well, with 99% of profitable weeks and DDs that low you could easily increase size each week. after 6-9 months your trading size would explode to over 1,000 contracts per system.
That's a fairly bold statement. My experience has actually been quite different. I've been trading an automated intraday system live for about 2 years now that generates 1000 trades/year with a positive expectation both live and simed. I reconcile my trades every day and my live executions match my sim executions 99% of the time. Edited to add: But I agree, Frosty's results look a little too perfect. His expectation of 4 ticks is slim enough that any slippage beyond his simulated slippage will degrade his results quite a bit (one tick additional slippage will reduce his expectation by 25%). But what he's trying to do is certainly not impossible.
For ES it is about 4 ticks net after the 1 tick slippage and commission. For CL as you can see it has a much higher average trade. All entries/exits are market orders. This ensures that there will be no missed trades. I have learned my lesson with fills on limit orders in the past. I give up more slippage, but I can be sure the trades will happen... The NN's are independent strategies, however there is a sort of "consensus" used among them.
I have played around with everything from 7 ticks to 25 ticks. This is one area of my strategy I am often tweaking. Likely the 15 tick variant will be used when I go live tomorrow.
All orders are market orders and 1 tick slippage is assumed on ES for each trade. Its possible I could be underestimating slippage slightly... That is one thing I can not be sure about until it goes live. I figured there would be some doubts once the equity curve was shown. I have sat here thinking "too good to be true" for a while now myself. However, after checking and rechecking it seems solid. Good news is because its profitable 99 out of 100 weeks, it will only take a few weeks to determine if this strategy is meeting expectations. I can greatly increase average per trade if needed, however I have to filter out a lot of trades. So if I determine I underestimated slippage I can easily re-adjust the strategy.
Having traded ES quite a bit in the past, I feel somewhat confident the 1 tick slippage is sufficient. The market is generally tight enough and its not like I am trading huge size. If anything 1 tick more may be required, which still leaves the strategy insanely profitable... For CL and 6E however I am new to those markets. What are the spreads generally like? How many ticks are needed to realistically represent the likely slippage?
In going live I would recommend either the 6E or CL over the ES anyday. Depending on the number of contracts you are trading, either of those should yield very small slippage.
Frost engine, How are you building NNs in Ninjatrader? Are you using a third party application then plugging them in to Ninja via the C# interface or are you doing something else? Runningbear