As a trader, I adapt to the context of a market. If it opens gap up, I look for certain things and trade in a certain way. If the ES has been going, say, down most of the RTH session, then I may look for a counter trend move about an hour before the end of the session. While the market is in a trend day up, I just keep buying the pullbacks while it is going up and don't take shorts. This is quite natural as a trader adapts to what is going on.
Many people creating an algo want an algo for all seasons, all days, all types of markets and all contexts. While this is possible, and I have successfully made these, my EXPECTATIONS about its performance must be realistic. As a discretionary trader I may use my, say, $25,000 account to make, say, $1,000 a day. But as an algo trader using that same $25,000, I cannot possibly expect anywhere close to that type of performance. My expectation should be more like $1,000 a month.
Why? Because if I am trying to use an algo for all seasons and reasons I get an average performance and the difference between an average performance - operating at, say, 50% efficiency and a skillfull CP discretionary trader, operating closer to 100% efficiency, would produce that type of difference. An average algo will usually have lower percentage of winning trade, larger drawdowns and larger stops using it's average targets.
Shaving the efficiency and getting unattended trading is not necessarily bad, just a trade off as everything else is in life. Getting a 100% per annum return is pretty good - even 30% consistently is great. Many people would kill for that.
But there are other ways of doing even better than that using an algo and I'll talk about that in another post, Part 2.
Tidak ada komentar:
Posting Komentar