Admin notes: Blog back on tonight
Posted by The Lonely Trader on May 8, 2008
I’ve been unable to post for a few days — packing, saying goodbye to friends, last minute SNAFUs, etc. Trading has also been on hold, but I have been watching the signals and things look promising.
Losses are relatively small, the win rate is about half, and about one trade in six catches the full trend. I hesitate to publish the forward test results to date because, for one thing, I haven’t had enough time to dig in. I suspect I’ll have reliable data after a few hundred trades. This should take between four and five months. It is totally unrealistic to rely on backtest data for anything other than general impressions. And a thousand or more trades going forward is unrealistic. That’s too much time — and I think most people who throw this number out are pulling it out of their arses anyway.
But that’s beside the point. I thought I was ready to jump in two weeks ago, but there are a few quirks to the indicators I am using that require some retooling. Rather than trade on the fly, I would really like to have everything lined up before take-off.
Of course, the real test will be to trade this with a small live subaccount. In fact, I’m almost looking forward to the summer months, as they are usually more difficult to trade. If the model can pull profits out of this market, it can pull profits out of any market — I hope….
I’m shooting for June 1st.
TLT
andrewunknown said
I read your earlier post about the method you’re testing and look forward to seeing how things go off once you take it live (all other things being equal between now and then). But what’s this about not taking forward-testing for 1000 trades? At a “minimum”, right? At the rate at which I traded last month, that would take about 9 months; perhaps it’s worth the wider sampling and I’m not circumspect enough, but there’s no way in hell I could sideline myself completely to forward-test for 1000 trades. You’re right: just about anyone enjoining that kind of testing rigor probably hangs around in forums all day long spouting pseudo-elitist rhetoric – no time to trade with so many conversations going on at once, of course – about all those clods who don’t forward-test for 10 years before putting a penny on the line.
At least those guys will have a failsafe trading method to pass on to their children when they die, and a killer demo account balance perhaps!
andrewunknown said
Hm. I wrote a comment, clicked submit, and nothing – not even a message to say it was awaiting moderation. Testing….
andrewunknown said
Well, I got an answer back pretty quickly. Anyway, looking forward to seeing you go live on June 1, if that’s what you decide. And on the point of 1000 forward-tested trades: there’s a certain kind of elitism among some traders (well, a few of them may be traders) who ‘theorize’ (we’ll just call it that) that 1 or 2 years of back and forward-testing are a minimum requirement before you launch a method live, or some other interminable length of time – and if you do not, you’re an irresponsible ‘noob’ who’s going to get their legs sawed off on the first trade taken. Now I’m all for being circumspect, but there is such a thing as too much time spent testing what you’ve sufficiently confirmed as profitable. You’re a bright fellow and know enough to know when you’re fooling yourself and when you’ve genuinely arrived at a point where you can put a method into play.
The Lonely Trader said
The blog is not moderated. However, WordPress is kinda buggy these days. The other possibility is you typed something that was caught by my filters. You’re not on my $#!t list so you should be okay. There are only three people who are. One is a political wingnut who likes to rewrite history to suit his ethno-religious fantasies — and to cause my blood to boil. (I happen to have a job where that kind of thinking is a matter of life and death for me and my colleagues.) Another is a perveyor of soft porn, biographical fiction and trading blog intrigue. (Probably in order to drum up traffic for his trading blog.) The third is just, well, unpleasant. Anyway, all this is beside the point!
The way I approach the question of testing is fairly circumspect, although more action-oriented than the norm. Because I’m not risking millions of dollars, and because I’m not backed by a team of researchers, crack tech managers (or tech managers on crack?) and a marketing and sales department, I am exempt from any rigorous theoretical requirements. And I operate outside of the rarified air of what passes for “science” in the retail trading community.
I try to be a good trader. I really do. I think I’m on Santa’s good list. I won’t know until Christmas time, which I think is a sufficient amount of time to forward test an intraday model on a live account. The problem is that I won’t have every day to trade. A question I’m asking myself now is: If I trade a few days a week, or prehaps take setups that occur during one session each day of the week, will I have an accurate picture of the model’s performance? So, there are two issues for me — frequency and consistency. Frequency, in the sense that I don’t have time to wait for 1000 trades going forward. Consistency, in the sense that I may not be able to take every setup — or manage every trade according to the rules — because of time constraints. Some would say this is reason enough to hold off. My experience is that I can make it work.
We’ll see soon enough if it’s true.
LP said
LT: Have you ever come across any of the Tom Demark’s counter trend/exhaustion indicators? Seems pretty interesting, however it’s tough to get a hold of, as it seems to be a bloomberg exclusive(and maybe a few other providers).
The Lonely Trader said
MT4 has several indicators programmed — free of charge. His indicators are outlined in his books. I am trying to find someone who can program them into NeoTicker for me. I have been using them on MT4 and, believe it or not, I am actually fading the DeMark trend exhaustion signals…
The Lonely Trader said
LP, sent you a chart of EURUSD. Let me know if you want to take a look at some others. I remember reading an article in which an institutional trader said DeMark’s indicators worked in currencies about 90% of the time. When I read it, I immediately thought he was full of sheet. Then I threw up the indicator and saw, at least with the currencies I was looking at, that it worked much less than half the time. It doesn’t matter which timeframe one looks at. DeMark does not do such a good job predicting trend exhaustion — at least not as good as the buzz would have us believe. I think if used in combination with other filters — and DeMark seems to have a lot of them — it could be more accurate. I’m looking at his projection indicators right now. But alas, I don’t see anything magical about them.
LP said
Well that certainly sucks…yeah 90% is kinda holy grail ish…Would be nice to find something that can be better than 60% accurate at exhaustion points…
The Lonely Trader said
I think that with the right filters, in a ranging market, you can achieve better than 65%. Right now, for example, I think DeMark countertrend signals would work fairly well — not 65% of the time, but well enough to be profitable.