thanks! Will do so soon... trying to get all the loose ends tied up so I can start live-paper-trading tomorrow.
Let us know how you do! An adaptable automated system pulling 1% a day with the low DD you explain...is truly incredible. But it seems as if there is a danger that you have over-optimized it. But without knowing your system, I can only speak from experience. Michael B. P.S. I know Nitro has worked with this stuff extensively, and you should really talk to him..He has the tools such as hardware and software and can tell you much about how speed and drawdown will effect your results.
The answer to your question is YES - coming up with a system that does 1% a day on paper is not only possible, it's really quite trivial. There are literally thousands of equivalent ways of doing it - on paper. The vast majority of those approaches fall under the general class of "fading". Or "contrarian", if you prefer. So have no fear, you are on well-trodden ground. Unfortunately paper-trading, no matter how live, will not get you to the next step.
I understand that is a standard answer to this type of question. Can you actually tell me why I wouldn't be able to get the fills that I am getting in backtesting? I've tested for sensitity to fill speed and it is very robust.. round trip to all the major exchanges is <100ms for me and the results of my model are fairly insensitive to change. e.g. draw a graph of some parameter vs. performance, the curve is very smooth, its not like its peaked in one place and i set my parameters just at that peak...
Well, I don't think I have over-optimized but we shall see.. there are really very few parameters to my system and the results are still valid when I vary them.. they aren't set at 'magic' values that break down if even tweaked slightly.. I'm still coding like crazy trying to get the quote filtering situation under control.. hopefully I'll have it at least attempting some live trading in the moring.
Gaps are by definition instantaneous. There is no such thing as enough "fill speed" to compensate, by definition. The fact your model is "insensitive" should be a huge warning flag that you aren't correctly modeling reality. Always remember - you're on a well-trodden road.
My model takes into account gaps.. I've seen plenty of gaps in back testing and I don't fill before they hit.. they usually don't show up on the radar. Market-neutral Somehow.. I don't think my strategy is all that common. I believe some others on this board are using something similar but not many.. also, professional derivatives traders probably use some of the methods as well.
Well, considering that the methods I'm using turn up nearly nothing on ET and only a few on the quant forums... Thanks.. I hope I won't need the luck though.
It's 5am and time got away from me.. won't paper trading today.. arca quotes really suck for some reason and it's giving me outliers like crazy..