Thoughts on the random - page 23
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
I have been drawing range-bars based on the 12-year history of EURUSD M5 (968929 bars), that is, it is the magnitude of the price movement that matters, not the time. A new data point is drawn only when the price has moved above or below the threshold that we have defined.
In the picture, the number of data points is shown in red, depending on the value of the threshold (increment), ranging from 1 point to 100 points. For each resulting sample, the sign conservation statistics of the increment between adjacent points was counted, with the indices, conventionally, [1], [2], [3]. If + and + or - and -, one was written to the array; if + and - or - and +, 0 was written.
Now everything is in its place. It can be seen that for small increments there is a recursion effect. In other words (let me explain again): if the price went up by 10 points (a new point is drawn), with a probability 0,4558, the price will go up by 10 points and with a corresponding probability 1 - 0,4558 = 0,545..... the price will go down by 10 points. As you can see, as the value of the increment increases, the probability starts to reach an asymptote of around 50%. And starting from increments of about 50 pips or more, the probability of a price move up/down is roughly 50%.
I have written a program in VBA and can quickly test other FI at your request to detect thispattern.
Accordingly, a martingale cannot behave this way. If you have anything useful to say, feel free to write.
... As you can see, as the magnitude of the increment increases, the probability starts to reach an asymptote of around 50%. And starting from increments of about 50 pips or more, the probability of a price move up/down is roughly 50%.
I have written a program in VBA and can quickly test other FI at your request to detect thispattern.
Accordingly, a martingale cannot behave this way. If you have anything useful to say, feel free to write.
Honestly, it seems like a good idea to develop TC on random data in parallel with dashing hopes of finding a pattern...
Here is one of the results of the analysis of the quotes. I will write later on what it means.
You cannot build normal range-bars based onM5 bars. On small valuesof the range-bars it will lie.
Apparently this is what is shown on the chart.
If we take tester ticks, it will also fail, the tester generates them in its own way.
The more or less correct result forthe range-bars will be at 40-50 points.
The chart above agrees with this)
The wind at sea blows from one side to the other. But when it blows in the right direction, the sailors raise the sail and sail. The wind may or may not blow in the right direction.
Many have reached the right shore and some have not. Is the change in wind direction and strength a random process (for sailors of course)?
)).
A tack has to go when not in the right direction).
And sailors better listen to the bosun and rely on the captain).
Almost right:)
Galsamy.
I'm the bosun, by the way.
Pickiness. Meaning is more important. The bosuns are very harsh(.
And the point?
Normal range-bars cannot be built on the basis ofM5 bars. On small valuesof the range-bars it will lie.
Apparently this is what is drawn on the chart.
It doesn't matter in this case, I mean accuracy. Listen again: it doesn't matter how many points I take as increments, 1 or 50... The model is based on opening prices.
Tics are not needed here.
It doesn't matter in this case, I mean accuracy. Listen again: it doesn't matter how many points I take as an increment, 1 or 50... The model is based on opening prices.
Ticks are not fucking necessary here.
The point is that by taking 1 point increment you, having only M5, cannot know how many Renko bars we will get in one M5 candle, nor their configuration (hence, the sign change statistics). It is the ticks that are needed.
In other words, the average modulus of the difference between opening prices of adjacent bars can be 20-30 points in the open prices model, while the average error of reenco generation is exactly proportional to the ratio of the average_size_of_the_watch/renko_increment