Bringing Sexy Back...to FIX

Discussion in 'Announcements' started by TT News, Feb 16, 2017.

  1. Drew Shields

    Drew Shields Sponsor

    Apologies for confusing matters. The new TT platform can play a part in every step in the 'pipeline' as you call it. But before I get to that... indeed the blog post was about FIX as a Service on the new platform and not algorithmic trading. It was your follow up question which raised the topic of HFT where you asked me to agree or disagree that our FIX offering is insufficient for HFT firms. Later you said "That was a rhetorical question as hft was mentioned by the TT executive which is obvious pure snake oil and marketing hype." I still don't see where I talked about HFT in the blog post and would gladly clarify if someone would point out the statement in question. I'm not in the snake oil business as anyone who has dealt with me (inc here on ET) will attest. I'm not afraid to say what TT does not offer or to identify the use cases where customers are better suited to go elsewhere.

    The only reason this conversation then shifted to automated trading is because you said to oversea "hft, as used by TT is misleading and should never be used by TT whatsoever" and "max 40 micro seconds to stay competitive" and "They [TT] do not have experience in this space (algorithmic trading)".

    Rather than let those statements sit out there, I felt the need to address the fact that your implication that TT has no business talking about algorithmic trading is incorrect. We have thousands of algorithmic traders using our Autospreaders, APIs, and ADL on our legacy platform and many are switching to the new 'TT' because of the performance gains. You quoted 40 mics as a benchmark for something and I thought it necessary to point out we beat that in some cases.

    So, to your final question here...what part of the 'pipeline' can TT provide? The answer is that we offer services at every step of the pipeline. For the sake of clarity, I'll only be referring to the new platform in the bullets below. We offer even more on the legacy platform but the performance, flexibility, and accessibility of 'TT' is what will set it apart from X_TRADER in the long term.
    • We can provide a user interface for controlling algorithms/autospreaders as well manually managing orders. The same UI provides charting, position management, and some lightweight trading analytics along with options valuation and soon options scenario risk.
    • In contrast to how you described TT, we have autospreaders and the Algo Design Lab which are high performance automated trading tools (and will soon offer a C++ sdk for algorithm development). Users use these tools to build their own custom logic and they indeed run headless in the colo as close to the matching engine as possible. We believe that we're the fastest 'off the shelf' platform for automated trading (with the added benefit for brokers/FCMs that they control pre-trade risk inline). Obviously, the largest HFTs or firms that can invest in FPGA development for custom algos can always find ways to be faster, but we think that we achieve a balance between flexibility and performance.
    • We offer pre- and post-trade risk management to brokers and FCMs which is fully integrated with the rest of the platform. Via FIX, our risk services can also be used for an entire firm's order flow regardless of how it is sourced.
    • We offer a 'forever audit trail' where every execution related message for all users on all accounts is housed forever in a data warehouse allowing us to be a clearinghouse for historical trade data for a firm. You don't have to use us for execution to take advantage of this. Inbound drop copy can feed this service if you don't use TT to trade.
    • We offer FIX services as described in the blog that got this entire conversation going. In terms of using us for market-access/connectivity via FIX, as I said in the blog, we have latencies around 50-55 microseconds today (from receiving a FIX message to getting it out to the exchange in the native format with a broker risk check in the middle). I expect those latencies to reduce significantly over the next 12-18 months as we invest in making our FIX solution truly viable for anyone with even challenging latency requirements.

    Three other comments I'll make regarding performance/latency. (1) We have rebuilt everything in the critical path of a trade to run on bare metal servers, c/c++ on linux, using solarflare NICs. While this doesn't make us 'better' than HFTs, it does make us better than most off the shelf players in the space. This applies to all FIX services as well. (2) You make a good point about the added latency resulting from the flexibility I described in the blog (using DNS, load balancers, etc). Where FIX customers need to squeeze out extra latency and value that over flexibility, we will be willing to work with them to make that happen by bypassing DNS and load balancers, but we always want to work to make our offerings as flexible as possible by default. (3) The new platform is colocated more than we ever were with X_TRADER/TTNET. We aren't just building high performance software and then letting it be deployed where it's convenient or cheap. We are investing heavily the in the fastest commercially available long haul network routes and in colo space for nearly every one of the 45 markets we support (growing to more than 50 over the next year).

    So TT is really a suite of services, not just a screen. We think firms/users can mix and match how they use our new platform in many different ways.

    Want to build and deploy custom algos? TT will give you a great off the shelf option that is truly competitive (30-50 mics for entire algos to run tick to trade and trade to hedge).

    Want to just use us for market access? TT will give you the fastest FIX offering on the market.

    If there are very specific use cases you, or anyone, would like to explore further, just PM me. Happy to discuss and I promise there will be no snake oil. If TT isn't the right choice for you, we'd rather be honest and transparent and lose some near term revenue for the sake of long term reputation and trust.
     
    #21     Feb 23, 2017