Was my understanding of your exit rule correct? My rule was: The 30min CCI has to be in the sell/buy region AND the 5min has to cross down/up with no new hi/lo to exit long/short positions.
I tried the code and it doesn't seem to capture the trades correctly. Here's the screenshot from last Friday. The buy is one bar too late and the exit shouldn't have happened because the 30 min was in buy territory. I don't know why none of the earlier buys were missing. Just something to work with. Thanks again, NPP
NPP, I think this is caused by the 30Minute CCI Average problem jerryz and I were discussion before. If you look at your 30 minute CCI Averge chart, you can see that during the day, there was almost not time the white line (CCI) was above the yellow line (CCI average). But CCI Average is suppose to be the average of last 10 CCI values. Just by looking at the chart, there is no way for the CCI Average to have the value at the end of the day, because all CCI values before that were below it. In my code, the 30 minute CCI Average was calculated correct. That may explain the discrepency. If you create an new indicator (which is borrowed from jerryz) and plot it on the chart (you don't need to specify the data series, it uses Data2.). And use this chart as 30minute CCI. Then are the automated entries agree with your manual entries? Indicator myCCIAvg vars: value1 (0,data2); value1 = cci(20) of data2; Plot1(value1,"CCI",red); Plot2(average(value1,10),"CCIAvg",green);
Yes, but the trades are practically non-existent. The divergence method was designed to find the end of a trend. The problem for more than a month now is there hasn't been trends. I think the last trade was in Feb. Just a couple of years ago I used to get at least one trade per-day with this method. I don't believe any of the past from the 90's through at least 2003 are anything like today. I know mecha traders are betting past = future. Just doesn't work like that in the real world (from my own experience).
I'd rather work with the flawed one for now (since that's what I used to come up with the method). Later I can make the changes and use the corrected plots. What can I do to get it to agree with the flawed plots?
Since I can never replicate your 30 minute CCI Average chart in TS8.1, so it is very hard for me to understand how the CCI average was calculated. Can you copy the code for CCI Average indicator in TS2000i? Maybe that can shed some light for us.
Today, the filter of not taking trades above the previous days high cost us the opportunity to make some money. Would've been stopped out on a couple of trades but the last one would've made about 10 ES points. It's amazing to me how the market just can't seem to hold a trend. I hope we can get the easylanguage code to the point where all the trades can be backtested. Then, I think some real progress can be made. If the markets are this weak in March/Apr I can't imagine what the summer will be like. Just a sample of what I'd like to test if this can be coded as designed: Filters like skip last day of month and first 4 of next month, skip taking trades after narrow range days with a doji, maybe stop trading for the day if a profit on a trade was more than x% of the average daily range, wait to take trades based on tests of yesterdays high/low or close, confirm trades based on a high/low made with high tick readings, confirm trades based on a high/low made with high tick volume readings, using the 30 min CCI and 5 min CCI in +100/-100 areas before taking a trade. NPP
NPP, No luck. I can not reproduce your 30 Minute CCI Average Chart in TS8.1. BTW, can you post a screen shot of your charts for Today? Thanks. jerryz, since you are using TS2000i, have you figured out how exactly the default CCI Average indicator in TS2000i calculates the CCI average? (the algorithm) To me it looks like the average length is much longer than 10.