NuclearPhynance Archive

I recently joined a venture, part of a group distributor of electronic components and appliances in high tech, aerospace and defense.

We are looking to partner with a broker, broker/dealer, to finish developing a hardware solution (fpga) for algo trading, market data/order handling and real time analytics (risk management, pricing etc).

The long term goal is using the technology for trading. Markets to cover futures US/EU/AU/JP, US equities, FX venues.



The deal would be having services and using the technology for free or a fee just to cover costs if they become too high. In exchange we need full support in the development, so api and network access and inputs for what is that you will be happy to see implemented.



For the second stage we will need help accessing more capital, we have some to start with and test the strategies.



Please get in touch if interested or please comment if you know some names.

I dont quite understand that and probably other people also.

From my understanding you want to build profitable trading strategy using fpga processors, when you dont have strategy or prototype of it yet, then in first place on what assumptions based on is your decision about using fpga for it?

Hi Svisstack



I will explain it better then.



The technology exists and works, we need to write the adapters for more markets (market data and orders transmission) and test it in a live environment.



The first goal is building an algo trading platform, real time risk management and pricing engines, reason why we need support in accessing exchange feeds, hence we were thinking of making a deal with a "first client"/sponsor.



Long term we would like to use it for trading, so yes we will develop trading strategies that use the advantage given by the infrastructure.



Please note that this is a company with substantial resources.

you're either a vendor or a prop firm, you need make up your mind. communicating your intention of bootstrapping from vendor to hft isn't going to win you any clients, for reasons which should be obvious.



secondly, if your company has substantial resources, then why do you need "first client"/sponsors? just sign up for and pull the exchange feeds. in the very least, it will be a good test of how willing your company is to actually allocate those resources.

Thanks for your comment, it is a good point, although working with brokers should not represent a conflict that much, you are right.



In any case we are going to be vendors that is the only certain thing.



I was thinking that a first client could be a way for both to save on costs (there are some issues/collateral damage for me on spending, especially if it can be avoided) and for us to understand more what is really needed, our fear is that we put too much effort in things that are not really needed that much.



I received some good advice privately too so thanks all for your help.

@zerohair



I had some conversations and you were right not much of a saving working with brokers, better working on our own and once developed and tested letting them help with the sales.



One point that came out was that what is really needed is low latency feeds and maybe a way to implement models on fpgas too. Orders not that much because it depends on hft regs.

You are talking about covering a very broad # of markets, all of which have VERY different MD/OR nuances. On some of the relevant exchanges, hardware can be quite tough to do such that it's a competitive edge and tremendously easy to fuck up and think you're fast.

We decided to prioritise the market data feed, as that is what is mostly required given our recent conversations, i.e. that's what we are more likely to sell in the short term.



Probably we are not going to make it with many exchanges either so let's see..., thanks for the heads up though

@radikal

could you please expand this (even privately)... tremendously easy to fuck up and think you're fast

It would be greatly appreciated.... and sorry for the late reply