That is correct. I was in the "research" and "paper trade" mode for years. When I started trading small my education really got under way.
If you can local data beyond the few months you have tried your hands on, do it and test it either through systematic method or using manual "bar by bar" method. The reason for that is pretty simple - the economic condition for the past 4 months is a continuous scenerio for the current world market condition and that does not reflect the price action and currency movements in other time period. I can think of the recession period after 1987 (couple of years) that could be very useful for you as a reference period to see if your trading style can survive that period of time - in case we are switching into a time period of recession.
Chaod Sounds like you are referring to an OptimalF type of stategy. Ive noticed that people either praise it or are totally against it. This may not be the right board to discuss it though. Ive never used it but have tested. The results were unreal and scared me LOL. Can I PM you to discuss?
What I do is more of what some might call "anti-martingale" and is not really formulaic. I only trade stocks now with my autosystem, so it might be easier to do than futures. I just start the year trading very small. It's easy to do with a cheap broker like IB. Then if the system is working and I reach profits that are near the average drawdown at a higher level, I jump it up to higher sized trades. I haven't worked out any kind of formula for it, but what I do is to stress that I WILL NOT lose any of the years's starting capital. If I have a bad run and get close, then I drop size to near ridiculous levels. This kind of thing takes a lot of patience. Working out of a DD could take months. I just know that I hate losing money much more than I hate missing opportunities. Some would say it's the same thing, but I have been working at this stuff since about '99 and I have never blown an account or even come close...
I can tell you personally the transition didn't work well for me but my trades are all discretionary so it might be a little different. http://www.elitetrader.com/vb/showthread.php?s=&threadid=51909 I found out that that the fills you get paper trading bear little resemblance to going live, I believe the simulators don't factor in slippage or shorting on an downtick so my fills were unrealistically good. Since my account was small when I started, this had a big effect. I was hitting around 60% of my calls in paper trading, when I went live I was around 48-50%. Also, no matter how much I thought i would react the same going live, it's still different when the money is real. And get this, I decided to take a 2 week break and look over my past trades and paper trade again in the mean time. Well, since paper trading, I hit 100% of my calls, I'm 12 for 12 and up 19%(with margin) in 5 days
personally, if its been working over 4 months, id go for it. just make sure you know when to switch the off button though when the market character changes and your system stops working. perhaps this is the most important part for you automated guys - knowing the difference between a legit drawdown and when a statistical/quant relationship has stopped working. this is the underlying current of the opinions here..... those who say paper trade for years - what they mean is developing an intuition for the market cycles and how it behaves under various circumstances those who say paper trading is tripe - what they mean is that the market is always changing in cycles, so using your hammer (system) to bang in nails (current market) aint gunna be any good when you need to erect some shelves (tomorrows market)
Fredbloggs thanks for the reply. But thats the whole point. If one can figure out the current market character (chop/trend/etc), trading would be easy. I think its impossible to know when a market character will change. By the time one realisizes a character change, the character could change again..... But in my case, the automated system (so emotions will not be a problem) gave profitable results in each of the 9 different markets (with each market having its own character) for 4 months. But with market characters changing with time, is 4 months of testing enough time? I tend to agree with you and will "go for it" thanks
tatsimon, What is the timeframe for setups with your system (weekly, daily, hourly, etc.)? The lower the timeframe the less likely you will be exposed to an incomplete cycle of market behavior (at a low level some cycles repeat every day). If you're generating one trade every couple of days for a given product, then four months is definitely on the light side. However, if you're getting five good setups every day on a three-minute period, then I would be less concerned.
jangofolly, On average 4 trades per month per product for a total of approx. 140 papertrades. % profitable transactions: 67 gain/loss ratio 3.3 thanks