Who offers true tick by tick backtesting

[investix]

MarketDelta will supposedly merge Trader and Charts, in addition to offer CQG integration, this quarter.

 

[lindq]

Quote from bundlemaker:



I know IRT and MarketDelta can do what I want (I've done it) but they offer no real time live execution solution that works properly.

Bob

My experience with IRT in live execution has been good (IB <> IRT), so long as you're trading just a few instruments.

Is it something you've tried, or are you just making an assumption?

 

[januson]

Quote from bundlemaker:

Perhaps, but it looks like it would be extremely difficult to code it. Other platforms (like Neoticker, MarketDelta, IRT) don't have to fool with multiple timeframes.

This is not difficult at all

but what do you mean by "don't have to fool with multiple timeframes" ?

 

[Eight]

i've coded multiple timeframes, it's a piece of cake. You hook any software up to tick data and you generate arrays of bars of any kind you like. It's trivial.. backtesting on that data is more certain than all the stuff the environments usually give you to "help" you because there is no ambiguity about what different sets of data are or how they are being accessed... I did that in TS2000 with EZLanguage and in Ninjatrader just because I couldn't wade through all the ambiguous documentation. The downside of building my own bars of data was that I had to build my own indicators to run on them, but whatever...

 

[cgroupman]

Just curious, do you guys ever get around to actually trading? Are you just 'back-testing' for fun and no profit? You do realize that many guys, much smarter than me, have explained that backtesting is just an exercise, not anything that will help you make money, right?

Don't mean to be rude, but seriously, time wasted not trading in the real world, hoping to find some magic in back-testing, is, well, you know.


c

 

[cgroupman]

Test your stuff in real time, with real money, even real small money. Much better than fantasy back testing. Test the automation, sure, but backtesting theories is, again, just hoping you will find some parallel that may or may not work in the real world. You cannot backtest enough environmental variables to give you a reasonable result.


c

 

[tradelink]

backtesting isn't mutually exclusive with live. (iow do both)

if you understand science and math, backtesting is a no brainer.

 

[Bob111]

not sure about futures and stuff,but backetesting\simulation on stocks(specially on low time frames) can be useless,compared with real trading. bots do react on your order,when you are out there,while in sim you just playing with last price(which not necessarily will be your fill)

 

[tradelink]

There is backtesting theory and backtesting implementation.

If you understand the theory of backtesting and the theory of your strategy, you should be able to select an appropriate backtester implementation (and data set) which provides useful data to your particular strategy.

The frequency should not matter, backtesting is useful and is used in HFT.

 

[cgroupman]

Quote from Bob111:

not sure about futures and stuff,but backetesting\simulation on stocks(specially on low time frames) can be useless,compared with real trading. bots do react on your order,when you are out there,while in sim you just playing with last price(which not necessarily will be your fill)

Right, but not just the bots. The whole trading environment. You simply cannot replicate real data input well enough to really matter. Why did the market crash 6 S&P points on the close today? Why did gold drop $50 or so? Who knows for sure, and you can't replicate that type of inter-market reaction. For every minor movement in one instrument, there are several chain reactions.

If you're testing software to see if your idea can be put to task, sure, fine. If you think you can determine any sort of profitability, no way IMO. But, if you do it for real, and it works, then hurry up and make some money before your idea becomes ancient history. The markets are extremely efficient.


c