Seeking partner for setting up a FX High-Frequency Trading Desk

I'm very sicked and tired for those bad design trading bots/EAs, that's just wasting money if not been deceived! So I'll try to seeking partners and rising funds ($10K would be sufficient) then setting my/our own HFT desk. If you're 'seriously' want to plunge into this business you can join the discuss. I have well connection with the 'industrial', but still need some seed funds to fire up. Any question and PM welcome. Please no broker or EA seller. And I don't sell any system! Thanks.

Interesting idea, is the $10k for trading capital or for your startup expenses?

Hi forexproptrader,
It'll be mainly spend on infrastructures (high performance trading engine, data analyzed tools, ticker plant, colo server, .. etc.), and first few months' regularly expenses. The system needed testing at least few weeks even months before trading real money.

Add: The personnel costs should be included, yet you always paid for what you got. Experienced personnel may cost more but will be worth.

What platform you will be using? If you are serious about HFT you need a DMA and institutional access to trade within spread, so 10k looks very naive in my opinion. Unless you mean something like MDP and MT4...

Hi MistiFx,
LMAX would be qualified. And yes, 10k for a HFT start-up could be quite low compared to those multi-million dollar ultra high frequency trading operations/firms. But you virtually could do sub-second high frequency trading (hundreds trading per second) with no problem by using many off-the-shelf solutions (of course they're far more sophisticated than 'MT4' and 'scalpers EA') these days. And for those firm which have plenty resources, they might be choose R/D their own tools and systems form the ground, and often need far more time and manpower for these works since they're almost trading as many instruments as they can to reducing risks. But for low budget operations which focus only on trading few instruments these custom-built tools/systems certainly wouldn't be economic effected. Indeed, there'll be always have some 'trade-off' factors.

10k is not nearly enough and for a FX HFT in the traditional sense you need access to an order book.

You have that?

Hi PipGnostic,
We could accessing 5 levels order book by LMAX's $1,000 account. That would be sufficient for the early stage testing and trading operations.

I see. Hope they're fast enough.

I like them actually by the looks of it and have my API in JAVA anyway, be easy to integrate.

I would submit though, that at this stage of the HFT game I wouldn't try take these guys on in their own game. Needs to be a new form of strategy that is not dependent on speed alone.

For forex you can't expect more than 20 000 maybe 40 000 ticks per pair per day, so there is only that many value changes. If you try and battle it out at the order book real HFT style on speed, most likely these guys are to far ahead to stand a real chance.

Tell you another thing, LMAX saying they update 10 times a second on that entry account, so they work in milli seconds rather than micro seconds BATS style, so here you're looking at 1/1000 of a second where there are people out there working in 1/1 000 000 of a second, so already this exchange is 1000 slower than the benchmark.

Yes, we'll need smarter strategies and algorithms for compensated our execution speed if we have to used slower trading systems or engines.

We don't need to dependent on LMAX's data feed. By ordering ultra low latency full order book stream, we'll get considerable advantage in this factor. And for the FX trading, it's inherent factor which's not been fully centralized, hence the processing speed can't compare to 'BATS'. This condition could be unavoidable at present.

I was thinking along the similar lines by utilizing order book and reference institutional feed provided by Black Light Data (orderbookfx), but in my opinion their signals only good for sub-pip, or 1-2 pips moves, so unless you pay no spread the strategy will be struggling to break even and then you add commissions and costs.. Would be possible to trade that way with an access like market makers have but it's a different level of the game, nothing to do with retail/prop trading.

I am building up algorithmic trading business myself (www.fxalgo.net) but actually surrendered the idea of HFT along the lines of market making in favor to less demanding strategies.

One interesting is about arbitraging between forex and fx futures. Execution for both can be done via FIX.

Damn difficult to do though in fx, be warned, will take a lot of time.

Fx, especially the HFT stuff is without doubt the most difficult thing I've done in my life. Maybe I'm just not clever enough, but I had to fight for it really hard.

OK, FX might be not very easy for doing HFT. It's could be harder than stocks and futures. I'm testing my pre-HFT strategy in E-Mini S&P futures by the demo accounts. The result was quite interesting. You could profit around $300 within 40 minutes. Please note this's done by 'manual'. Since I'm not a coder/programmer, I'll need someone who can do the programming works team up with me for implement/converted the strategy into algorithms and automatic trading systems/platforms. If you're the people who can and willing doing these jobs it can be further discussed.

Well I had a hard long look at LMAX .

On FX they're missing the point. Smallest units would be 1000 units or 0.01 lots on a 100k lot sizes. Which is Metatrader sizes. Just not good enough. Oanda breaks their trade sizes down to 1 unit, which is what one really needs. And the API trades about the same speeds as LMAX, if not faster.

why on Earth you need to trade 1 units? Oanda is a market maker of it's purest form and hedges only overall exposure, they don't transmit orders anywhere, this is why they offer good speed. But when you start winning using scalping with Oanda they will do everything to block you as they will lose, since they don't hedge their short-term exposure only long term. That is the reasons of 50 pips spreads during news, that kills many technical positions too.

Most interfaces good enough for HFT (FIX for example) trade minimum 1 lot, which is just fine if you ask me because you aren't shooting for hundreds of pips with relevant huge stops for for a couple of pips usually. 100k minimum account to use FIX (normally) would be trading 100 lots positions otherwise it would make a very little sense..

I have a good algo for HFT, makes miracles on demo, similar to MDP but not MDP and I can have it coded for FIX or JForex or whatever. The question is where to find a good execution? I love Dukas but if they slip 1-2 pips too often it would kill everything. And they do slip more than most MT4 retail brokers.

I trade 1 unit if I can cause I can make a profit on it. Means I can get exact trade sizes and values. If I want to throw 1 more Euro at the market I can.

Ok. Can you share your experience and concerns about LMAX besides lot sizes? Basically what about theri multicharts and their proprietary API? I have never heard of them before actually. Worth looking deeper? Slippage? Ease of use of their API?

I didn't look further Misti, I saw the trade sizes and stopped there.

API is API I guess, they seem awfully proud of their execution speed and claim to be an exchange as apposed to an ECN broker. But to me it just looks like a faster MT broker with the option to use other languages.

They were measuring in milliseconds as apposed to micro seconds. So nothing special on the speed side. I guess I just switched off when I saw I couldn't pursue my strategy on it.