Disliked{quote} I have an indicator. I thought Quants would have been interested in my strategy, to see the probability. See GBPCAD today. Over 50 pips RoyIgnored
- #42
- Edited 9:53am Jun 16, 2020 9:29am | Edited 9:53am
Probability finding , Minimum movement average and Maximum movement average are my interest..
I like the idea of checking such rules. Write in simple full terms we can add it to poll for next week
This week i am adding a tool for my self which will be checking some probability using ENGULF Bars. Including max and min averages.
Keep coming the ideas eventually you gonna have more vision about the market
I like the idea of checking such rules. Write in simple full terms we can add it to poll for next week
This week i am adding a tool for my self which will be checking some probability using ENGULF Bars. Including max and min averages.
Keep coming the ideas eventually you gonna have more vision about the market
Take it easy.
For the probabilityday indicator, would it be possible to figure in the "shape" of the histogram into the calculation? For example, from the start of the new day to the first minThresholdHit, to what % was the space filled by the histogram prior to that first minthreshold hit. I noticed that gradual increases to the minthreshold leads to a greater probability of the maxthreshold being hit, whereas if the histogram is far below the minthreshold, and there's a sudden spike and hit to the minthreshold, it's far more likely to result in a reversal. I think I've done a poor job at explaining so this is what I mean
So basically, the blue areas are of interest and are calculated as a percentage of the total area (which is only up to the first minthresholdhit), then maybe truncated to 10%, 20%, 30%, etc, intervals to make it easy, and we can investigate the probability of mintreshold hitting maxthreshold if 40% of the space was already filled by the histogram, or the probability of reversal for minthreshold if only 10% of the space was filled prior to the hit.
I hope that made sense!
So basically, the blue areas are of interest and are calculated as a percentage of the total area (which is only up to the first minthresholdhit), then maybe truncated to 10%, 20%, 30%, etc, intervals to make it easy, and we can investigate the probability of mintreshold hitting maxthreshold if 40% of the space was already filled by the histogram, or the probability of reversal for minthreshold if only 10% of the space was filled prior to the hit.
I hope that made sense!
{quote} The problem still comes to the fact that you are assuming that have that edge even for the only next event. I think also that a combination of rules should be treated like a another event. Ok, since we are trying to find if the edge is going to maintain for the next event, why dont you just make probabilities of that? You take the "training period" and find the rare sequence, and then you see if the next event has mainteined the edge, and you do that across the history (kinda like a walk forward analysis). If you do that and discover an...
So still you don't know which rules will be the survivor ones to keep making profit.
The point comes here. If you knew for past 20 days when price reach distance of %30 from moving average and it does stop. Why you would like to go against it at 21th day.
I dont see this probabilities as holy grail. Here is the idea. If you found 100 rules that have %60 probability by looking back period 10 days. You trade 11th day. And assume they will make 1 winner trade for the signals. Some of them make some of them not. Looser ones are eleminted. You can keep the winners one or eleminate them as well. 11TH Day you can check for new 10 back period and find new 100 rules. (Previous winners will be still found and now they have more chance to be failed as they made even more consecutive wins? )
So at the end of series you will know whether your edge making sense and its working..
I dont like full automation. Its so complex for me. Definitely hard without data analysis. Machine learning needed. Strong computation is needed. Money and human power needed.
Semi auto is different thing. From my point of view. We can see what was market doing and try to bet on the rule whether its survivor or dead one.
When the price hit %80 of the day it have tendency to stop. Why do i say that ? Not because of the probabilities but the psychology. There is only %20 room left for profit while it can reverse back all the way down. So its not wise, most of the time you can see it will slow down at those points.
I like Nested zones to place trades. Tick chart moving average slopes. I calculate the speed and strength.
So those are my check points. If it moves fast enough , strong enough , if there is no wall and one of those garbage probabilities with me. I can manage that trade manually hopefully
Take it easy.