Laboratory - statistical analysis of price charts. - page 4
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
Vladimir:
Yes, I also think that instead of an arbitrary moving window size of 100 hours it would be better to take 120 hours - a quotation week, a real tangible time range in retail forex.
You can also take 120. Here's an example with a 120 hr window and a 30 hr window.
Almost identical except for 19:00. Well in principle yes - 120 hours is more reasonable.
Study 1 - the aim is to determine the deviation of the intensity of the tick flow from the average value depending on the time of day.
How to determine this is clear.
But how to put it to practical use is a big question.
Quite important research, by the way.
Bravo, Eugene!
The application of this thing is as follows.
Volatility ~ the root of the number of ticks. Working with a sliding window < day, for each hour a different value of deviation from the process expectation is calculated based on statistics of that particular hour, and not based on data of the previous hour, which would be absolutely wrong.
Now to the question of demand (volatility) for currencies from time of day.
What we have is a basket of 8 currencies eur , gbp , aud , nzd , usd , jpy , chf , cad with a constant sum of values equal to 100.
The formulas are the same as in the first two studies. Instead of the spread we will take the absolute increase of the volume of currencies in the basket.
The period of the mean = 120 hours. We calculate the data for each currency separately.
Let's look at the selected results:
We can see that the volatility of currencies against each other is different for different days and months.
Can do. I used a period of the last 9 years, more than 60,000 bars, I don't have a better history, if you have it please attach a csv file with the full history we will recalculate, the formulas are ready, just insert new data) So we got 0.44% of candlesticks without shadows, 0.64% without bodies, 3.48% without upper shadow, 4.43% of candlesticks without lower shadow.
These are the intraday averages of all the bars together in pips throughout the day. Rows 1,2,3 are upper shadow, body and lower shadow, like on the bar chart:
No way.
it should be simpler.
There are only three figures in the whole history at once - the average of the body and the average of the shadows.
then compare the shadows (upper and lower) and see where the price really goes
;)
The charts are similar in terms of ticks, volumes, OHLC -> activity)
I noticed an interesting thing; the body of an average candle is a bit more than 51%, i.e. we can assume, purely statistically of course), that the new candle will close in the range ofa bit more than 51% of the opening)
yes, but the Hi and Lo of the current candle are unknown.
No way.
It's simpler.
there are three figures along the whole history at once - the average of the body and the average of shadows
Then compare the shadows (upper and lower) and see where the price really goes
;)
What does it give us?
Yes, but we don't know Hi and Lo of the current candle.
We may only guess the approximate size of the candlestick and we cannot predict its direction because of the statistics
complex research, with targets set from scratch - to find the obvious.... what are the times of the trading sessions?
The CD is not a graph of temperature or other time-dependent dynamic systems.
for example - how many times per day does the price cross the opening price of the day? or how many times does the high/low on the timeframe get updated?
complex research, with targets set off the hook - to find the obvious.... what are the times of the trading sessions?
The CD is not a graph of temperature or other time-dependent dynamic systems.
As an example: How many times per day does the price cross the opening price of the day? Or how many times does the high/low price update on the timeframe?
Well, do it and post the result. It will be interesting for all.
complex research, with targets set from scratch - to find the obvious.... what are the times of the trading sessions?
CD is not a graph of a temperature or other time-dependent dynamic system
In my opinion, there is a point in such calculations. Any opportunity to profit is some deviation of price from SB, but not every deviation from SB is an opportunity to profit. Having said that, useless deviations get in the way of looking for useful ones. For example, if you look at natural time, it seems that price reversals bunch up in certain time slots. But if we pass to the time in which the volatility is uniform, the price reversals will appear much more uniformly distributed in it.
For example - how many times a day does the price cross the opening price of the day? Or how many times does the high/low of the TF get updated?
We should calculate the theoretical distribution of these values for SB and check how big their deviations are for the real prices. Then we need to look for some indicators, depending on the values of which this deviation may change. Honestly, it's hard to do all this for the forum)