My Own Broker Arbitrage | Page 6 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Arbitrage with no-interest broker 147 replies

eur/usd arbitrage broker? 4 replies

does you broker ban arbitrage on manipulated price quotes? 2 replies

My idea for broker arbitrage 13 replies

Broker arbitrage BIG TIME 4 replies

Attachments: My Own Broker Arbitrage
Exit Attachments

My Own Broker Arbitrage

  • #101
  • Dec 24, 2011 12:37am Dec 24, 2011 12:37am
"sockets from MQL?" from terminal to terminal I assume. But please tell me you're not talking about named pipes. Are you?

Quoting ianj1
I'll try and dig some code out to use sockets from MQL if you want - there are particular threading concerns within MT4 - especially if you want to emit a feed AND respond to order processing requests in a timely manner. I don't have any reusable code OUTSIDE of Mt4 - my stuff is all Java and part of a different system.

Of course this might sound like teaching a grandmother to suck eggs ..
  • #102
  • Dec 24, 2011 4:17am Dec 24, 2011 4:17am
I do not understand what you are looking to do here?

Why not simply have a prime broker and make the trade-off between the provider of liquidity? (Lava / Currenex / Hotspotfxi / EBS /etc etc)?
  • #103
  • Dec 24, 2011 9:54am Dec 24, 2011 9:54am
Quoting surfeur
I do not understand what you are looking to do here?

Why not simply have a prime broker and make the trade-off between the provider of liquidity? (Lava / Currenex / Hotspotfxi / EBS /etc etc)?
Capital, or relatively lack thereof.
  • #104
  • Dec 24, 2011 12:12pm Dec 24, 2011 12:12pm
Quoting Ronald Raygun
Capital, or relatively lack thereof.
U need 20K$ for londoncapital..
  • #105
  • Dec 24, 2011 4:40pm Dec 24, 2011 4:40pm
Quoting ianj1
My assumption is that nearly all BROKERS (certainly retail MT4 brokers) will last look - its the ones that are slow that are the problem (most people don't realise that 90% of the prices on Currenex are last looked and even the last bastion of institutional NON last look brokers - Hotspot - is suspected to offer limited last looking to some privileged liquidity providers) Of course there are techniques to reduce the likelyhood of a last look, but they will not work on an MT4 broker (in fact they will not work in most cases unless you are talking...
if you were to deal with an aggregrate pool of prices amongst mutilple liquidity providers through say ebs hotspot and lava why would you worry about last look? Most likely your market maker will be different across both or all trades so why would the bank cancel the trade theough last look if they are not exposed to an arb based solely on their own pricing? The time i see lastlook coming into play is if there is an arb based on one provider and he chooses not to respect that order as it exposes the provider. Anyone shed light on this? Also arbing across retail brokers your not only fighting broker spread markup and manipulation but also higher latency as each broker has to send the order to its liquidity provider. If thhe broker is not an ecn they will prob ban you since they wont allow arbing
  • #106
  • Dec 24, 2011 4:45pm Dec 24, 2011 4:45pm
Quoting sidhujag
if you were to deal with an aggregrate pool of prices amongst mutilple liquidity providers through say ebs hotspot and lava why would you worry about last look? Most likely your market maker will be different across both or all trades so why would the bank cancel the trade theough last look if they are not exposed to an arb based solely on their own pricing? The time i see lastlook coming into play is if there is an arb based on one provider and he chooses not to respect that order as it exposes the provider. Anyone shed light on this? Also arbing...
What the broker doesn't allow is time-shifting. That is to say, using a faster feed to predict a slower feed.

True arbitrage should not affect the broker provided they are truly a broker in that they pass through your orders.
  • #107
  • Dec 24, 2011 5:28pm Dec 24, 2011 5:28pm
Quoting Ronald Raygun
What the broker doesn't allow is time-shifting. That is to say, using a faster feed to predict a slower feed.

True arbitrage should not affect the broker provided they are truly a broker in that they pass through your orders.

yup thats the problem, not many true brokers. If they hedge your position and they find out you are arbing they may ban you. That may be an urban legend.

You are still fighting spread markup on 3rd tier brokers.

You also have the backend management to deal with. There are brokers out there now that offer 2nd tier liquidity with DMA access and backend support. Im sure that some of the feeds offer last look and it is used, but that is a mute point as it will happen across any broker aswell. This will allow tighter spreads, but at the cost of higher capital startup (nothing comes for free). Also a note about trading thru the FIX api through the different exchanges like HOTSPOT and LAVA and EBS is that I heard there is a server induced latency of 10 or 15 ms.. not sure why, as arbing should create a more efficient market so why not? Probably the big boys trying to create a monopoly on the strategy. Proof of this and you can really make some money in court

I heard ProTrader allows a dual broker arb, the type of arb you are looking for. Its all built in, and you can connect Dukascopy and say MBTrading feeds and let it run all day. It might be a marketing strategy to sway traders away from Mt5 but def worth a try.

I myself am more intrested in arbs with more than one leg, if you can do a simple arb, why not use FPI (fractional price inefficiency) to have much more arbs, and they usually last longer? Even the effects of the spread are negated by the quality of arbs that come up. Ofcourse you have to hope that your trading protocol allows you to trade with fractional lot sizes which itself is another problem in that strategy lol.

Why is it always so hard?
  • #108
  • Dec 24, 2011 6:03pm Dec 24, 2011 6:03pm
Quoting Ronald Raygun
Capital, or relatively lack thereof.
Would a UK based member capital pool allow you to overcome the investment size issue and make the job any easier? We may have
the method to do this....not multi-millions but say $25K to $100K or a
bit more?

Atlas1
  • #109
  • Dec 24, 2011 7:49pm Dec 24, 2011 7:49pm
Quoting Atlas1
Would a UK based member capital pool allow you to overcome the investment size issue and make the job any easier? We may have
the method to do this....not multi-millions but say $25K to $100K or a
bit more?

Atlas1
It's a start. But before even approaching that, I need to get this prototype working

My main problem is making the shared pipes reliable. I haven't quite figured that out yet.

Someone privately mentioned to me putting all the slave terminals in the expertsfiles folder of the master terminal, so it's just a matter of navigating to the files folder on the slaves. I'm using that as an absolute last resort, because then I'd have to worry about lock files and other nonsense.
  • #110
  • Dec 24, 2011 10:51pm Dec 24, 2011 10:51pm
Quoting nondisclosure00
"sockets from MQL?" from terminal to terminal I assume. But please tell me you're not talking about named pipes. Are you?

I suppose you could open sockets between terminals but this would not offer much flexibility with multiple brokers and might run into much more complex state management of messaging/threading

I am suggesting using a central server that arbitrates between an arbitrary number of MT4 instances. I currently use such a system to make N:M copying between MT4 instance servers possible (and can include Mt4 brokers if suitable adapters are written)

The threading/asynchronous issues can then be dealt with independently of the fragile Mt4 threading model

My experience is in Java - but you could equally write it in a weapon of your choice (e.g. dotnet, python etc)


Real network sockets from MT4 is not a problem - you can locate the server anywhere reachable
  • #111
  • Edited Dec 25, 2011 5:58am Dec 24, 2011 11:00pm | Edited Dec 25, 2011 5:58am
Quoting surfeur
I do not understand what you are looking to do here?

Why not simply have a prime broker and make the trade-off between the provider of liquidity? (Lava / Currenex / Hotspotfxi / EBS /etc etc)?
I think because:
1. That is beyond the reach of most retail traders - Have you tried to open a real prime broker relationship ? Have you tried getting access to EBS/Reuters ? (Come back when you have) I believe the key (perhaps i misunderstood) unwritten constraint here is RETAIL brokers - please correct me if i got it wrong
2. MT4 brokers do sometimes publish 'attractive' prices - the question is whether they are hit-able (slippage on market orders)

And then of course the question is how long you will last on institutional platforms - i can count the days your accounts will be liquid on about three hands. The account relationships with a prime brokerage solution are much more personal - you can't hind 'under the radar' in the same way you MIGHT be able to do in retail
  • #112
  • Dec 24, 2011 11:14pm Dec 24, 2011 11:14pm
Quoting sidhujag
if you were to deal with an aggregrate pool of prices amongst mutilple liquidity providers through say ebs hotspot and lava why would you worry about last look? Most likely your market maker will be different across both or all trades so why would the bank cancel the trade theough last look if they are not exposed to an arb based solely on their own pricing? The time i see lastlook coming into play is if there is an arb based on one provider and he chooses not to respect that order as it exposes the provider. Anyone shed light on this? Also arbing...
The main problem is 'souring' latency, not cancellations (but they are sometimes a problem that must be settled with a 'gentlemen agreement' where goliath usually wins on a dispute over an out trade) - its like picking pennies off a steam roller drum - if you miss your exit - you can be squished! Most instututional arb platforms have a min size of 10 lots or so - arbing 10-50 lots is normal. If a last looker does not like your price and 'sits' on it for 500-1000ms you could be screwed (try it on USD/ZAR or USD/MXN) On odd occasions a participant might DECLINE the trade and it will return to the matching pool - if another participant also sours the trade you could be looking at seconds

You have two basic choices in simple arbing - Wait for leg1 to complete before starting leg2, or execute leg1 & 2 together

* With market orders you might as well execute both together
* With limit orders you have to be careful because either leg1 or leg2 could 'miss' and you could be left with an uncomfortable position. Last look on leg2 is poison as you are then chasing to cover position
  • #113
  • Dec 24, 2011 11:22pm Dec 24, 2011 11:22pm
Quoting surfeur
U need 20K$ for londoncapital..
It is difficult to arb effectively with $20k in institutional. $50k-$100k min is required to get them to give you access to multiple platforms - and that will NOT include EBS/Reuters

Prime Broker relationships offer much lower leverage - 15-30:1 is the 'norm'

HotspotFxi in particular has a min match specifier on posted orders. Many market makers will specify a min size of 5 lots which means you may not match if you are under 5 lots, even if the price is good - so anything under 5 lots and your odds start sliding
  • #114
  • Dec 24, 2011 11:38pm Dec 24, 2011 11:38pm
Quoting Ronald Raygun
It's a start. But before even approaching that, I need to get this prototype working

My main problem is making the shared pipes reliable. I haven't quite figured that out yet.

Someone privately mentioned to me putting all the slave terminals in the expertsfiles folder of the master terminal, so it's just a matter of navigating to the files folder on the slaves. I'm using that as an absolute last resort, because then I'd have to worry about lock files and other nonsense.
I don't think you gain any benefit (other than for simple display), and have lots of minuses in running the master script within MT4. A simple console application functioning as the master would seem the way to overcome much/all of your difficulties. In fact, the more processing you can remove from MT4 the more control over the outcome and performance you should have, including multithreading.

edit: I notice Ian said the same thing basically.
  • #115
  • Dec 24, 2011 11:56pm Dec 24, 2011 11:56pm
Quoting sidhujag
Also a note about trading thru the FIX api through the different exchanges like HOTSPOT and LAVA and EBS is that I heard there is a server induced latency of 10 or 15 ms.. not sure why, as arbing should create a more efficient market so why not?
(
Coz the banks don't like being dumped with 'hot potatoes' for being 3ms too slow. The brokers are generally fine with it, but if they do not tow the line they will not get the bank liquidity and their spreads will suffer accordingly.

Price feeds are conflated (Combining a number of price updates together and emitting them in batch, say every 50-500ms) and sometimes intentionally delayed for 'other' reasons, but i am not aware of core matching induced latency, but liquidity provider 'induced' latency - almost certainly. Of course if am not speaking of retail systems again (sorry - i should stick to context)

Quoting sidhujag
I myself am more intrested in arbs with more than one leg, if you can do a simple arb, why not use FPI (fractional price inefficiency) to have much more arbs, and they usually last longer? Even the effects of the spread are negated by the quality of arbs that come up. Ofcourse you have to hope that your trading protocol allows you to trade with fractional lot sizes which itself is another problem in that strategy lol.
(
I would have thought that was minimised with 5 digit price feeds (but then are you considering 4 digit feeds as well?) I guess you are on about opening, say, a 3 leg triangular arb when 1 leg is out of alignment - and then closing when out of alignment in the opposite direction ? I would agree that it is merely a variation on a theme and slightly more likely to work if you can overcome the cost of the third leg, but most of the technology required is the same. On the fractional lot size requirement you only have to support 'normal' (leg 1-3) trades approximately - you don't need to trade an EXACT triangle - any differences can be picked up in a small 4th leg that is far less critical in price.

So as long as your leg 1-3 is 10-20+ times larger than the min lot size the costs are manageable
  • #116
  • Dec 25, 2011 12:08am Dec 25, 2011 12:08am
Quoting ianj1
And then of course the question is how long you will last on REAL platforms - i can count the days your accounts will be liquid on about three hands. The account relationships with a prime brokerage solution are much more personal - you can't hind 'under the radar' in the same way you MIGHT be able to do in retail
To the last point, this is the reason I believe that speed alone with the straight arb isn't the answer to long term success. I'm working on the premise that an algorithmic stat arb approach is required to stick around more than three hands worth of days. I'm well down the path with forward testing and platform tuning and am hoping to be in position to test a small live account in the next few months to determine if this concept will work with larger funds and to also gauge how this approach is received on the retail level.

As things move from retail dealers to retail brokers to DMA any perceived advantage of executing from home off the DSL or cable modem would seem to evaporate and strategies like FPI don't seem viable when competing with colocated servers for choice fills. This aside form the points regarding the evaporation of liquidity when attempting a straight arb as the sole approach.
  • #117
  • Dec 25, 2011 2:34am Dec 25, 2011 2:34am
Quoting ianj1
The main problem is 'souring' latency, not cancellations (but they are sometimes a problem that must be settled with a 'gentlemen agreement' where goliath usually wins on a dispute over an out trade) - its like picking pennies off a steam roller drum - if you miss your exit - you can be squished! Most instututional arb platforms have a min size of 10 lots or so - arbing 10-50 lots is normal. If a last looker does not like your price and 'sits' on it for 500-1000ms you could be screwed (try it on USD/ZAR or USD/MXN) On odd occasions a participant...
do tou have proof that there is latency incclusion to the execution of the strategy
  • #118
  • Edited 6:01am Dec 25, 2011 4:19am | Edited 6:01am
Quoting sidhujag
do tou have proof that there is latency incclusion to the execution of the strategy
Don't be silly !(no offence intended) Remember it is probably not introduced by the broker, but possibly by the liquidity provider - and it could be applied to anonymous accounts or simply individual trades, so it may not be 'personal' It might be unfair practise, but its not illegal practise - so there are no come backs for it.

Even if i had proof, it would be professional suicide to discuss it in detail

In any case, its all part of the game - we are predatorial, they are defensive - they are tools of the trade. There is no point in pursuing it - just being aware that it might happen. And retail may be completely different (sorry for banging on about this separation)

What is really unfair is trade breaking - that can crucify you ! Fortunately it does not happen often
  • #119
  • Dec 25, 2011 1:16pm Dec 25, 2011 1:16pm
All this talk about how arbing wont work makes me wounder even more if what Jason fielder (forexArb) did is really just a 2k scam?
I think now we know why he did not automate it.
I'm glad i kept my money.
  • #120
  • Dec 25, 2011 2:06pm Dec 25, 2011 2:06pm
Quoting tommbstone
All this talk about how arbing wont work makes me wounder even more if what Jason fielder (forexArb) did is really just a 2k scam?
I think now we know why he did not automate it.
I'm glad i kept my money.
i'm glad you kept your money. it's surely a 2k piece of ... whats the point of buying and selling on two brokers to lock in a pip when you still need to close the two positions anything can happen in the mean time.
1 trader viewing now