Any good structure for a simple self-made backtesting platform?

I cannot apply my strategy in retail level or open source trading systems, the logic is totally different from the straight forward apply one indicator or few indicators then backest it with price/volume.

Quote from comintel:

Agreed.

Just as one simple example, no available retail or professional platform I have found handles systematic options strategies back-testing. This makes them all useless to me.

They also tend not to cover less-common futures etc.

And all of them, including the open source ones, are now excluding from their open source most of their broker connectors, which they are now licensing and charging for. (The IB connector might typically be free but they charge for most everything else).

So if you use them, you are forever dependent on unknown code developed by somebody else that might stop working any day and you can do nothing about it. Their connectors tend to also be out of date and incomplete and require old levels of the broker API's to limp along.
More...
 
Quote from j2ee:

Your reason may be good only if you do backtesting but not live trading because there is no official live trading support with matlab.

As I knew matlab language is not that easy too, may be it is easier to build a new backtesting system with C#/Java then learning matlab codes. Well matlab codes mean something in job industry but not a lot.
More...

Interactive Brokers have an ActiveX API for live trading (I've used it for intra-day trading for a year now and I'm 100% satisfied).

If you insist on C/C++/C#/..., my recommendation would be to do prototyping in Matlab and then port to any other language - in this way you'll save yourself a lot of time. Learning Matlab is way easier than e.g. C++. You don't need the whole OOP thingy with simple trading scripts anyway (and Matlab has that too just in case).
 
Quote from QuantWizard:

Interactive Brokers have an ActiveX API for live trading (I've used it for intra-day trading for a year now and I'm 100% satisfied).

If you insist on C/C++/C#/..., my recommendation would be to do prototyping in Matlab and then port to any other language - in this way you'll save yourself a lot of time. Learning Matlab is way easier than e.g. C++. You don't need the whole OOP thingy with simple trading scripts anyway (and Matlab has that too just in case).
More...

So you are saying after someone spends a lot of money buying matlab to just do backtesting, then he needs to build his own live auto trading system with IB API from scratch. If someone knows how to build a live system with IB API, of course he knows how to create a backtest system as well, then no point to buy matlab or whatever.

If someone buys a retail system, of course he wants the auto live trade part too, not just a backtesting system like matlab.
 
Quote from j2ee:

So you are saying after someone spends a lot of money buying matlab to just do backtesting, then he needs to build his own live auto trading system with IB API from scratch. If someone knows how to build a live system with IB API, of course he knows how to create a backtest system as well, then no point to buy matlab or whatever.

If someone buys a retail system, of course he wants the auto live trade part too, not just a backtesting system like matlab.
More...

You don't need to build from scratch - there's plenty of source code available out there. Anyway, I'm not a salesman so if you're not yet convinced I'm not gonna argue - I'm just saying what works for me and it's your call what to do next.
 
Quote from j2ee:

So you are saying after someone spends a lot of money buying matlab to just do backtesting, then he needs to build his own live auto trading system with IB API from scratch. If someone knows how to build a live system with IB API, of course he knows how to create a backtest system as well, then no point to buy matlab or whatever.

If someone buys a retail system, of course he wants the auto live trade part too, not just a backtesting system like matlab.
More...

I suggest R instead of Matlab for those looking at Matlab for whom cost is an issue.

The two are equally (and widely) popular for financial modelling.
 
Quote from comintel:

I suggest R instead of Matlab for those looking at Matlab for whom cost is an issue.

The two are equally (and widely) popular for financial modelling.
More...

There are free Matlab alternatives as well, e.g. Octact - not sure about compatibility though.
 
Quote from QuantWizard:

You don't need to build from scratch - there's plenty of source code available out there. Anyway, I'm not a salesman so if you're not yet convinced I'm not gonna argue - I'm just saying what works for me and it's your call what to do next.
More...

If someone can understand source code and work with it with programming skill, why would he choose matlab but not build his own system?
 
Quote from j2ee:

If someone can understand source code and work with it with programming skill, why would he choose matlab but not build his own system?
More...

How about saving time and money perhaps?
 
You must log in or register to reply here.
Back
Top