Arbitrage system forward test

[Znail]

Not sure why you assumed that I ever bothered to run a arbitrage based system in a live environment -- and if I did, it wouldn't be on Windows, in fact it wouldn't be running in software at all. Likewise you shouldn't assume running on a Linux VPS is low latency. I'll repeat what I said before: It is not impossible to make money on arbitrage. The problem is that the majority of people who come on this forum and talk about arbitrage think 10ms is fast. Most of the advice given on this forum, in general, should be implicitly assumed it is directed towards a typical retailer who lacks the resources (both technical and financial) to compete at the scale required to do arbitrage in a competitive market. If you are doing this against a counter-party forex broker, then you are not doing arbitrage in a competitive market. In this case, you don't have to be fast, you just have to be faster than the broker.

I don't know the details of your infrastructure or the design of your system (I don't even know if it's hardware or software based), but this was your intention -- to be vague. For all I know you have $100B backing and have designed a system that will respond in < 100nano seconds. I haven't the slightest clue what the time scales that the top competition in arbitrage is running at these days because I don't work in finance, but I'd guess it's around 100ns on a pure arbitrage system in equities markets.

I originally wrote a lot about my background, but I removed it. Lets just say I know latency more than most. Most people on this forum who talk about low latency probably couldn't design a system that emits an order < 10ms after receiving a quote.

It's not about what I know, my caution was with regards to what you know since you were so vague it's hard to tell. You gave absolutely no details, so I suggested you be cautious and don't get too excited because of the following:

1) I do truly understand how fierce the competition is in arbitrage and I also recognize that the vast majority of people to even bother posting questions about it here are not likely to understand it to the degree that I do.
2) 6 days is too small a sample size to definitively know why you had profit. It is absolutely plausible that you got lucky with prices moving in your favor when you executed your trades and therefore that it's a statistical anomaly.
3) you may be using a counter-party forex broker that has a shitty market making algorithm. If this is the case, you're not actually doing arbitrage in a competitive environment. You've simply found a forex broker with a bad market making algorithm and you're exploiting it -- you're not competing against other arbitragers, but instead against the broker. In this case, you're not fast, your broker is just slow.
4) if you're using a broker as described in 3), you have a very real possibility of them preventing withdrawal or "fixing" their system so that you can't do this anymore.

What I'd like to know is:
1) is your forex broker your counter-party?
2) what is your system's latency from last byte of quote received from the network (hardware level) to last byte of order leaving the network (hardware level)?

Hi, Lee , Thanks for your informative reply, to cut thing short, Hedge-Fund using code base progress, nano seconds execution, 6 digits quote, and fast computer, to deal with LPs . For retail level, I admit that not all brokers are Arb able because they are (most of them are SCAM). As you know that the most effective strategy in financial market is the latency trade and even faster these days, but brokers prevent their pockets on B-book by just trading against you. Also, LPs and Aggregators plug-in your executions on prevent their pocket. So at retail level to make ARB to work, one should read client agreement of regulated broker before commence anything else, and make sure that you don't do strategy against agreement. Then with a little application of Hedge-Fund methodology to trade them. We will file lawsuit to their regulator to shut them down or until they pay us, if they refuse to pay.

Brokers has been robbing traders for a long time, name it, B-book, plug-in, spread widening, slippages, etc. How long are we going to let this keep going.

 

[Znail]

I don't want your precise strategy, just a rough idea of what you are doing. That will give me some idea of whether your type of strategy really is do-able (which is what you say you want to prove).

You posted a .jpg that could be mocked up in 5 minutes and proves nothing. Plus even if it was genuine the results wouldn't be statistically significant. As you say plenty of research and all common sense tells us that significant pure arbitrage opportunities are incredibly rare and do not persist for very long; you'd need some pretty strong evidence to back up a claim otherwise.

GAT

Hi GAT, the best way to demonstrate the test is to post my fx book with verified account and track record, but you should know why i will not do that. I will say it again, my intention here is only give market idea IF ARB DO-ABLE, that's all. I will keep update the same way, fake or not please use own judgement and be patient.
Plus, ARB opportunity is not as rare as you think, trust me. There are many fake opportunity, most old methods
take all opportunities and failed. you need to select them carefully.

 

[Znail]

Dear all,
After many researches, everyone insisted that Arbitrage is impossible. So I would like to open a topic to run Arbitrage system forward test and see if the result will keep on the growth (and see how long it will keep rolling) ?
So a permission to share my live account results;

day: 6
starting deposit: 995$


P.S. this is not an advertise thread, I will not give out the name of system or broker or account information. I only want to prove if Arbitrage is do-able these days.

Starting Test Date : 18 APR 17

 

[Znail]

Hello Znail,

Welcome to ET. Thanks for sharing your forward testing with the Arbitrage system.

May I ask a few questions as I am also beginner programmer with NinjaTrader learning to program and properly test automated trading systems before going live with my own money.

1. What does Arbitrage system mean? Can you please explain to me what this means?

Thanks

Hi SimpleMeLike, definition of ARB is easily found on internet, the problem is how to implement it in real world. Some think that it just buy/sell assets knowing that price will rise/fall (At institution grade, e.g. brokers, you can do that to trade against client on B-book), but at retail level it needs multiple strategies to make it happens.

I am not claiming myself professional, but when it get to coding, please concern these factors;
1. efficiency of coding language, some easily code but slow execution, some are fast without platform to support.
2. Some platform offers institution grade interface, but lower standard when comes to execution.
3. all depend on your strategy to work comfortably with which coding language.

 

[truetype]

nano seconds execution

When someone claims his tick-to-trade is in the nanos, there's no point reading the rest of his posts.

 

[Lee-]

When someone claims his tick-to-trade is in the nanos, there's no point reading the rest of his posts.

Well, I sort of disagree with you. I could provide an arbitrage solution that is sub microsecond measured from last bit of the frame that contains the quote message to first bit of the frame that contains the order, but this would be a hardware based solution. In general, I agree with you though. He basically said he's using software and probably has no concept of the issues that arise when dealing with nano (or even microsecond) scale such as inaccuracies in software based methods for measuring latency, the impact of the scheduler, or that context switches are a bitch.

I don't doubt that there are counter-party forex brokers out there that have bad market making algorithms that can be exploited in various ways (including arbitrage). If OP found one, good for him!

 

[SimpleMeLike]

Hi SimpleMeLike, definition of ARB is easily found on internet, the problem is how to implement it in real world. Some think that it just buy/sell assets knowing that price will rise/fall (At institution grade, e.g. brokers, you can do that to trade against client on B-book), but at retail level it needs multiple strategies to make it happens.

I am not claiming myself professional, but when it get to coding, please concern these factors;
1. efficiency of coding language, some easily code but slow execution, some are fast without platform to support.
2. Some platform offers institution grade interface, but lower standard when comes to execution.
3. all depend on your strategy to work comfortably with which coding language.

Thank you Znail for response,

I use NinjaTrader8 to program and back test different strategies ideas for me to make money. I only see to trade 1-3 contracts.

 

[Znail]

When someone claims his tick-to-trade is in the nanos, there's no point reading the rest of his posts.

Hi Truetype, please kindly carefully requote in full phrase, and do not leave out important message that will distort meaning. "Hedge-Fund using code base progress, nano seconds execution, 6 digits quote, and fast computer, to deal with LPs" and "Then with a little application of Hedge-Fund methodology to trade them".
I would not be here posting here if I am Hedge Fund.

 

[Znail]

Thank you Znail for response,

I use NinjaTrader8 to program and back test different strategies ideas for me to make money. I only see to trade 1-3 contracts.

Hi SimpleMelike, from my experience with Ninja platform, coding them is not a hard task, the hard bit is speed of people who process the code outputs. As I say the platform is superb, but broker who use them slow their performance by process not platform structure. One reason I can think of is that Ninja provide many market information that crucial and important to trading strategy, so broker might has to put little magic to the executions and you know why.

 

[Znail]

Starting Test Date : 18 APR 17