How to get your Quant programming project privately funded through Linked In
Here is conversation found on a Linked In group. This shows you how it is possible to get a quant based coding project privately funded by a prop shop.
What could you do if you had a trading platform with “Tick-To-Trade” latency of 5 microseconds?
“Tick-To-Trade” latency is a period of time from the moment a Market Data Update message appears at the platform’s input port, to the moment when the last byte of an order message resulting from that update leaves the platform’s output port. would be very interested in discussions if it were futures based
It could be just as easily as equities or options or FX or…….
The impressiveness of that speed and its utility really depends on what’s happening in those 5 usec. Is there a precalculated model for trade generation? Is the model useful/non-trivial? Is risk management/hedging considered? I could whip up a low latency app that executes a trade every time there are 10 consecutive up ticks. Sure, it may be very fast, but it doesn’t implement a reasonable strategy so it’s not useful.
Though the specific market isn’t important in principle, there are many practical differences in strategy development. If you’re doing options, how are you getting theos? Is that speed based on a lookup from a cache? What about cache misses? I suppose you could operate only on the chart, but few firms have such simplistic strategies.
There are no precalculated or trivial algorithms, market data is kept in full depth Order Book, real strategy algorithms of medium complexity then use it to manipulate orders while utilizing risk management and hedging in real time.
Options are not currently a priority, but theoretical values are calculated in real time utilizing few tricks.
This is not a general purpose platform to trade everything under the sun simultaneously by an IB, but rather precisely crafted tool best utilized in targeted areas by a small to medium prop shop.
Does it have connectivity to Globex? If so please get in touch.
I have many questions, some simple like XXX about what exchanges you currently have connectivety agreements with. It seems that your post question has generated some interest, it might be the time for you to follow up and let us know if you have a product, what it is, show a demo and start doing business, or if you have just poised a question
This was definitely not a rhetorical question.
It is a living, breathing (with its cooling fans) system.
The initial area of trading is futures, but due to the design it will accomodate other asset types with equal effectivenes. The obvious extra effort would be connectivity gateways, but I deem it to be of minor size/scope.
In fact many exchanges, for example NASDAG employ protocols magnitudes simpler than the ones required by CME for example.
This system is very finely optimized at all levels: hardware, operating system, application software, because I have been involved with all of these components throughout my long career, if I told you how long you’d think I’m a living dinosaur, suffice to say I was playing with individual bits on 4 bit processors.
In the course of that I learned some tricks that they don’t teach anymore but surprisingly are still as useful as ever or even more so, because spoiled by the enormous processing power of current processors, developers manage very successfuly to waste most of it.
sounds great XXXX. what are you offering/selling/looking for?
Initially I was looking for some feedback and discussion, but so far I have only been answering questions.
I am curious what asset type would be best suited for this type of system. Personally, FX (not FX futures) has always been a mystery to me in terms of how it actually works, how to connect to it, and if latency is as relevant as say for equities?
I am not asking for any confidential or proprietary info, but more in terms of something generic like, does it even matter that the latency is 5us? I had a casual conversation with some equity arb traders, who said they wouldn’t even know what to do with 5us latency, because they don’t have algos for such short time horizon.
Below, I share some more info about my goals and design philosophy I adopted before the inception of the project.
My #1 requirement and goal was the lowest possible deterministic value of worst case latency.
Best case latency is almost meaningless because it occurs when it’s hardly needed, when trading volume is low and few profit making opportunities exist.
The true measure of systems latency is the worst case scenario, when trading volume is at the maximum level and there are vastly more opportunities for profits. That’s when many systems choke, unable to handle peak trading loads and thus unable to exploit these opportunities.
Hence my near obsession with minimizing the worst case latency and making it deterministic.
The only source of latency jitter in this system are unavoidable, occasional cache memory misses.
My approach was that latency trumps features. Features can be added later, but significantly reducing latency after the system is developed, requires major if nor complete redesign.
Right now everything is compiled, there are obviously different configuration utilities, but no SDKs.
There is a rich set of APIs for getting/updating various information from the Order Book, strategy scheduling, list processing (yes I wrote my own link list implementation) and memory management (I also wrote my own, deterministic memory management system).
I spent a lot of time thinking, and very little coding. If I worked for one of those companies who keep metrics about their developers including number of lines of code written, my career there would be very short lived.
The result is, that the entire executable code including gateways, several strategies, risk, compliance and order submission/execution is on the order of 20,000 lines of source code.
Are you hoping to make this an open source project? Do you also have a link for a more detailed description of this project you are working on?
I am afraid not, the system was designed to push the latency limits in trading environment in a software only computing environment.
The goal however was to provide competitive advantage for a trading company as opposed to being publicly available.
So are you selling this as a commercial product to various trading companies or just as say, have one private trading company fund this project.
Let me know if this is something you would like to further discuss. i would have interest in funding such a project.