Accuracy for Testing Trades

[mickro 10]

I stick to an attitude of automate/test everything and use redunduncy, but this goes counter to paper trading systems when I feel I cannot trust their reliability.

 

I feel there has to be a solution for this. Even if you have to roll your own to look closely at book data I think it would be worth it.

 

Right now, I'm just using the Interactive Brokers API and paper trading account for testing new systems... It's good, but does anybody now of a really good system? It doesn't have to be a broker. I'm even welcome to ideas, b/c I'm tempted to roll my own if it'd be considerably accurate to a real trading system.

[henryduncan12 10]

There will be some tools or API as well to test new system. Just try in google or in this forum as well. I think, you will get something useful.

[BlueHorseshoe 10]

I stick to an attitude of automate/test everything and use redunduncy, but this goes counter to paper trading systems when I feel I cannot trust their reliability.

 

I feel there has to be a solution for this.

 

If the part after the comma is the problem, then surely the part before it is the solution?

 

You need to be looking for net profitability across literally thousands of trades before you can have any real confidence in a strategy. So unless you are trading in an extremely high frequency fashion, there will be no way to accomplish this with paper trading in any reasonable amount of time. Hence you should automate and backtest on historical data rather than paper trading.

 

Paper trading is useful in that it allows you to become familiar with the nuances of your strategy's order entry and execution foibles; it is not a sound way to validate a trading system.

 

Hope that helps.