backtesting platforms...

[cdcaveman]

more and more research..

Ninjatrader.. and other colorful related pieces of software don't seem to add up to what Amibroker is doing.. .

i am at the beginnings of my work with developing signals .. i can not build a entire system from scratch at this point.. i want to learn programming and become proficient.....

Amibroker seems to be the winner.. Ninjatrader seems like a great platform if your selling trading systems and indicators to the public .. i want something powerful .. and this guy who developed Amibroker seems legit.. everyone praises his response time etc.. ..

any thoughts.. still need a data provider..

active tick.. iqfeed... ... help??? tick data as far back as i can afford on a 50 to 150 a month budget.. i want to collect data from a good feed and keep it.. .. some of the platforms keep the data proprietary..... esignal i think does that.. anyway..

this is intraday pattern recognition at the most granular level.. i'm not a HFT... i need to find a good enough place.. i've been stuck in analysis paralysis with this for a while..

i could go into depths about what i want to do with the platform..
but its similar to what alot of people do...

what kind of thing scans real time an entire list of underlying for patterns?

can you download options data from a feed like IB in Amibroker?

 

[Tonkadad]

Have you become a member of the yahoo groups "amibroker".

http://finance.groups.yahoo.com/group/amibroker/

Very devoted crew of users. If you want to do it, more than likely someone has already done it.

Like anything, do as much leg work as you can, if you repeatedly ask questions that show no effort on your part, be ready for a smack on the nose.

 

[dom993]

Re. plateform ... AmiBroker, NinjaTrader, MultiChart, TradeLink, etc. ... they all have pros & cons. none of them is perfect.

You need to know what's important for you, in the short-term & in the long-run, and analyze those plateforms in this light.

I can say for myself, object-oriented programming was a strong requirement (speed of development, which includes testability & re-usability). I also wanted a plateform commercially supported, with the largest installed base possible so that most bugs would be fixed or at least known (I still found a few "brand new" ones).

At the end of the day, the same strategy will make the same backtesting profits when correctly coded on any plateform - and no plateform comes with a money-making strat, you will have to create that yourself. So this (the R&D part) is a key aspect to dig into. The other aspect to investigate is operations. Once you go live, how will the platform support your trading. Multiple accounts, multiple datafeed, multiple strategies, failure scenarios, remote monitoring, etc.

Do your analysis, then pick your poison, then deal with the issues & limitations that come with it.

And PLEASE buy 10+ years of quality historical data for all markets you intend to trade. The cost of that is ridiculously small, compared to the cost of *not* spending that money, using limited (in quality & scope) historical data and ending-up with systems that lose money in forward execution.

 

[cdcaveman]

thanks dom... i don't plan to go automated with a strategy... thats one big thing..

 

[dom993]

Quote from cdcaveman:

thanks dom... i don't plan to go automated with a strategy... thats one big thing..

I posted in another thread that, to me, backtesting has value only if it shares the same code-path as live execution. If you are not going automated, then IMO backtesting is a waste of time ...

- because you won't be able to trade live all the setups that would be found later in backtesting (murphy's law, you were in the bathroom or on the phone or still in bed or having lunch or ... when the setup occurred).

- because you won't trade live those setups you'll enter in the same way you would later backtest them.


Does the attached backtesting P&L curve look like something you would like to achieve in your trading ? It covers 6 years for 3266 trades. That's 1.5 trade average per day. But it would drive me crazy to spend all day (and night, that system is 24/7) just waiting for those 1.5 signals per day.


Anyway, good luck in your trading endeavor!

 

[gmst]

Quote from dom993:


Does the attached backtesting P&L curve look like something you would like to achieve in your trading ? It covers 6 years for 3266 trades. That's 1.5 trade average per day. But it would drive me crazy to spend all day (and night, that system is 24/7) just waiting for those 1.5 signals per day.

For the first time, I have seen your backtested results. Every other time you posted them on other threads, when I clicked on attachment it used to say some kind of "double header error".

Looks v.good. Who did you use as your data provider for last 10 yrs?

 

[cdcaveman]

Quote from gmst:

For the first time, I have seen your backtested results. Every other time you posted them on other threads, when I clicked on attachment it used to say some kind of "double header error".

Looks v.good. Who did you use as your data provider for last 10 yrs?

i've had those double header errors lately looking at pics here to lately...

dom... i get exactly what your saying....

 

[dom993]

Quote from gmst:

For the first time, I have seen your backtested results. Every other time you posted them on other threads, when I clicked on attachment it used to say some kind of "double header error".

Looks v.good. Who did you use as your data provider for last 10 yrs?

I have no idea what those errors can be. Most often I check my attachments after posting, never found any issue. Does this attachment give you an error?

Funny how a P&L curve can tell a story better than the corresponding performance report. P/F 1.38 isn't exactly great, neither is 51% win-rate - and I am sure most everyone will run away reading the max.DD at -24k (for 1 contract !) (at least those figures account for 1-tick slippage on entry, same on exit, and $5 comms / round-trip). But what makes the P&L is avg/trade * number of trades.

Re. historical data, I bought mine from TickData. They are simply the best (that I can afford ... I did inquire once directly to the CME, clearly their prices are not geared towards individual traders).

 

[gmst]

Quote from dom993:

I have no idea what those errors can be. Most often I check my attachments after posting, never found any issue. Does this attachment give you an error?

Funny how a P&L curve can tell a story better than the corresponding performance report. P/F 1.38 isn't exactly great, neither is 51% win-rate - and I am sure most everyone will run away reading the max.DD at -24k (for 1 contract !) (at least those figures account for 1-tick slippage on entry, same on exit, and $5 comms / round-trip). But what makes the P&L is avg/trade * number of trades.

Re. historical data, I bought mine from TickData. They are simply the best (that I can afford ... I did inquire once directly to the CME, clearly their prices are not geared towards individual traders).

Thanks dom.

Yes, I can see the attachment.

I think its a good decision that you are trading your strategy using full automation - due to it trading 24 hrs and because PF is a bit low, so margin of error is small. But great work overall on strategy development and developing all the automation required to run it.

Yes, tickdata is the best out there.

You planning to stick with Ninja for long-term or you think you will be doing some changes in Ninja as your progress in your trading and your requirements become more advanced?

 

[cdcaveman]

so theres no use in signal creation on a larger time frame? doesn't make sense.. ive heard tickdata so many times.. they must be good..