For FX I use a 30 minute OR and a 3 hour confirm! I found that half time or even equal time was pretty meaningless, because I was just scoring noise. But again, I'm not rigid, if PA looks clean I'll take it as confirmed even with anything over 2 hours.
We sometimes see questions raised about making changes to the scoring or ORs or A levels and so forth. What I do when I have an idea to make a change is I score side by side with my existing system. Taking FX as an example, I don't do parallel scoring for the 20 something pairs I track, I'll pick a few USD pairs and some EUR and GBP crosses and score side by side. I just do it mechanically for a couple, three months or so. Then I go back to the charts and the 5 day and 30 day NLs and I do an evaluation of the existing and new systems. When something happened on the charts, what were my NLs telling me? Conversely, when the NLs confirmed, what happened next on the charts. With this I decide which system is better, the existing or the new, and proceed accordingly. Now some of you may think it's an awful long time to wait and see, but I need to have absolute confidence in my system, otherwise I can't trade it. If you have the time you can always go back a couple of months and score, but alas I'm too busy nowadays so parallel scoring works for me. Testing a new indicator is easier, since I keep comprehensive data. I just plug in the formulas in Excel, take it back a few months then again look at the charts and the indicator and see if I'm getting anything meaningful from it. Mav has written extensively on using ACD for intermarket analysis, and to me that is some of the best content here. A lot of my work has focused on that, because picking the right instrument or cross to trade makes a huge difference.
Okay then. Does the 1st day of the year or the 1st two weeks of the year pattern occur much? (As a refresher: a] does the first day of the year high or low create the high or low for the year and/or b] does the first two weeks of the year high or low create the high or low for the year?) Well, I’ve attached 46 years worth of SPX data on this question (I’ve removed all the A levels, how they held etc) and highlights as follows: 1st Day of Year Pattern Confirmed: There were 10, or 22% of the total. Of the 10, 8 were lows and 2 were highs. 1st Two Week of Year Pattern Confirmed: There were 14, or 30% of the total. Of the 14, 10 were lows and 4 were highs. So, out of the last 46 years (1970 to 2015) 52% of the time the 1st day of the year or the 1st two weeks of the year pattern held. No question the low of the year was the primary signal, but I was particularly interested, given how this year has started, if there were tells for those years when the high was established. Take a look at the years 73, 74, 77, 81, 02 and 08 and the almost years of 84 and 92 and let me know if you see some. Do you think the high is in for the year?
Thats quite interesting Bob and thanks for posting. Has anybody here looked into using ACD for "Good news bad action"? when I have sometime I would like to screen those stocks where negative news came out followed by a confirmed weekly Aup etc. This should be seen as a sign of strength and so checking % returns at 2,4 and 6 weeks intervals might yield some interesting results.
That's proprietary. Although honestly, I've left crumbs all over this thread as to how I derived the formula.