As trading latency continues to plateau, startup alternative trading system operator OneChronos looks to take speed out of the equation.

The new platform, which is slated to start trading US equities in early 2019, relies on a distributed architecture that will address latency issue, Richard Suth, a co-founder of OneChronos, told Markets Media.

OneChronos will have multiple points-of-presence in New York, Chicago, as well as other financial centers, he said.

“We probably will launch in NY4 or NY5 and add Chicago shortly after, but that is not carved in stone,” said fellow co-founder Stephen Johnson.

The orders that come into a PoP from a subscriber are time stamped at single-digit nanosecond accuracy before the PoP routes it on to the matching engine. The platform provides a few-millisecond window before each auction on ATS so that the orders that are traveling the longest can arrive and participate in the auction.

“The width of the window is adaptive and is based on current network path latencies and our historical estimates,” added Kelly Littlepage, another co-founder of OneChronos.

To improve the accuracy of its timestamps, OneChronos confirms the timestamps at the network layer so that they are optical and electrical signals rather than touching a computer to apply the time stamps, he said.

However, the sensitivity in applying those timestamps are fairly low, according to Johnson.

“As long as we get within tens of nanoseconds, or even hundreds of nanoseconds, it will not have a material impact on order entry since the auctions are not time priority,” he said.

The auctions will use a price-size priority and run an optimization process across all of the stocks using the most recently timestamped market data to find matches.

“No one has taken the same optimization-based approach before,” said Johnson. “It allows the trading of portfolio and market impact curves as well as allowing parent-sized orders to interact with each other in a way that could could provide a lot more liquidity.”