HFT Orders Getting filled against market

Hi Guys,
there was one thing that I noticed from the OPs initial post - he's connecting via PrimeXM so he is using a aggregator, this means he will lose some speed here (it adds an extra layer of communications to the trade execution process).

There are a couple of things that he could do to decrease his latency, so for example he says he's in NY - that's great but it's not a village. If you're trying to reduce your latency and you want to go the collocation route then you don't just go for the same city you go for the same machine room.
So most of the big boys use rack space so potentially you want your infrastructure to be hosted in NY1 or if you're Primexm instance is in London you're hopefully in LD4 (both of these machine rooms have a lot of the big boys infrastructure).

Other things to check are that the PrimeXM instances and the LP instances are in the same machine room.
I've worked with the PXM guys before and know that the likes of Clive are more than capable of advising you on latency and connectivity issues.
For those of you who haven't come across PXM before one of their best techy guys is a guy called Clive - ask for him by name if you're having trouble.

Hope that helps,
Paul.
 
volpunter said:
Joe, Euronext is an exchange where listed products are traded. Spot FX is an OTC product, an entirely different animal. You can't compare the two in terms of market makers or liquidity providers because they fulfill very different functions and are bound to entirely different regulations. Nobody in OTC space commits to anything, trust me on that, I have traded currencies professionally for a very long time. In spot fx space a liquidity provider can come and go as they like at their own discretion. That is why I pointed out there is a huge difference.
More...

There's no avoiding the fact that the other side of a retail bet is a market maker. Calling them liquidity provider doesn't change this fact. Nor the fact that they come and go prove they are not making markets.

The OP has no way of beating even a temporary market maker. They have to make a living and they have no reason to give out free money. What they do is they take the OP's business and moves the price against him. This stops him from stealing their business and at the same time potentially force him to regurgitate what he has bitten into at a loss. This is in fact what he is reporting.
 
My last comment to you because you seem clueless: Check out LMAX, nobody withdraws, nobody marks up, liquidity providers do not even see client information (this fact is ensured through official audits plus LMAX is a reputable ECN). No last-look provision. Clearly you come from the equity world but demonstrate you have zero knowledge of the spot currency world. You are simply posting factually incorrect information.

ExperiencedJoe said:
There's no avoiding the fact that the other side of a retail bet is a market maker. Calling them liquidity provider doesn't change this fact. Nor the fact that they come and go prove they are not making markets.

The OP has no way of beating even a temporary market maker. They have to make a living and they have no reason to give out free money. What they do is they take the OP's business and moves the price against him. This stops him from stealing their business and at the same time potentially force him to regurgitate what he has bitten into at a loss. This is in fact what he is reporting.
More...
 
Wow you guys are brutal. It is tough hearing how stupid I must have been to think this would work. But it is definitely what I need.

The 8 microseconds is really just detecting the opportunity, not to send the order. I did say my ping times are 1 ms. So if you really add it all up, it takes .5 ms to get the data, 28 microseconds to decode the fix feed to data your app can use and 8 microseconds to detect the opportunity.

@volpunter, You are right, I have a masters in CS. But that doesn't really help with this, they teach you how to think, I taught myself how to code for the real world (not boasting, but it was out of necessity). I did research the algorithm. There is only so much you can get from the internet. If you look at my post history, you will see back in April I asked for practical help. I knew there was a huge gap between the theoretical knowledge and the practical knowledge I needed. Almost everyone was willing to help once I started making money. All I have to offer is that I'm a decent programmer. So I worked with what I could get.


@ExperiencedJoe, I made my whole system. It wasn't something I promised anyone. I basically read on reddit how this guy (a CS guy), made a system and was cleaning up with his system. I had a little money that I could put into it, so I decided to try. I read about all I could about it. But there is a lot you just don't get from books. Also, all the legs of the strategy are conducted at different venues. I wouldn't buy from one person and then sell back to them at a higher price. My strategies crossed the market, I would buy at the ask and sell at the bid. Not that it matters now, I just wanted you to know.
 
I'm not built for manual pattern trading because of the psychological aspect. I tried it and lost a bunch of money. I figured since I'm a programmer and another programmer made it doing this, why not. He had a mentor (big difference).

So where do I go from now? I'm looking for a direction that could make use of my skill set.
 
PM me if you are interested in exchanging ideas or potentially collaborate. You seem to be a decent guy who has no problems wrapping up his sleeves and getting to work. I do have my own trading system architecture but am not co-located and I do not trade ultra high frequency strategies but I am very interested in hearing how you set up your system. I do run algorithms in futures but mostly in spot currency space. We could share ideas or just exchange thoughts. I have no interest in harvesting freebies. Your call.

max_w said:
Wow you guys are brutal. It is tough hearing how stupid I must have been to think this would work. But it is definitely what I need.

The 8 microseconds is really just detecting the opportunity, not to send the order. I did say my ping times are 1 ms. So if you really add it all up, it takes .5 ms to get the data, 28 microseconds to decode the fix feed to data your app can use and 8 microseconds to detect the opportunity.

@volpunter, You are right, I have a masters in CS. But that doesn't really help with this, they teach you how to think, I taught myself how to code for the real world (not boasting, but it was out of necessity). I did research the algorithm. There is only so much you can get from the internet. If you look at my post history, you will see back in April I asked for practical help. I knew there was a huge gap between the theoretical knowledge and the practical knowledge I needed. Almost everyone was willing to help once I started making money. All I have to offer is that I'm a decent programmer. So I worked with what I could get.


@ExperiencedJoe, I made my whole system. It wasn't something I promised anyone. I basically read on reddit how this guy (a CS guy), made a system and was cleaning up with his system. I had a little money that I could put into it, so I decided to try. I read about all I could about it. But there is a lot you just don't get from books. Also, all the legs of the strategy are conducted at different venues. I wouldn't buy from one person and then sell back to them at a higher price. My strategies crossed the market, I would buy at the ask and sell at the bid. Not that it matters now, I just wanted you to know.
More...
 
max_w said:
@ExperiencedJoe, I made my whole system. It wasn't something I promised anyone. I basically read on reddit how this guy (a CS guy), made a system and was cleaning up with his system. I had a little money that I could put into it, so I decided to try. I read about all I could about it. But there is a lot you just don't get from books. Also, all the legs of the strategy are conducted at different venues. I wouldn't buy from one person and then sell back to them at a higher price. My strategies crossed the market, I would buy at the ask and sell at the bid. Not that it matters now, I just wanted you to know.
More...

You buy from one bank and tries to sell to another bank. That has no chance to work. If the belief is that there are others in there like you who you can take advantage of, that won't work either because such persons would have been eaten by the banks long ago. In the end the game is just you and the banks.

The inter-bank connections are in the nano-second scale. Your 8 micro seconds is a bit slow to beat the banks.
 
then how come it worked in as simple as spot fx years ago? It worked according to atticus (and unless a lot of his other crap, that particular thread sounded pretty legit) with some fx broker that offered wrongly priced fx options. I believe it may not work anymore today but nothing speaks against doing it from a technical perspective. I have earned millions in dollars arbing brokers in the professional space on the options side. Sure, they may not like it if they find out but as long as you play the game smart it will take quite some time before someone notices. (and there is nothing whatsoever illegal in doing so, and I would argue it is not even unethical) Some individual trades netted me hundreds of thousands of dollars. Even firms like Goldman begged at times to cancel the trade. Sometimes I let them out because they obviously mispriced, at other times I held them up to a done deal. That was only a short few years ago. So, nothing speaks against opportunities existing as long as one finds a way to execute and take advantage of them. My example stems from the discretionary world, and I admit such opportunities may not exist in the algorithmic space anymore.

ExperiencedJoe said:
You buy from one bank and tries to sell to another bank. That has no chance to work. If the belief is that there are others in there like you who you can take advantage of, that won't work either because such persons would have been eaten by the banks long ago. In the end the game is just you and the banks.

The inter-bank connections are in the nano-second scale. Your 8 micro seconds is a bit slow to beat the banks.
More...
 
The opportunities do exist, but the infrastructure to take advantage of them is prohibitively expensive. Believe it or not there have been some trades that worked, maybe they weren't paying attention or something went wrong on their end. I did buy from one and sell to another. But more often than not it didn't. But stuff like that gave me hope until recently. I would probably have to be like Virtu and have direct cross connects to the banks.

So I'm trying to avoid this happening again. I could go online, find some price action or market taking algorithm, try and implement it (wouldn't be hard to do), but without someone giving me real world input, I'm bound to repeat the same mistake.

I'd love to be someone's apprentice, and heck work for free (this is wishful, but it would be nice). I'm an independent contract programmer, so I make my own hours. Barring that, I'd settle for someone pointing out the direction.
 
You must log in or register to reply here.
Back
Top