Fluid CEO: Half a second of latency equals around 2% gap risk in BTC

Ashton Addison: What is Fluid, and what problem is it solving?▶ 00:40

Ahmed Ismail: I was a senior investment banker on Wall Street for 15 years. I then co-founded Haven, the first fully regulated OTC in the Middle East. Last year I branched out to set up Fluid, because one of the big problems in our industry is liquidity. Liquidity is the lifeblood of crypto trading. It's the same in equities, FX, fixed income and commodities. Without it, you can't trade. Crypto liquidity is super fragmented. It's a trillion-dollar industry with the same liquidity attributes as equities about 30 years ago. There are liquidity aggregators around, but none of them are solving the problem. If they were, crypto wouldn't be so illiquid. An aggregator is like Skyscanner or Expedia when you buy a flight. You get the best price and a verified way of buying, and it keeps competition healthy by bringing all the marketplaces together. That's what we do as a liquidity aggregator. Our users are institutional trading houses and the tier two, three and four exchanges that need more liquidity to thicken up their pairs and get better price tension.

Ashton: Bitcoin is already the most liquid coin. How much difference can Fluid make there?▶ 13:51

Ahmed: In the near term we're focused on the top 20 tokens. More than 85% of BTC trading happens on Binance today, according to a statistic released last week, which shows how fragmented the market is. On our website, fluid.finance, we have a price disparity calculator that shows the difference in price across the top six exchanges. For Bitcoin, the gap at any given moment is anywhere between $20 and $180, and that's on a typical sideways day when pricing should be really tight. That's a massive arbitrage opportunity, but it also shows that users are being short-changed by the big exchanges. A liquidity aggregator takes out that arbitrage opportunity. It makes market makers across all these venues honest, because they have to offer a better price to get the flow. That adds competitive tension and harmonizes pricing. It also makes markets less susceptible to flash crashes. You'll never get rid of them completely, but the more real liquidity aggregators there are, the more efficient markets become.

Ashton: Why does Fluid use AI, and why does latency matter so much?▶ 04:21

Ahmed: Latency is the biggest problem in the aggregator industry. Even in the most liquid of markets, such as BTC, 0.5 seconds of latency equates to around 2% of gap risk. If you're an OTC desk or an exchange with a half-second round trip, you have to price 2% wide to compensate yourself, which means you're hurting your customers. You're probably not doing it intentionally, but otherwise you can lose out. That's one of the big reasons crypto is so illiquid. We've invested very heavily in ultra-low latency infrastructure. We use methods from traditional finance, such as co-location and smart order routing, and overlay them with a prediction engine that uses AI.

Ashton: How do your AI models work, and how accurate are they?▶ 06:11

Ahmed: We have partnerships with two academic institutions, Imperial College London and the Blekinge Institute of Technology in Sweden. Around 15 post-doctoral researchers work for us full-time to build hybrid AI models that predict order books on exchanges. They go beyond the simple statistical methods most high-frequency trading houses use, like MACD and Bollinger bands. On live data, we can predict order books up to 30 seconds ahead. Accuracy is up to 99.5% for up to 8 seconds, and about 95% between 8 and 30 seconds. We also train the models on sentiment analysis, using things like Twitter hashtags and Discord boards. Crypto markets are far from linear, and the features that affect them are very complex. We feed in unstructured data, including commodity prices, the CPI and equity markets. The model finds patterns, and with a high degree of confidence we can eat away at that latency. In the end, it's a way of giving you as a trader best execution, meaning the best possible price on any market in the shortest time.

Ashton: Could your AI call market tops and bottoms? Would you trade on it?▶ 09:55

Ahmed: In theory, we could probably use our AI to trade. But we don't, because trading is a whole different ballgame. You need to look at risk, risk management and compliance. And you can't run a broker agency business and a proprietary trading business under one roof, as we saw with FTX. We took that on board. We don't trade people's money. We simply provide best execution.

Ashton: Did the 2021 boom make a visible difference to liquidity compared with now?▶ 17:07

Ahmed: With the downfall of FTX, the market has become massively fragmented. That means the problem we're trying to solve has become even greater, and markets are even more inefficient than a year ago. For us, that's a massive opportunity. Volumes have dropped, and retail traders have been spooked. Even institutional users have been spooked by the cowboy nature of many operators in the industry. I've been very vocal in calling for better regulatory oversight. A lot of OG crypto people are libertarians who want to change the system through self-regulation. But FTX showed that this doesn't stop the human element of fraud. The venues people trade on need better oversight. As a liquidity aggregator, we only work with verified, compliant, best-in-class exchanges, and there aren't many of them at the moment. Anyone who did diligence on FTX saw very quickly that things weren't right. We take controls, policies and procedures very seriously internally. Things take a little longer to set up, but they'll be done right.

Ashton: How long has the team been at this, and where does the product stand?▶ 11:31

Ahmed: We started in early 2021. The team is made up of ex-bankers and people with 15 to 20-plus years of experience. Our CTO built electronic trading systems for Goldman Sachs, JP Morgan and UBS. Our head of product spent 20 years at Bloomberg, Lehman and Citi working on electronic trading. Our head of data science is a professor who runs the two academic partnerships and has the 15 post-doc researchers working for him. We've now released our MVP, which showcases our order book aggregation technology and the ability to predict order books and prices. For now it only shows BTC/USDT. The vision is to branch out to ETH and the top 20 currencies. In maybe two years, we could extend to any tokenized market, whether that's real estate, NFTs or fractional ownership of equities. Anything with a tokenized, digitized marketplace is somewhere Fluid can become the liquidity aggregator.

Ashton: Looking to 2023, what does the team need to roll out next?▶ 21:13

Ahmed: We're now building execution. The MVP aggregates the liquidity and showcases our AI capabilities. The product suite is built on an execution layer, so you can actually put orders into our smart order router. You won't need to set up accounts with 20 exchanges. You set up one account with Fluid, and through our institutional accounts at these exchanges, which we do a lot of diligence on, you get the best possible price. That's what we're building now, and hopefully it will roll out in Q1 next year.