Favorite Futures: Futures - bonds, currencies, index
Posts: 288 since Oct 2010
Thanks: 70 given,
Well thanks for answering. You're the only one so far.
It sounds like because your testing was manual, you couldn't really cover enough market conditions to be confident. Can we assume that your backtested trades were very similar to what you could have accomplished live?
I know exactly what you mean about the drawdown dilemma. My systems have small drawdowns for that very reason, so that I'd be able to spot quickly if it wasn't performing as expected. The thing is though, you can identify what sort of market conditions you've tested, and after you've done a lot of testing, you develop an intuitive (ironically) feel for whether your system and tweaks are curve fitting. So, if you've covered all general market conditions, eliminated curive fitting and have mentally OK drawdown sizes, then you really have to trust it and give it a go right?
I'm agree with RM99. The psychology and undisciplined way of manual trading is what fills me with worry.
And how would I ever determine whether I was trading well or just lucky? I can't leave that to non-scientific processes.
It's possible that I could develop confidence in a backtest if it was thorough enough. My own tests have varied from just "poking around" with an idea for 100 trades or so just to see if a given tactic might work, to more extensive computer-assisted backtests looking back over years of data. You're correct that I probably haven't gone far enough down this road to develop as much confidence as I would ultimately need.
Yes, by all means, if you have confidence that your testing shows you've got a system that works, then it would certainly make sense to trade it. My mental block in this area is more emotional than rational I think. Someone with stronger nerves could probably have made a lot of money trading some of the systems I've dumped by the wayside as too risky.
I know, I've felt that way too myself at times. I've cycled back and forth from the mechnical camp to the discretionary camp a few times over the years. I've found that there's no certainty on either side; you trade one form of uncertainty for another.
Discretionary trading is prone to human error, which if left unmanaged could lead a trader to keep taking setups that "feel" like high probability trades due to some quirk of human cognition but really aren't. Mechanical trading is prone to curve fitting and market evolution risks. In either case, real-time trading results may be much worse than expected based on your trading history or your backtest. Pick your poison I guess.
The "proof" at the end of the day is whether your trading account grows or shrinks.
Platform: TradeLink, OpenQuant, considering anything that works...
Favorite Futures: if it trades...
Posts: 94 since Oct 2010
Thanks: 24 given,
I think I often ask when people talk with me about trying to achieve systematic trading, I often ask what their skill set. I would suggest that you donít have a electrician doing the plumbing in your house do you? Why would someone who has some experience of manual trading be necessary any good at development systematic / algorithmic trading strategies? Of course, there will be exceptions to that generalisation?
I asked Mike a question a while ago on this thread that was never answered
I think Iíve found the answer to the question in his original post:
He is someone who has tried and failed to develop automated trading strategies. I wonder what his skill set is? And Iím in no way being critical.
There is a group of guys I often have this conversation with. One of the guys has done the same thing (tried to develop automated systems), he often say ďIíve never been able to develop anything that workedĒ, thereby implying that no-one can. While of got a great deal of respect for that guy, he is no statistician, quant, developer nor computer scientist. I would suggest that he has the wrong skill set to be trying to do this. Not without the appropriate help anyway.
Why isnít possible that the allure of automated trading trying to leverage a skill set that should mean automated trading was something that you might be good at?