Proof Trading Progress Update: 27 Months

Wow, this past quarter has been crazy exciting and crazy intense. It started off with a bang with our first ever live trade, and it closed with an even bigger bang as we turned on and began ramping up our first pilot customer. The final tech push preparing for “customer go-live” was super high intensity. It is one thing to build a fully functional trading system from scratch, but conducting war games and testing each infrastructural and logical failure scenario and its remediation to exercise each of the many layers of safety mechanisms and redundancies, whew. It’s a mouthful just to even say. I will never forget the raw adrenaline of launch day at IEX back in 2013, nor the brutal months leading up to it, and likewise we will never forget this time period at Proof. Unlike IEX, Proof doesn’t match buyers and sellers, so we have the luxury of a slow and deliberate ramp up as opposed to a single big bang launch. But even with just 1 live pilot customer, it is imperative that Proof’s platform be safe and resilient. From the moment we started designing the system, safety has been our top priority, even above the fancy quant models and trading features.

Our team has built some excellent and heavily used trading platforms in the past, but in terms of the core technology, Proof is probably the most impressive thing we’ve ever worked on. We have already published a few pieces detailing various aspects of our build, and now that our CTO Prerak is through most of the heavy lifting launching the technology platform, he plans to write up a deep holistic overview of our entire technology stack. Stay tuned!

Major updates from Q1 2021

• Proof Services LLC, our broker-dealer subsidiary, executed its first trade and onboarded and began ramping up its first pilot client with our first algo offering: VWAP.
• Hired Brian Foley as our Head of Sales.
• Integrated and certified with Refinitiv Autex and Enfusion.
• Published content on how we connect to the stock market, our values, and the various external vendors we have engaged at Proof.

Goals from our previous update

Technology

Harden production environment and deployment processes.

• (10/10) This is what we spent the last two months on. We believe that our current production environment setup is secure, reliable, and sufficiently redundant. We’ve used “Infrastructure as Code” tools to create a replicable and stable environment — this includes not only the servers, but also all of the networking, security, and storage components. We have also designed our deployment and operational processes to be a good balance of automation and human checks.

Automate post-trade processes.

• (9/10) We have automated a whole slew of post-trade processes, including Clearing, OATS, CAT, Supervision reports (such as Daily Activity Report and Time Synchronization Report), as well as the backup and archival processes as required by FINRA/SEC regulation.

Integrate with EMSs used by our pilot customers, including any required FIX networks.

• (10/10) As mentioned above, we have integrated and certified with Autex and Enfusion, and are fully live with production orders coming through those vendors.

Product/Research

Finish design, implementation, and testing of our second algo.

• (2/10) Our second algo is taking longer than expected to design, as the amount of noise in market price movements makes it difficult to tease out performance differences between different combinations of features in the underlying model. But we’ve learned a lot in the process, as well as built some sophisticated tools for quickly testing various features. We are confident that we will complete the design as well as implementation and testing in the next quarter.

Publish research/content on information-leakage metric, intra-order trading model, and algorithm design.

• (2/10) We’ve begun drafting this content, but this was pushed back to focus on the design of the second algo.

Design further tests that can be run on our live trading data (once we have it) to continually validate, evaluate, and improve the algo design.

• (8/10) We have developed a suite of robustness checks for our performance metrics to help mitigate the impact of noise. We nonetheless expect our early sample sizes will be too small to draw nuanced conclusions anytime soon. Stay tuned for a blog post describing our techniques.

Sales/Marketing

Go live with initial pilot clients.

• (10/10) We launched with our first pilot client!

Connect/reconnect with potential pilot and early adopter clients.

• (7/10) Bringing Brian on the team combined with actually having a live platform to talk about has led to a significant increase in client dialogue.

Launch a new website and a new framework for how we think about progress and goals going forward.

• (0/10) We kicked the can down the road on this goal (how meta), but we do still hope to do it this coming quarter.

Expand upon our article on how to obtain a b/d license to create a multi-part series covering what it takes to start an institutional equities broker-dealer from scratch (e.g. technology, team, vendors, financials, etc).

• (3/10) We published an article detailing our vendors, but we did not complete the entire series. We hope to write the remaining posts in Q2.

Operations/Regulatory

Post job listings and ramp up interviewing for senior sales, technology, and operations roles.

• (8/10) We filled our senior sales role, and we posted and began interviewing for a senior technology hire. We decided we were not quite ready to make a dedicated operations hire, but we may start looking in the near future.

Productionize/automate ongoing regulatory and compliance requirements.

• (8/10) we have automated the majority of regulatory and compliance tasks (e.g. CAT, OATS, daily activity reports), but we still have a couple remaining manual processes (e.g. KYC/AML).

Goals for the next 3 months

Technology

• Integrate with the EMS vendor(s) used by the next pilot customers.
• Add features that enhance system scalability.

Product/Research

• Finish design, implementation, and testing of the second algo.
• Publish research on information leakage and the design of the second algo.
• Analyze data from early trading activity and incorporate any resulting lessons into the process of improving the algos.

Sales/Marketing

• Go live with at least 1 additional pilot customer, and have at least 5 pilot customers in the onboarding pipeline.
• Launch a new website and a new framework for how we think about progress and goals going forward.
• Complete our multi-part series covering what it takes to start an institutional equities broker-dealer from scratch (e.g. technology, team, vendors, financials, how to obtain a b-d license, etc).

Operations/Regulatory

• Continue to automate operational processes, including integration with CAT CAIS and 606 reporting.

Progress Summary

Additional thoughts:

At various stages in the life of a startup, different members of the team are required to step up and carry the weight of the entire company on their shoulders, where failure might mean the startup’s demise. An easy example is fundraising, when all eyes are on the CEO. Over the past quarter, and really the past 6 months, our entire tech team has been running on all cylinders, and specifically our CTO Prerak was the one required to step up, leading both our technology and operational efforts. Prerak not only hand-built a huge swath of the system himself, but he coordinated the entire team/build, and managed nearly all of our external integrations including client connectivity, market access, clearing, and regulatory reporting (all with a newborn at home!). I can’t stress enough how monumental this effort was — there are so many mission-critical moving parts — and the end result has been nothing short of miraculous. We seek to have one of the most powerful, modern, and robust trading platforms on the street. We obviously have a long way to go to turn Proof into a lasting business, but Prerak and his team have delivered on their part with flying colors (although of course there’s always plenty more to be done). Now the focus shifts back to Allison and me: we need to take this amazing platform we’ve been handed, and layer on thoughtful and highly effective trading algorithms, and then continue iterating until we have something truly special. We believe we have the right skillset and approach to get there, but the research process is never 100% predictable.

Here we go!