Frost, why don't you program your own little class keeping track of orders for a given strategy (could be a few dictionaries with orderIds/positions etc.). That way you at least know for sure that when you didn't place any orders, your position is flat.
Hmm. Are you saying that the phantom orders are actually being placed by your strategy in response to a corrupt state? If that's the case, why can't you prevent the orders from being placed in the first place, rather than trying to close them after the fact? Are you concerned about the time cost of double checking your open orders before entering/exiting a trade? Sorry if this betrays my ignorance of how NT works, I have no real experience with the platform.
Hey Frost, question for you, did you ever get the backtesting versus real money trades straightened out? What's the cause for the descrepancy? Slippage?
Yes the system has still been running. Its been up and down. The verdict is still out on whether or not its a profitable system over time.It is certainly close. I am currently working on improvements to several of the strategies which I plan to put live in about a week. The phantom trades are mostly taken care of,but there still exists a condition where my strategy does not auto exit them. Need to sit down and figure out what that is. That particular glitch cost me over 100 pips in a single day a few weeks ago. Attached is the current PIPS performance chart. Since going live with this system it has performed 933 trades.
7 of the 10 strategies in use have been modified for this week. They were updated with newer strategies spit out by my strategy mining efforts. Overall the per trade expectancy increased 10 to 20% on most strategies. My goal is to increase per trade expectancy another 10% by end of year.
After 933 trades I have netted 452 PIPS. At my current leverage, my money doubles every 3,250 pips. Which means my return has been 14% over the last ~130 days. My max drawdown has been 461 pips, which is also roughly 14% Not bad, but like I said,the verdict is still out...
I am currently trading 10 strategies. If I eliminate the 2 worse performing strategies, this is what my equity curve looks like. Something to strongly consider.
Are you still employing that technique where you were basically applying a neural network to some random grab bag of technical indicators until you had a decent set of out-of-sample results?