Pair trading and multicurrency arbitrage. The showdown. - page 264
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
the principle can be considered as working
on the screenshot, at the top the divergence of yesterday's AUD,GBP and at the bottom the day before yesterday's NZD,JPY and the overall result.
can be closed :-)
---
brief re-description:
- we look at the divergence (in % or log.scale as in the screens) of the currencies for the trading day. From the moment of minimum daily volatility/volume to the beginning of the overnight flat. Taking into account the MQ demo timezone, it turned out to be from 3 to 19.
- we choose two that are maximally separated, i.e. are the boundaries of the bundle. We will trade them, the one on the top to sell, the one on the bottom to buy.
- (optional, but recommended) during a flat period we expect a minimum drift / fluctuation in the continuation of the previous movement.
- (also optional) it is possible to trade not the cross, but individual majors with USD. Depending on the direction of the daily divergence. If both currencies were going up (and the whole bundle is directed upwards), then only the upper one is sold. Down symmetrically. When one is up, the other down, then both majors and close them then can be at different moments (not necessarily simultaneously as a cross).
notes:
- If there was a correction on any currency (i.e. it alone changed the price significantly), it should be excluded from consideration, and it is better to skip the day altogether. If you look at the bundle, you can see it very well - everyone goes very synchronously (so synchronously that it is a mystery how low correlations are obtained), during correction the currency moves quickly and without fail across everyone, and it is as if they do not notice.
- if there were serious economic/political news in the evening and there is no decrease in activity by 19, it is also not unreasonable to rest for a day.
- It is absolutely impossible to hold a position for a long time. Even if it is in profit. A day or two at the most. And two days, if the selected currencies were going nostril-to-nostril. There are objective justifications for this, I'll give them later on
the principle can be considered working
on the screenshot, on the top the divergence of yesterday's AUD,GBP and on the bottom the day before yesterday's NZD,JPY and the overall result.
can be closed :-)
---
brief re-description:
- look at the divergence (in % or log.scale as in the screens) of currencies for the trading day. From the moment of minimum daily volatility/volume to the beginning of the overnight flat. Taking into account the MQ demo timezone it turned out to be from 3 to 19
- we choose two that are maximally separated, i.e. are the boundaries of the bundle. We will trade them, the one on the top to sell, the one on the bottom to buy.
- (optional, but recommended) during a flat period we expect minimum drift/ fluctuation in continuation of the previous movement.
- (also optional) it is possible to trade not the cross, but individual majors with USD. Depending on the direction of the daily divergence. If both currencies were going up (and the whole bundle is directed upwards), then only the upper one is sold. Down symmetrically. When one is up, the other is down, then both majors and close them at different moments (not necessarily simultaneously as a cross).
notes:
- if there was a correction on any currency (i.e. it alone changed the price significantly), it should be excluded from consideration, and it is better to skip the day altogether. If you look at the bundle it is clearly visible - all go very synchronously (so synchronously that how low correlations are obtained is a mystery in general), in case of correction the currency moves quickly and without fail across everyone, and they as if they do not notice.
- If there was serious economic/political news in the evening and by 19 pm there is no decrease in activity, it is not bad to rest for a day either
- It is absolutely impossible to hold a position for a long time. Even if it is in profit. A day or two at the most. And two days, if the selected currencies were going nostril-to-nostril. There are objective reasons for this, I'll give you some more details later
Super! TC - working!!!!!
Come on! To yourself! ) Cool!"!"! It is possible to be in touch and to address still can for clarification TC!!!?????
Where it is a joke and where it is not....).
you can formula how to optimally count relative movements of currencies, somewhere there was and did - still in the archive ... and update the data like the beginning of the bundle at 24 hours? This was discussed with Alexander in another thread.....
there is something like current price - price at the beginning of the interval of each currency and lot (coeff) of its weight must be linked here, like from the cost of tick and volatility.....
The relative movement of a currency pair, expressed in pips, is calculated as the difference between the closing price and the opening price multiplied by the appropriate coefficient depending on the currency pair and lot size. In general, the formula looks like this:
Movement in pips = (Closing price - Opening price) * Coefficient
Where:
- Closing price - the price at which the trade was closed (or the current price if the trade is not closed).
- Opening Price - the price at which the trade was opened .
- Coefficient - a number, depending on the currency pair and lot size, that converts the price difference into pips .
Coefficient Calculation:
The coefficient depends on the currency pair and lot size. For most currency pairs with the US dollar as the quoted currency (e.g. EUR/USD, GBP/USD), the coefficient is 10000, as one pip is usually equal to 0.0001. For pairs where the quoted currency is the Japanese Yen (e.g. USD/JPY), the coefficient is 100, as one pip is usually equal to 0.01.
Example calculation:
Suppose you opened a EUR/USD trade at 1.1000 and closed it at 1.1050. The lot size is 100,000 units of the base currency (EUR). In this case, the quoted currency is USD.
1. price difference: 1.1050 - 1.1000 = 0.0050
2. Coefficient: 10000 (for EUR/USD )
3. Movement in pips: 0.0050 * 10000 = 50 pips.
----
divergence (in % or log scale - I don't know how to do log scale - I can do it in per cent... if anything, please suggest the best option using the formula!!!! sps
you can formula how to optimally count relative movements of currencies, somewhere there was and did - still in the archive ... and update the data like the beginning of the bundle at 24 hours? This was discussed with Alexander in another thread.....
there is something like current price - price at the beginning of the interval of each currency and lot (coeff) of its weight must be connected here, like from tick cost and volatility.....
The relative movement of a currency pair, expressed in points, is calculated as the difference between the closing price and the opening price multiplied by an appropriate coefficient depending on the currency pair and lot size. In general, the formula looks like this:
Movement in pips = (Closing price - Opening price) * Coefficient
I will try to decipher the question and give an answer :-)
1. I have written in black and white several times - the beginning of the control bundle from the minimum of the daily volume (it is approximately from 3 o'clock at night) and to the beginning of the night flat, it is up to 19 o'clock. That is, we look at how the currencies separated during the active trading day
2. count movements in % or logarithms of prices (movement = log(total_price_to_usd) - log(initial_price_to_usd) , instead of usd you can take any other).
In them all currencies move proportionally. Points are the tenth and unimportant thing, you (and everyone) are interested in how many times, by how many per cent, the capital will increase if you invest in this or that currency. And not the points
3. no coefficients when switching to log scale or percentages, lots do not play a role. 1 lot of gold rises/falls by the same percentage as 10 lots.
3. no coefficients. When switching to log scale or percentages, lots do not play a role. 1 lot of gold rises/falls by the same percentage as 10 lots.
oops.
yes - that's a yes. I had a sneaking suspicion..... it's relative, yeah.... )
I confused the ghost into a single coordinate system... in general with some other calculation... spreads....
(movement = log(total_price_to_usd - this apparently means the current price at the time of construction and consideration of the bundle of each currency pair, i.e. the price at the current time?
(movement = log(total_price_to_usd - this apparently means the current price at the time the index is constructed?
Yyyyyy
for example:
1. take USDCHF price at the moment of 3:00. We convert it to CHFUSD=1.0/USDCHF.
2. we take the logarithm (so that the numbers are not too small, I use log2, although it is not crucial) : CHFUSD_begin = MathLog(CHFUSD)/MathLog(2)
3. also get the logarithm of the price at the end of the day, i.e. at 19:00. CHFUSD_end
4. CHF movement for the trading day, CHF_move=CHFUSD_end - CHFUSD_begin
repeat for all majors.
choose the pair with the largest and the smallest XXX_move - this is the pair that has diverged during the trading day.
Yippee
For example:
1. Take the USDCHF price at the moment 3:00. Bring it to the relative dollar CHFUSD=1.0/USDCHF
2. take the logarithm (so that the numbers are not too small, I use log2, although it is not essential) : CHFUSD_begin = MathLog(CHFUSD)/MathLog(2)
3. also get the logarithm of the price at the end of the day, i.e. at 19:00. CHFUSD_end
4. CHF movement for the trading day, CHF_move=CHFUSD_end - CHFUSD_begin
repeat for all majors.
select the pair with the largest and smallest XXX_move - this is the pair that has diverged during the trading day.
Oh, that's it! )
I've got it now... thanks I will implement it in my code!!! thanks again to you very much!!!!!!
if there is anything else there for a while, I will ask for an explanation..... thanks )
I get it - but not immediately - I'm getting the hang of it....)
I am in the subject - absolutely immersed and trading......
I also wanted to clarify (if there are magic secrets, if not difficult - you can write in a private message), what I remembered - such background to your model can be viewed, only not to interfere, if it will interfere, and do not need to.....
Here I have asked in google (I remembered the formula which was used earlier in different threads on spreads), I just need your opinion:
Yes, the formula for calculating the relative movement of a currency pair, as you suggested, is correct and corresponds to the rationing formula. It expresses the change in the price of a currency pair relative to the initial value and shows the relative change in per cent.
The formula is as follows:
Relative change (%) = ((Price at the current time - Price at the beginning of the period) / Price at the beginning of the period) * 100%
Where:
- Price at current time
- is the price of the currency pair at the moment.
- Price at the beginning of the period
- is the price of the currency pair at the beginning of the period you are interested in (for example, at the beginning of the day, week, month, etc.).
This formula allows you to evaluate the dynamics of the price of a currency pair, regardless of the initial price, and express it as a percentage, which is convenient for comparing relative changes between different currency pairs or periods.
---
Immediately - no jokes here and please do not consider it a mauveton to ask for clarification from the chrome AI...)
Where:
- Price at current time
- is the price of the currency pair at the moment.
- Price at the beginning of the period
- is the price of the currency pair at the beginning of the period you are interested in (for example, at the beginning of the day, week, month, etc.).
This formula allows you to evaluate the dynamics of the price change of a currency pair, regardless of the initial price, and express it as a percentage, which is convenient for comparing relative changes between different currency pairs or periods.
---
Immediately - no jokes here and please don't consider it a mauveton to ask for clarification from the chrome AI...)
Seriously, I don't understand what needs to be clarified...
scales in %% and logarithms are the same thing. Log. it's easier to calculate just MathLog, less confusion.
you can do both. It is absolutely the same thing, just think that the words are different.
As a more complex alternative you can take relative weights of currencies, I gave the formula above in the discussion. But there will be no fundamental difference
The usual y=MathLog(x) is simpler, there will definitely be no confusion with it