and white. I think it is important to be able to clear multiple FCMs - for the reasons you all noted. Thanks again - I may reach out for more direct questions.

Hi Scott. Are you an active NinjaTrader user? I'm considering some platform options, of which NT is one, and have some questions on your experience with it.

Try our http://stocksharp.com Free for private trading. Have visual designer for non professional programming. And Hydra - automatic tool for downloading market data. Available with full source codes.

What are your thoughts on Collective2: https://trade.collective2.com/

TT is a great option... also check out their ADL which holds logic server side... good for someone on a budget who wants to focus on logic and doesn't want to worry with tech side... of course it all depends on latency sensitivity... but reality most are not equipped and or capitalized to compete in the latency game...

I like the topic

Hi, you can checkout NanoTrader. It is a German built trading software for advanced automated trading on futures. https://www.whselfinvest.com/en/trading_platform_cfd_forex_futures_06_automated_trading_backtesting.php . It is based on "sentiments". An indicator can generate a sentiment, that opens, holds or closes one or multiple positions. It is also possible to create your own indicator / functions etc.

I am using CloudQuant to backtest algorithms these days. I introduced this to a bunch of people at the Chicago Python users group and gave them a challenge to improve an old algo that hadn't produced alpha in years. The source code to the algo is in the public scripts directory at CloudQuant. One data scientists was able to modify the algo to generate a 4.25+ sharpe ratio.

One of the interns, a masters student in Financial Mathematics, wrote up her experience improving the algo at https://info.cloudquant.com/2017/08/improving-td-bear-flip/

The interesting thing was she used the social sentiment data to prevent losing trades.

Hands down, drop the mic - https://www.iqbroker.com/

Perhaps others have experience with Ninja Trader? I'm considering it for swing trading. Any thoughts on this?

So, what is it that distinguishes a "retail" trading platform from a professional "platform?"

Great question, Brad... it seems the flexibility the software allows? Also pricing! Would be interested to hear a more detailed explanation from someone with experience using both.

It depends on your/their coding level and experience. In my opinion www.quantopian.com is very good.

Hi, I like NinjaTrader as you can build basically any strategy on top of that and have all the order management functions ready (no need to write and tested well) and use basically all the standard indicators immediately and add and use your own indicators if needed. Brad, perfect for swing trading.

Have you heard about Blue Trading Systems? Kevin Darby might be able to provide some insight.

Hi Dominic , get in touch if your client needs advice or the system coded

The differences between a retail platform and a pro one will be : Speed of calculation and execution, management of arrays, ability to connect many feeds, almost no bugs, hotline efficiency. Try www.multicharts.com 20 years of experience... 11th version.... with many releases in between.. avoid MT4 wich is purely for retails...

Dominic,

the right system to implement next depends on what is currently in use, and the AUM.

If they have $5m AUM and their model is all in excel, then there are some simple inexpensive solutions to implement until they grow to the next order of magnitude. it really all depends on the budget and what limiting factors are keeping them up at night.

happy to discuss privately with you

Regards

Brian

I will mention some code I have been working on for a while: https://github.com/rburkholder/trade-frame -- I started it on Windows, migrated to Linux, and with a few tweaks, should run on Windows again as well. It is written in C++. Primary data stream is DTNIQ/IQFeed. Interactive Brokers is used for executions. The code is modular, and has a defined interface for writing additional providers. A simulation provider is built in. The code is designed to work with equities/futures/options/futures-options. Quotes, Trades, Greeks, LevelII data elements can be persisted to disk and evaluated via the simulator. The greeks are calculated locally based upon the iQFeed Libor feed for interest rates. US treasury rates can be used alternatively. Code can run with or without a gui. If you have some questions, let me know. The project is effectively a series of C++ libraries fronted by a bunch of test projects for experimenting with trading scenarios. Just needs an algorithm.

Henk Vedder

Here's a link to a blog post with a short description of our API beta program. Give us a call for more info. We'll be sharing more over the coming weeks.

http://www.bluetradesys.com/news/82-new-edge-api-beta

https://youtu.be/Eax1AtEwLhk

https://youtu.be/sHddjSfyfZ0

Tindeco VISION might be the right solution depending on the type of client. We serve professional clients ranging from family offices to institutional investment management houses.

My only comment is that you should provide specifications and functional requirements before making a decision -otherwise you will be bombarded with lots of options and could well purchase a system that is unsuitable. Many people have great difficulty describing what they actually want; and end up buying a Porsche when they could have bought a VW.

I am available to facilitate this process if required.

It all depends on the underlying strategy and liquidity and frequency of trading. We can implement any code on almost any platform. Costing has to be done based on Api, strategy algo etc.

We've used Ninja and Multicharts or own Api. That is not the primary issue. Integration with the price and liquidity source is key.

A client with a futures trading strategy that is not yet automated. The question then would be if the manual strategy can be automated at all. The human instinct of the trader could be involved in the current strategy, and that may be hard to catch into rules. How clear are the rules of the strategy?

What style strategy you looking for. Does it have to be bought and sold in mill sec. then you need co-location. If you have a strat that is more of I am buying here and will sell at a given point then. Most platforms will work. Ninja is a good platform, had a few on there. I have one linnsoft. But functionality isn't is good. I have used most platforms. Esignal, motivewave, Thur the years all will generally work good if you get out of mainstream. I have know idea. If you need specifics glad to help good luck on the search.

If you are trading forex then MT4 is the platform I use for automated trading. It is a free system and not difficult for a programmer to code. Sample code is available online or can be purchase cheaply from software suppliers.

http://www.financial-hacker.com/hackers-tools-zorro-and-r/

Of course I'm biased...

Multicharts

Sterling

Have you looked into Trading Technologies automated trading solutions, including ADL (Algo Design Lab)? Easily build your own algos, trade anywhere, and take advantage of their co-located servers.

https://www.tradingtechnologies.com/solutions/algo-trading/

if it's a portfolio strategy and you don't need intraday gsphere.net is powerful, flexible, efficacious and would not require programming skills

Bloodhound is supposed to allow creation of custom strategies with no programing required. https://www.sharkindicators.com/bloodhound/

But really, the question posted is too vague to be answered, in its current form presented, because a "best platform" will be dependent upon the specific requirements of the strategy. Perhaps Dominic can be a little more specific about the requirements of the platform if he wants credible answers.

Happy to throw my hat in the ring here; I've recently made the transition to MultiCharts and can vouch for it. Great for trading a portfolio of strategies, back testing is a breeze to use.

If you are not a programmer (such as myself), there are plenty of folk that can fill in the gap for you... So long as you have the logic/rules properly documented.

We've built our own platform

I have very negative experience with multicharts. I have a lot of experience with programming of strategies with AmiBroker, TradeStation, NinjaTrader. 3 years ago I decided to try Multicharts, C# version of it. I decided to order them to program very simple strategy, which I implemented already in AmiBroker and NinjaTrader. They told me that I have to buy Multicharts first, then to fill the application. The estimated cost was about $1000. I was told that it will take them about a week. After a week they finished the project, but it did not work as expected. Day after day I had almost the same problem. They tried to fix the problem on my computer remotely, next day I had similar problem again. After 2 monthes they told me that they can not implement the project in current version of Multicharts. I asked them to return my money, they refused. I stopped to work with them and I am not using this product any more.

Just out of interest, what was that strategy? We would charge $1000 maybe for a relatively complex options strategy. And what was the problem that they could not fix?

www.tradingblox.com for end of day strategies using back adjusted continuous contract data through CQG. Shorter term strategies, probably Ninja or TradeStation.

RightEdge (rightedgesystems.com), simple and very customizable but require c# proficiency

There are many platforms that have been successful. Here is the one I have been using for the last 6 months. http://cpptraders.com

Can you elaborate on the strategy and software in place now.

At the risk of sounding glib, I've found that the best system is the one you understand and can manage the risk on. I've seen very simple systems operate fantastically with the right people at the helm, and very complex ones fail utterly because no one knew how to run / diagnose / troubleshoot / operate them through a crisis.

at that moment I'm looking at the protrader platform, but I still do not have enough data to affirm the quality of it. It is programmable in C #

I can tell you one think, any of those commercial programs you might use, have one thing in common: share your intellectual property with platform developers. This can only work for somebody who want to do RD or rapid prototyping. if you think you can get industrial grade platform to trade real money where such operation will last longer that 6 months (assuming it is profitable) you are out of your mind. furthermore if such platform uses .Net or Java it is not worth risking a penny.

Why people are so irrational: most investors have no issue with loosing 10% of portfolio, but I have not found one wise enough to spend in advance 5% of portfolio to avoid such loses.

Guys I have been there and struggled for decade, if you are beginner AI developer: one free advise: unless you you can invest $1-2M i RD and spend 5-7 years on development, better forget about whole thing. you go to war (conflict of interest) with most sophisticated well funded and connected crowd. Good Luck !

Good point Martin

@Martin - why do you believe .Net and Java are a waste of time? Individuals can compete with those better funded if they have an edge of some sort and unless you are talking true hft, either C# or Java will perform just fine. I can go tick to trade in 40 µs in Java on a decent linux box. While that same logic could likely be done in 8 - 10 µs in C++; truly, who cares, I am not trading at the microsecond level. I don't have to beat the top firms, I just have to win more than I lose.

Other than that, I do agree it takes vastly more effort to create your own platform than most people realize. Had I known I likely wouldn't have started and by the time it became clear, I was better off finishing what I had started.

Well... time/ performance is much less important then reliability and security, if you consider you computer a pyramid, the C / assembly written operating system drivers, kernel core etc.. then c++ MFC layer is level two, then virtualization app level 3 nested Linux os kernel level 4 then, Linux foundation lev.5 then java environment is top level, now imagine one of this components have exploit, update, error possibilities of failure grows exponentially with the level as higher level components rely on lower level.. hence such pyramid is to reminiscent of house of cards, not to mention you don't own/ control other elements that may use it and jam it (or exploit it). Successful trading hinges on ability of unexpected trade that benefit you, and not to share in public domain how you allocate asset and let market participants to follow or counter trade.

PM me please.

@Martin - possibly you are missing one key piece of information. In a co-located scenario on extremely stable Linux releases, at least in the case of java, none of what you describe is applicable. Actually the lower the level the code, the harder it is to maintain and therefore the more easily it can be exploited (i.e. buffer overflows are an issue with assembly language but not so much in higher level languages).

C/C++ is great in theory but unless you are at the level of worrying about cache misses in your quest for speed, java will get you where you need to go. As to the garbage collection argument, well designed java code can virtually eliminate gc.

Lastly, as I have said many times, I could tell someone in intricate detail what I do and by writing tens of thousands of lines of code over several years and having that code battle hardened by actually trading with it, they too could execute my exact trades.

Hello my Teri Workman

I am new to this group.

Like many of you I to work in Trades

We monitize SBLC,BG,MTN.

And than get them in to Trades.

Also,Cash trades as well as Forex .

I am a product specialist at CQG.

Depending on the complexity of the program, the trader can be up and running in hours on CQG. We will help you to program your system a no additional cost. Most trader programs are not difficult to translate into CQG language. Jim Stavros 312-939-1574

AUTOTRADETECH.com

Dominic,

you might like to have a chat with Ilya Holeu at Spotware (they're the company that created the cTrader platform - which includes cAlgo as the c# based development platform for quants).

The reason I suggest them is that firstly the product is great but perhaps, more importantly, you can take advantage of their infrastructure and execution (speed/reliability being of the essence for most auto trading systems) - they're also a very cost-effective solution. Hope this helps. Disclaimer: I do not work for Spotware.

Try Python base robot framework

I use automated trading and it works fine for me though.

Its simple and very consistent ..Its easy to set-up and off course not for sale

But everyone can use it....

Depends on their programming ability.

If it's Futures you want to use Multicharts.

Use saltstack.org for automated ma agent and big data platform to trigger events for automated responses.

Would recommend to consult: https://www.mql5.com/en/job?utm_source=mt5terminal&utm_medium=special&utm_campaign=pt.mail.freelance