Hi Mark, we do have a user friendly and powerful back-testing platform that shall match your requirements and help you building your project.

http://systemathics.com/backtest.html

I suggest that we discuss your project in a bit more details over the phone or skype.

Good morning Mark,

I am sure you know all Oanda, Bloomberg, Saxo and many more .. but mail to Gregor and ask what you need, he is an expert on this field - Gregor Koželj, e-mail: gregor@bulldy.com

or skype: bulldozer_hq

I hope I could help -

regards from Switzerland

Rudolf

rudolf.strobl@fx-projectmanagement.com

The best platform for backtesting is excel. If you are well versed with formulas and macros, you can do it yourself. If you'd like to leverage us (Alphamatters let us know)

MultiCharts platform is the best choice for backtesting single strategies or portfolio trading. It is available in EL, VB.net and C# languages.

I am very happy with it.

You can use platforms like Amiborker, Ninja Trader, R or Matlab for you backtesting

Hi,

I have some model done for NinjaTtrader, where the most of the "normal" ones (at least those based to any standard indicators/timing/break) can be tested quite easily and cost effective and optimise some values at same time without additional programming although that's not a problem either.

Please, contact me if interested.

Hi Mark,

eSignal is another good platform with an extensive back testing feature with detailed analysis and graphs. Their scripting language is based on JavaScript. They have a delayed data version that starts at $45/mo.

Have you tried portfolio123.com?

If you're comfortable with Python you could try out Quantopian (quantopian.com), or the open source engine they're developing to power it called Zipline (zipline.io).

If you have Bloomnerg, there is an APP that can help you. It is called Strategy Explorer and it uses Bloomberg market data for your backtest, also handles survivorship bias

what's best for nasdaq and new york equity TICK data? Thanks

Hi Yechiel, try Interactive Data (IDC)

Thank you

QuantConnect has free back testing using 10 years of tick data, if you are comfortable coding in C#.

Thank you very much

hi mark we can help you to backtest and optimise your strategy

The problem, for options at least is that, there is no trades while the underlying moves.

You have a bid and ask spread, thats stays fixed while the underlying moves, they mean nothing. The market maker loads up, and then lets go with the whip. But with historical data, you work with time and sales. So you work with "after the fact data". You actually are doing a spurious regression, and are predicting the past. As is often the case with most models. For example you have a last sale at say 100.25 and a bid at 100 and ask at 100.50. Then, the next sale is at 101.30. Nothing to do with anything. Unless you, do a bootstrap recall. That is why you get the determinant of the information matrix reaching zero. What you need to do is to relate actual trades with a time lapse related to the heart beat, the load and dump of the players who make their living on market mechanics and not on forecasts. The you get really strong determinants, and you can get reliable matrix inverses and valuable solutions to your differential equation parameters. Market makers often use bid or ask available quantities as a lure to get the volume on their side.

Should you wish to discuss this further, please contact me on stefhardy.com