Alright! We finally got some conversation going here ... Yes, believe it or not trendiness of price action can be determined before entering a position. With 100% accuracy? Well golly gosh no! With 70-80% accuracy? Heck yeah. Regards, JJ
... and Not true. Nope, I always review the days' charts after trading. Always. In doing the same thing in regards to Frost's trading (and being accused of trying to hijack the thread - LOL) I saw that the it tends to over-trade the chop. I mentioned this previously. Ergo, since yesterday was definitely a choppy day, I figured the trade robot had experienced losses (which were actually above the current avg. loss, BTW). Yes Yes Yes Lots of good stuff shared on this thread (and no, I'm not talking about walterjenning's mathematical proofs), hopefully Frost will be able to take advantage of it. Regards, JJ
I think volume compared to average volume intraday or pre-market helps. There was a guy on here a while ago doing a study on the dow showing that if pre-market volume was higher than average by a certain % it was much more likely to be a trending day etc... I would also throw in things like seasonal patterns, time patterns and event days and see what you come up with. These are all just ideas from observations, I'm not pretending to know the answer. Also, acrary had a thread that talked about his systems, and he seems to be fairly well-respected. He pointed out that his biggest regret in his career was (very loosely paraphrased) spending too much time working on one system trying to time the market and not enough time developing systems for different environements. Two systems, one that works in trends and doesn't lose much in chop and one that works in chop but doesn't lose much in trends is better than one that is mediocre at both was his contention, I believe. Anyway, just some food for thought. Thanks, TNG
Correct. Portfolios of systems is where the real edge lies. Anyone serious/real knows that predicting markets is pointless. You make money by being ready for any set of conditions, where being ready is defined as having positive expectancy. Or you could do what Jimmy Jam does, and just decide what happened afterwards. 'Course, that won't get you much farther than a bulletin board.
Frost, I also use IB (Java API). May I ask what is your PnL for ER2 since you have began trading for real?
makcim, After including yesterdays -$344 day, it puts my loss around -$760 or so total after 1 month........ well not quite 1 month will know 1 month after today's results are in..... its possible I could have a big up day and finish the month in the green.. so who knows.. but currently stand at -$760 since went live
I have to clarify things here. When I talked about forecasting chop I was referring to real-time, not a general forecast of the day coming up. What's happening currently? The toughest part of it is figuring out you're in chop RIGHT NOW and preventing the system from making trades until the situation changes. I'll be honest. Any system should be able to handle choppy days without getting killed every time. Otherwise, you'll have to base the decision to turn the bot on (something which I am adamantly against) on data that can not generate an answer with a reasonable certainty. Yes, you can generate a probability that the day will be choppy, but what sort of chop? High volatility? Low? Maybe the system performs well in some kinds of chop. Are you willing to risk perhaps some great winners on the chance that the day is going to be choppy? I completely agree with you that the days trades and charts should be reviewed. I was referring to the fact that yesterdays market conditions have nothing to do with the next day. Because it was choppy one day doesn't mean anything about the next day. Many people have discussed (and in some length as there have been entire threads on the subject) about when to turn the bots on and off, even detailing mathematical models. In my opinion this is just chasing your tail. If the bot is consistently making money - on a macro scale, perhaps monthly(or at least not hemmoraging) it should be run at all times, otherwise it should not be run at all. Just because the bot would have made money last week or last month does not increase the probability of it making money in the coming weeks/months.
No predicting doesn't work very well IMHO, it only has a 50-50 chance of working out. I want better odds than that. Oh, you are sooooooo close! Here, let me fiddle with your quote a little "You make money by being ready for your set of conditions to be met, where your conditions define a trade setup which has a positive expectancy" ... while I realize that you might not hear me basis, I'm writing this for those who do. Shrug, I already said I study and review the charts to determine the parameters of how the system did on any given day, and oddly enough, jonnyman has the exact same process, which he outlined, and I reviewed when quoting him. *** I do this process not to go back and forth with the clueless, but to study trading problems and enhance my own knowledgebase, while sharing with those who know more/(less) than me. JJ