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

[QuantWizard]

Quote from j2ee:

Look like s very curve fitting graph. Anyone can create super profit backtest result with luck or curve fitting. Show us some stats showing your backtest has good chance happen
in real future.

No curve fitting is taking place.

 

[j2ee]

Quote from QuantWizard:

No curve fitting is taking place.

Yes or no doesn't depend on you just say no.

 

[Gonz]

Quote from QuantWizard:

Not sure whether I fall into this category, but if my back testing is half as good in reality (see 2008-2009 performance attached), and if my system continues equally well as in the past, I'd me more than happy. The only thing I know though, is that I won't fall victim to my own psychology.

Nice curve. Is it the results of backtesting + walkforward testing? If it is only backtesting results, then it is meaningless.

 

[Gonz]

For example, I have backtest APPL stock from 1984 to 2012 (as shown). The returns is 20,000% including commission & compounding.

Nevertheless, I don't think I will be able to get that kind of results if I continue to trade the market. Reason is because it is backtest optimized.

 

[j2ee]

Quote from Gonz:

Nice curve. Is it the results of backtesting + walkforward testing? If it is only backtesting results, then it is meaningless.

Exactly. Even just walk forward may not be enough, show us more stats.

 

[GloriaBrown]

Quote from dom993:

j2ee:

Your problem is you have no idea what you are doing. And how would you? You don't have any experience in automated trading. You are making this really impossible for yourself.

Step back. Start using an existing platform, develop a few profitable strategies, in the process you'll learn quite a bit about what services a platform should provide. Then start trading live - on a simulated account of course, I recommend using IB sim account for that. And you'll learn some more doing that. At some point, you'll have a complete understanding of what an automated trading platform must offer to be viable.

Cheers

You sound like a good man.

 

[pairsarbtooo]

Quote from flakac:

I decided to use Matlab for backtesting&trading system development. Yes, the Java (or any real programming language is better) but I had no experience to work in anything else then Matlab.
My system consist of several objects, they are connected to each other trough event -listener mechanism (=nightmare for debugging and source of instability) The first objects reads the data and the last objects creates the trading signals. (the objects between do data processing, machine learning, etc)
The connection to broker I haven't done yet..
Matlab pros:
- You can do what you want, the programming is easy and quick
- Plenty of ready made methods you can use (like machine learning methods in statistic toolbox) You do not need program this yourselv.
Matlab cons:
- Mechanisms common in other programming languages are missing
- Matlab calculation are vector based, which is good for simple ccalculations but if you want incremental steps (data arrive - data are processed) then you must handle it accordingly.
- No multithreading (or at least it is difficult)
- Big memery demands
- Very slow
And even the programming in Matlab is easier then in other languages (at least for me..) the time demand to develop something useful is huge huge....
If you want you need OWN back testing platform, think carefully why not to use something what is available already. Because development something from scratch is really not easy. If I know it beforehand, I never do it again ...
So if you really need your own backtesting, do the flowing:
- Hire a project manager
- Hire some programmers

you guys make this WAYY TOO COMPLEX. If your strategy has so many indicators that its too difficult to easily develop in C++, your system is not robust enough to trade profitably. Reduce the amount of your indicators, create a SIMPLE system, create a trading frequency and volatility you can personally manage, and then go trade. JEEZ...these posts are generally very far off the mark with what little needs to be done to test strategies.

The most successful trader in history developed their systems with OLD languages with punch card output. What makes you think just by having more complex tech you will surpass them as traders? This is such a common mythology ..I could go on and on and on...but few will listen.