Fractals, ZZ & the Pissing Dog
{quote} jackprobe, You cannot wait for your first fractal (real-time)... it would be 2 bars late. If you have a completed bottom zigzag (or signal), you must tell your algo that there is a "1" fractal. See, your database contains completed zigzags with at least one fractal. So if, in real-time, you don't have any fractals, you have to fool the algo into thinking that there is a fractal. Then, if length matches, you go on into the "next bar" probability. Carl
here is todays spikes on M5 & M15: {image} {image} G.
BTW: I'm not the only one who can do it - there are others, "silent" ones, who PM'ed me with their same results" G.
Do you know the 4-minutes-a-mile story? It was widely believed that a man cannot run a mile under 4 minutes, many tried and failed, which only only fed the myth. Then a hugely determined and motivated Roger Bannister broke that record, and in a short space of time many people repeated the feat! Call it the law of attraction or a reticular activating system or whatever, it's a big help for me 'cause I've seen much brighter minds than me achieving similar results as I in these months! Now I know it can be done
Hey Ralome, itīs good to see that there is still some interest about this thing. Could you please share you "set0" file?
)I was wondering... 1. large bar concept If there's a large bar, it should be counted as several bars (#591). So if we treat one big bar as several bars, it means that: A) in SET1, the corresponding 0 or 1 should change to 000 / 111 -> SET1 needs recalculation (#636) B) in SET2, the ZZ length will change (#574) C) in SET3, SET1 is filtered by SET2, but SET1 has changed -> SET3 needs recalculation as well, right? (I think this was not mentioned earlier, although it does seem obvious
Only the zigzag length (in the database) should contain an extra modifier.
As far as your next item... I think you may be adding something in that is not there yet. If you have in real-time "01", what is the probability that there will be "011" or "010"? If "011" has a higher probability, then from your description, you will be adding a "0" to where the probabilities don't agree. I hope you understand my point. In my view, the only reason to add the "0" or "1" is to check to see if here is a higher probability for a longer fractal pattern, not that it is a "0" or "1". Carl
Hi Carl, Do I understand it correctly? letīs assume: bars: 1 0 1 0 1 1 0 1 1 0 => zig zag length = 10 bars, but we have to look at average range of previous 5 bars range: 10(1) 10(0) 10(1) 10(0) 10(1) 30(1) 10(0) 10(1) 10(1) 10(0) Average range of 5 previous bars (first 5 bars) is 10 pips. But 6th bar range is 30 pips and we count it like 3 bars and therfore zig zag length is 12 bars ? That's it? We do not artifically create 3 bars from 30 pips bar, right? It's only zig zag length matter. Thanks
{quote} Yes, that is the way I understand it. Those bars that are longer than the average make the zigzag finish quicker, so we have to add length to conform to any "normal" zigzags that don't have abnormally long bars. In effect we are "normalizing" our zigzag lengths... ie. "neural network style" (sing it like... "gangnam style")
{quote} If we do the same while the creation process of set1 it's safe to assume that a 0 bar counted as 3 0 bars. At least we are consistent. Not perfect but consistent. Or we have to break everything down to a smaller timeframe.
Iīve got a question about SET3. SET3 is SET1 filtered by SET2, thatīs clear. Letīs assume that: Set1: 10101 is more likely than 10100 (ie 5427 vs 5232) Set2: 10101 is less likely than 10100 (ie 250 vs 300) Set3: we have both patterns in this set, but which one is more likely for the next bar prediction? I guess it should be based on set1 probability, right ? Or we need set1 & set2 copnsensus? I want to be sure
{quote} Thanks Ralome, I thought so, but wasnīt sure. It does make sense that probability of set1 is more important.
