Thanks for the reply Sci. Yes with sharing code I have the same concerns. I was really referring to code that was already publicly available. With the systems I’m developing the code belongs to me as I’m the author. I am however more than happy to share the principles on which the system is based.
I can remember VV trying to get me to purchase additional add-ons, but I can’t remember an Auto-Timer. I think that’s something new. Yes it had a backtester, but I’m not sure about the portfolio and genius functions.
Yes Amibroker has a Montecarlo testing function, but in its basic form it provides “trade shuffling” functionality only, i.e. it won’t randomly select a subset of the trades from the trade list in its analysis. The Amibroker argument in favour of this is that the System Developer would already have used trade ranking functions to select the best trades in the first place. There are ways around this by introducing randomness into the ranking function code prior to doing the MonteCarlo analysis. One smart thing that you can do is to introduce randomness into the trade buy and sell prices, i.e. simulate slippage, via coding and then use the Amibroker optimiser to generate the trade list and then run MonteCarlo analysis over that trade list. All very flexible and powerful stuff but I haven’t used it as yet.
The autotimer and portfolio functions sound really worthwhile. The way my system will work is that I’ll run an “Exploration” each night that will return the next day’s trades and all the data required to set up the orders. It’s just a matter of doing a copy and paste of that data into an already formatted Excel spreadsheet which then calculates all of the relevant trade data. The part of the process that will take most of the work is then placing all of the conditional orders into Iress. When I paper traded a system a few months back I was getting up to 30 trades to open each night. This would have been unworkable for me so I stopped and started to reassess the system with the view to reducing the number of open trades at any point in time to between 10 and 15. I’ve been working on the trade ranking indicator(s) since then. I’m now in a position to hopefully be able to backtest the system again with these trade ranking functions implemented.
I agree with you re needing a system with and edge and then just repeat. This is definitely the golden rule. Forget about the magic indicator that gets a correct answer every time, it just doesn’t exist. The one thing that really does scare me about going live is what happened to you regarding experiencing that maximum drawdown almost immediately. I keep on asking myself what I would do if that happened to me. The answer is that I would immediately start to question the value of the system. But as you point out if you do all of the professional system development and testing work prior to going live you know what to expect and you can make a rational decision regarding the health of the system. The fact that you experienced this drawdown so quickly is just bad luck. Thanks for sharing your initial trading results Sci. I look forward to learning out more about your system development approach in the future.
(20min delay)
