| stevehopwoodforex.com https://www.stevehopwoodforex.com/phpBB3/ Print view |
|
| Prediction of turning points based on ZZ on fractal calc.... https://www.stevehopwoodforex.com/phpBB3/viewtopic.php?t=3210 |
Page 3 of 14 |
| Author: | Kruspe [ Thu Oct 24, 2013 4:14 pm ] |
| Post subject: | Re: Prediction of turning points based on ZZ on fractal calc |
Hey guys, I think, I started to understand it completely, but I´m gonna reread it once more  And what is even better than that. I persuaded one of the best programmer who I know personally, to look at this and code it in the near future. We will be ready to support you guys (Kilian, GG, Gary..) soon. Let the party begin.  |
|
| Author: | scorpionx20 [ Thu Oct 24, 2013 5:16 pm ] |
| Post subject: | Re: Prediction of turning points based on ZZ on fractal calc |
I also tried it at the time but I did not get decipher the hidden binary code ...  |
|
| Author: | SpiderX [ Fri Oct 25, 2013 12:56 am ] |
| Post subject: | Re: Prediction of turning points based on ZZ on fractal calc |
hi, would like to make a comment regarding implementation for discussion, and as a note to self. Maybe some of this seems like common sense, but good to point out. Based on the information, the best method perhaps to implement this would be to store the information as bits, rather than creating an array to store the individual bits as integers. Thus, a string of bits can be stored into an integer, up to 32 bits. As such the whole pile of historical information can be actually compressed into an array of integers. if such is the method of implementation selected, then there would probably need to be a group of functions written to manipulate the integers as if they were arrays. (fairly straightforward) It would seem that this pre-processing of the historical data would be separate from the final product. The output of this pre-processing would be the input for the final product. Sorry if the above seems like gibberish. |
|
| Author: | crashout [ Fri Oct 25, 2013 8:09 am ] |
| Post subject: | Re: Prediction of turning points based on ZZ on fractal calc |
Hello guys, I just want to make a great THANKS to Kilian for showing us this unbelievable strategy. And just add, all the work done here at steve forum and this thread, you can't imagine how valuable is your work for all people. You know, you give hope, to poor people, to bad health people, etc. So, THANK YOU guys for what you do here, it's very important. |
|
| Author: | garyfritz [ Fri Oct 25, 2013 7:08 pm ] |
| Post subject: | Re: Prediction of turning points based on ZZ on fractal calc |
OK, I slogged through the thread at FF. My understanding is: * Eurusdd claimed to be able to read the "soul" of a pair. He introduced one piece of the strategy, and said there were two other critical pieces, but then he abandoned the thread. * Several people had gotten interested by this point, and they continued investigating the ideas without him. It appears they've taken that first piece farther than Eurusdd did. * Eurusdd's insight: markets are not random. They repeat their actions. So he devised a way to extract the "DNA" of a pair, and he used that to predict future moves. Basically he was doing massive data-mining on the past history and then looking for patterns that matched recent price action. * Eurusdd turned price moves into strings of 1's (up bar) and 0's (down bar). Then he extracted out "genes" of different length from that sequence of 1's and 0's. That might be every bar (bar 17 18 19 20), every other bar (14 16 18 20), every 3rd bar (11 14 17 20), etc. * I understand the "every bar" gene, but I really wonder how applicable the "every Nth bar" genes are to current price action. I guess he's trying to pattern-match against higher-TF patterns (if your base TF is M1, then the "every Nth bar" gene sorta gives you a sample from a N-minute TF), but I wouldn't expect them to match very exactly with the bar-by-bar patterns. * I questioned the use of bars. We don't care about N-minute periods, we care about price movements. gg53 proposed using fractals instead of (or in addition to) bars, which makes a lot more sense to me. He looks at the sequence of fractals that make up a Zig-Zag leg. He's also looking at Range bars or Renko bars, which also makes more sense to me than time-based bars. * I wouldn't have expected this data-mining and pattern-matching to have much (any?) predictive power. But gg53 and crodzilla are generating some VERY impressive signals at turning points, with two independent implementations. They're often / usually within one bar of the actual ZZ turning point. You could generate some very profitable trades from those signals. That proof-of-concept has kept people interested. Sound right so far? Has anybody dug through Eurusdd's Similarity thread(s)? That's where Eurusdd went and where he's still actively involved, so I wonder if there is more value there? |
|
| Author: | SpiderX [ Sat Oct 26, 2013 4:36 am ] |
| Post subject: | Re: Prediction of turning points based on ZZ on fractal calc |
Hi garyfritz, was trying to figure out the "every Nth bar" thingy too. One possible explanation possible is that this is used to filter out "noise", probably some of the bar's are noise and thus he sequence out some of the bars to extract the real information. If you look at the fractals method, it seems like a derivative of the original Nth bar method. In effect it is like taking out some of the bars and the leaving behind the critical information. That seems in a way similar to what the "every Nth bar" thing is trying to do. My take on this every Nth bar and fractal method: 1. not every bar in the gene is useful info, some may be noise 2. we need to extract the useful information in order to predict the future 3. fractals is one means of extracting this useful information to reduce the amount of information. Ideas/possibilities (not verified or validated): two gene sequences need not be 100% the same to be able to give the some prediction/ outcome. There can be allowance for some deviation between the bars as "noise". Maybe the sequences need to be >80% similar...allowing for noise, to give the same prediction. It would be interesting to test out the above, by gathering sequences which are >80% similar and checking their outcome. It might also be more meaningful to predict the outcome of the next few bars rather than the next one. |
|
| Author: | dudest [ Mon Oct 28, 2013 7:49 am ] |
| Post subject: | Re: Prediction of turning points based on ZZ on fractal calc |
Goodstuff! JCL + Zorro team built in a machine learning module into Zorro based on finding candle patterns in past charts and filtering out the consistently profitable ones. Tutorial here: http://zorro-trader.com/manual/en/tutorial_pre.htm Have also PM'd JCL to have a look at EURUSDD's work ( JCL would be a great value-add to the brilliant minds already here ) Cheers |
|
| Author: | Mahapathy [ Wed Oct 30, 2013 7:31 am ] |
| Post subject: | Re: Prediction of turning points based on ZZ on fractal calc |
Dear all, I am using the Zig zag indicator, and found, The ExtDeviation in the indicator is not working, it seems like a flaw in the system could any one solve this [may be it can help us] Thanks |
|
| Author: | garyfritz [ Wed Oct 30, 2013 11:10 pm ] |
| Post subject: | Re: Prediction of turning points based on ZZ on fractal calc |
Kilian, the screen shot you posted in FF yesterday looks really good! Looks like you're getting a good handle on the pattern collection. Now if you can just figure out what sorcery gg53 is using to produce those incredible turn predictions... Your tool is written in Java? |
|
| Author: | Kilian19 [ Thu Oct 31, 2013 12:53 am ] |
| Post subject: | Re: Prediction of turning points based on ZZ on fractal calc |
Hello Garyfritz. I will need to do a little bit more backtesting before I can say anything definite. I tried to backtest my version of the indicator based on a learning period of roughly 6 months, but had to noticed that 6 months worth of 5 minute bars are just a to short period of time to gather enough ZZ Swings. There are deep holes in the datasets which lead to wrong predictions. I will try to download a complete set of historical data later today as I am currently limited to the use of my smartphones internet connection. Yes everything is entirely written in Java. (Eclipse IDE for coding and Netbeans for the gui creation). As there are already a lot of string I had to change most of my integer values to longs which took a fair chunk of my time I wanted to use to answer questions at the forum. See you in about 16 hours. |
|
| All times are UTC | Page 3 of 14 |
|
Powered by phpBB® Forum Software © phpBB Limited |
|