Arbusers

Hi,

I am working on a Swap arbitrage system that is working on multiple CFD providers, mostly forex brokers. I am wondering if someone is doing this aswell and/or if you have any tips about improving the profitability.

This system takes advantage of the different swaprates (intrest) that brokers take/give on open positions, if held overnight. On some postitions you will receive a positive swap when you are shorting the asset and paying if you are longing the asset. But there are differences between brokers, some are paying a high amount compared to some that will have quite low swaps. If you take a long and short position you won't have any price gain or loss. This way you will, after commission and spread, only get a positive result over the received swaps.

An example:
Broker A:
EUR/USD: longswap: 2 pip, shortswap: -3 pip.

Broker B:
EUR/USD: longswap: 1 pip, shortswap: -1 pip.

If you trade 1 lot (100k, with leverage), a movement of 0,0001 will be around 10 euros. This 0,0001 is called a pip.

So if you are taking a long position on broker A you will receive 20 euros, and then taking a short position on broker B you will have to pay 10 euros to keep the position open. This way you will have a 10 euro arbitrage. On some forex pairs you are able to make between 3-10% each month if the positions are kept open.

Risks
There are a couple of risks involved:
- Counterparty risk: Mainly about the brokers that do not allow this. They might close your account, make the trading conditions unfavorable.
- Liquidity: When a stoploss is hit/ a position is liquidated you immidiatly have to close the opposite trade. Liquidity might make this difficult.

My version on the strategy
The brokers have a hard time to intercept such trades because a position looks like a swingtrade. From what I heard was that the brokers do not like this strategy because of all the rebalancing of your accounts. They are paying a fee on the withdrawels, and if you only trade once or twice you won't make them enough in commission and spread to cover these transaction fee's. This way I want to hide my trades and use another system that trades frequently, just to "give" them some commission. Since the commisions are lineaer; I want to put in 5-10 smaller orders spreadout over a couple of days instead of one big bulk order. This to make the detection a bit harder without extra costs.

Since the volitility plays a big part of this strategy you need to adjust your position size to the volitility. In a dream senario you will put as much leverage on these trades as possible to maximize your position size. But this means that you will be liquidated quickly. I am currently using the expected value of x-percentage stoploss. I am doing this by calculating the runtime of a trade at multiple stoploss percentages for every daily closing price. This way I get the average runtime for a certain stoploss percentage, and with this you can calculate the expected return per stoploss percentage.

I am planning to use only regulated brokers, this means that I will get less return but my funds are somewhat safe. There will be probably 5-10 brokers connected since only a couple of brokers will give a positive total swapvalue. I am working on a way to make this somewhat of a dynamic system with all the brokers. For example:

You have three brokers: A, B and C.
Brokers AB, AC and BC have trades running. But the trades on A are losing, so the cummulative loss on A is 100. But on B and C the trades are both 50 in profit. Since the system trades lots of small positions the trades on A will be stopped out, so there is a surplus of money on B and C. Normally you would withdraw 50 on B&C and deposit this 100 on A. But this causes lots of small transactions (withdrawels and deposits) that the brokers dislike. So the system will search for a trade that can be done between B and C (with a minimum treshold of returns because the system does not need to trade when the expected returns are below 3% (or whatever you like). By doing it this way I might be able (still testing the basics) to reduce the amount of withdrawels/deposits by a significant amount.

Combining the small transactions instead of one bulk order and using a small gridsystem to make a lot of transactions will hopefully reduce the chance of a broker closing the accounts.


I am completly basing this system on some basic expected values from historical data. I am wondering if there are any improvement possible to maximize the returns. I hope you can give me some inputs on this.

Daan