Pair trading and multicurrency arbitrage. The showdown. - page 103

 
Maxim Kuznetsov #:

Suppose that we invest in one basket, i.e. we buy a weighted basket of currencies. After time T the price of the basket will change and the balance will be disturbed.

Yes, it should be tracked and possibly predicted

Maxim Kuznetsov #:

Re-balancing is again done in three ways: 1) add the missing in proportions, 2) redistribute the existing 3) remove the surplus in proportions

In the classical portfolio, everything is done as in point 2)

there is an allocated amount of say 100k and everything is redistributed within this amount.

Maxim Kuznetsov #:

and reducing the list in the basket to a minimum of 3, you can get the algorithm "im.Renat" ; but 3 will have to sit in ambush for a long time.

Is there a single fact that proves that his system exists?

 
Maxim Kuznetsov trading baskets, in the second option, the total basket will have balances/shortages and they will not be balanced (regardless of the method).

suppose there is a market and baskets are traded, then

a) change in the exchange rate of e.g. USD to USD

1 It is wrong to open and close a basket at once. Orders are opened and closed by signals of the algorithm when conditions are met.
Even in five days you can not close the basket. There will be a signal - then it will be opened.
2 When an order is closed, the counter trade is not always opened. It is necessary to use 2 timeframes - optimally major 240 and minor 15
3. Take profit and stop loss should not be put or put it only for a long distance t=10080; t=43200; (remember the CHF sweep in 2014?)
4. If there are no orders in the basket, it does not mean that the basket is incomplete. Usually this does not happen - there is always a normal number of orders in the work. An order is closed, it is opened.
Martingale can also be used, but with closing on a signal if it is against the trend. Martin is rarely opened if the algorithm does not "rattle".
Indicator rattling is treated by algorithmic formulas with floating dynamic coefficients.
5. There is no sense to determine the fastest currency - the momentum of the whole market is calculated, the bets that run faster are taken,
all orders are opened with the same risk, and therefore with different lots.

 
mytarmailS #:

Yes, it needs to be monitored and possibly predicted

In the classical portfolio, I think everything is done as in point 2)

There is an allocated amount of say 100k and everything is redistributed within that amount.

Is there a single fact that proves that his system exists?

A portfolio, not quite just by reallocating within it. The price of the portfolio has increased by 1000 usd, part of it is withdrawn, part of it is redistributed to keep the balance (and something goes to expense/spread/commissions). In the end you have 800 in your pocket and the same 100k in a balanced portfolio.

Otherwise balancing will be at a loss: initial portfolio 100k growth by 1% = 101k and we rebalance. Then a 1% drop and we are at a disadvantage from the initial value.

 

Almost an algorithm:

1. we create a basket based on days and ln(x)

2. when the basket price deviates upwards, we withdraw a part of the leader's volume in its national session (at the maximum intraday volatility)

3. when the price deviates downwards, we add to the least laggard outside its national session (at the minimum intraday volatility).

By "deviation" we mean a statistical deviation, not just a price change.

it's just a small matter of calculating the basket, deviation limits and volumes of deposits/withdrawals :-) and a little bit of programming.

---

PS/ if the algorithm is more or less working, it will also show how money is withdrawn from national economies in the USA.

In national sessions there is not only speculative part in the total volume and it also falls into withdrawals, and replenishments take place outside the sessions from speculative money.

PPS/ and if replenishments take place in the evening, and alternate following the sun, they collect more than one (up to 2.5) discount rate per day. Money overflows in a wide stream

 
Maxim Kuznetsov #:
The money's flowing in.
That's it, pick your yachts!
😂
 
Sergey Gridnev #:
That's it, pick your yachts!
😂

no way...this is more of an alg.of very rich uncles(and aunts).

1) we are not the recipients of the bet but on the contrary minus their difference 2) we have no insight, no influence on events, no technical capabilities and will be the last to enter/rebalance. 3) we have high profit percentage requirements and commissions too. 4) high leverage does not allow for long trades

Our way is to find the sharpest movements and take them with maximum risk.

For us, alg. is more of an academic nature.

 
Maxim Kuznetsov #:
create a basket on the basis of days

1. there should be 4 sub-baskets in the basket!
if you don't observe condition #1 - the deposit will plum, fast or slow - it doesn't matter.
sub-baskets are divided into 4 types
gap1,gap2, cross1, cross2 - (compression/stretching and width of the channel how much more 100%/less than 100%) wrote a couple of pages ago about it.
national sessions - to put it mildly to the back, you should take when the signal, it can be in Tokyo, or in the afternoon it does not matter.
2.. in baskets there is a differential! that to buy weak or to buy strong, to sell weak or to sell strong -
it is necessary to do it when there are certain conditions for it, but it is correct to do it specifically depending on what movement. There should be 4 baskets.
If you will not change the conditions for opening orders depending on gap1, gap2, cross1, cross2 - then the depo drain in a multi-trade basket is ensured.

 
Maxim Kuznetsov #:

No way...it's more like the alg. of very rich uncles and aunts.

Our way is to find the sharpest moves and take them with maximum risk.

That is, we'll be draining as before.



 
There is also a strategy for very lazy traders, in the algo-trading mode. The theme is as follows: you trade the basket from the maximum width of the channel - from the point of reversal to the maximum compression. For the timeframe H4, it is somewhere one or two weeks to move.
At first you hammer in one side on channel compression, when it is equalised at 100% width, then you trade in both sides with the use of martin, but not evil martin, and it is opened only on the repeated signal. You use trailing stop and takekporofit, and at the maximum compression of the channel, you take a viprigit or sit further. There's no stop, no stop, just a take. The equity grows faster than the drawdown. By the end of the second week with a half-locked basket you come to the centre of the channel, and then you stop trading and sit until the channel expands to 100% again. Here on the screen is the result for today.
 
Alexander Pryakha #:
There is also a strategy for very lazy traders, in the algo-trading mode. The theme is as follows: you trade the basket from the maximum width of the channel - from the point of reversal to the maximum compression. For the timeframe H4, it is somewhere one or two weeks to move.
At first you hammer in one side on channel compression, when it is equalised at 100% width, then you trade in both sides with the use of martin, but not evil martin, and it is opened only on the repeated signal. You use trailing stop and takekporofit, and at the maximum compression of the channel, you take a viprigit or sit further. There's no stop, no stop, just a take. The equity grows faster than the drawdown. By the end of the second week with a half-locked basket you come to the centre of the channel, and then you stop trading and sit until the channel expands to 100% again. Here on the screen is the result for today.

)))

serious statistics