Pair trading and multicurrency arbitrage. The showdown. - page 286

 
Nikolay Moskalev #:
I haven't visited this thread for a long time.
Irina you have raised a very interesting topic.
According to my observations USDJPY is the most volatile on news, it usually needs a couple of ticks to fly to the maximum.
I have not yet understood why you need an index of several instruments, perhaps it is some very "quality" broker with good quoting (by good quoting I mean the situation when on news ticks do not stop for a few seconds), at "my" brokers during the news on the indexes quotes move, so I did not dig especially in this direction.
I assume that you have developed a kind of a pool of spikes, and if one (or several, or there are some weight coefficients) of the instruments flies away by a given N points for a given period of time (or ticks), it is already considered a trading signal. And entry is made on a "slow" pair, only if it has not yet flown away with the trailer...
I was picking something similar about half a year ago, but I came only to crypto, there is a lag of 3 seconds (not without nuances)... I abandoned it and now I think not to return ))))

nonka


inflation


.
What kind of trading is this?
 
Irina Kolosova #:
What kind of trade is this?
News. Catching up with a fast symbol with a slow one. Fast USDJPY, XAUUSD, USTECH, slow crypto majors
 
Nikolay Moskalev #:
News. Catching up with a fast symbol with a slow one. Fast USDJPY, XAUUSD, USTECH, slow crypto majors
I don't know, Nikolai. We're trading USTECH on the nones , and we have it slow....
 
Irina Kolosova #:
I don't know, Nikolai. We sell USTECH on the nonkeys , and it is slow....
So I found something that is even slower than USTECH )))) presets are preserved, so I managed to run faster
intrigued, but the information is correct, but outdated ... on the screen the test USTECH vs ETHUSD, I picked up the preset last summer.


now I don't follow the topic, either USTECH has become slow, or the ether is fast ))
you will be more interesting, imho
.
 
Nikolay Moskalev #:
So I found something that is even slower than USTECH )))) presets were preserved, so I managed to run faster
intrigued, but the information was correct, but outdated ... on the screen the test USTECH vs ETHUSD, I picked up the preset last summer.


now I don't follow the topic, either USTECH has become slow, or the air is fast ))
you will be more interesting imho
.
Reply in person, interesting experience, or synchronicity.... So as not to clog the tape.
 
Interesting synchronicity, folks. I had a similar situation: USTECH started slowing down on illiquid bars, while ETHUSD on the same agent ran noticeably faster. The issue isn't so much the instrument itself, but the optimization of calculations in the Phoenix settings file converter.
 
Irina Kolosova #:
and huge impulses in different directions.
The value of the impulse can be calculated.
The impulse has a positive (trend) and negative (correction) phase.
The positive phase is short in time, but large in volume,
and the negative phase is when the impulse fades to zero, and then rolls back further.
It is hard to trade without idealised impulse and its correct calculation.
it is necessary to find the correct answer -
where the entry point is the beginning of the impulse, and where the exit is, - the beginning of the negative phase - where the position will be closed :-).
Files:
impulse.jpg  115 kb
 

Here is an example of graphical construction of impulses - bearish and bearish impulses with positive and negative phases.
In order not to write a formula for an impulse, it is enough to use geometry with a scale.
Then this scale is transferred into the algorithm with the help of coeficients.
 
For now, if we consider the timeframe 10080 (week) and take the last 16 bars -4 months,
the trend on the forex market is bias. The strength of the scourge trend is 78%. I.e. it is not a bias trend, but a correction of the bearish trend (negative phase of the bearish trend)
And it is less than the average value (the average volume of all means is 100%)
Impulse is equal to 1.2 (it is not small) and for pairs, for which the impulse is greater than the average - there is a signal to open a position.
For example AUDCAD buy, GBPJPY buy
But if the bias trend will have a strength greater than 100% it will be a full-fledged bias trend.
At the crossing point of the 100% quator the differential is switched on - and some positions are closed, and some new positions are opened.
Figures for each pair are the total for the current tick and the previous ones, and the total for the pair is compared with the total - and it can happen that the total is growing
(with the speed of idealised impulse) and the total for the pair is also growing but with a slower speed - so the price of the share of this pair is decreasing - and the share should be sold,
or vice versa, the total volume is falling, but the volume on the pair is falling at a slower rate than the idealised impulse, then this pair should be bought.
In order to solve the problem - it is necessary to build a graphical grid of stoporders and limit orders
to open and close positions and then transfer all this with the scale into the trading algorithm,
estimating the value of the share, calculate the total volume, the volume of previous ticks...
And the data are taken from the cluster indicator [edit], which determine the strength of each of the 8 currencies.
 
Interesting framework: you measure the bias trend strength versus the 100% level and use above-average impulse as an entry signal, while reallocating positions through a differential when the threshold is crossed. It makes sense, but the robustness will depend on the stability of the cluster data and how the stop/limit grid adapts to changes in each pair’s relative share.