The recent outages at some of the world’s largest centralized cryptocurrency exchanges have mainly been caused by algorithmic trading firms, Ivo Crnkovic-Rubsamen, the chief strategy officer and technical lead for trading at the dydx exchange, told Cointelegraph in an exclusive interview.
“Because there’s so much retail interest and the price action is moving so fast, all of the algorithmic trading firms are vastly increasing the rate of order placements and cancels they want to send to the matching engine to maintain their positions […] It’s common for a trading firm to 20 times the output of orders and cancels at a very busy time.”
Some of the world’s most prominent exchanges suffered technical issues in the past week, including Binance, Coinbase, Kraken and Bybit, just days after Bitcoin breached $60,000 for the first time in over two years on Feb. 28.
“We do see this pretty much every bull market or every time there’s concentrated retail interest and big price moves,” added Crnkovic-Rubsamen.
Following the temporary outage, investment research firm Citron called for a short sale on Coinbase stock. Its shares rose over 11.36% in the 24 hours leading up to 12:25 pm UTC to trade at $229.15, according to Google Finance data.
Unlike decentralized exchanges (DEXs), some centralized exchanges could set custom trading limits for individual market makers based on trust assumptions, which contributes to the growing workload under bull market conditions, according to Crnkovic-Rubsamen.
“[Centralized exchanges] may have one market maker they know really well… and they’ll set their rate limit 10 times higher… That’s perfectly reasonable. but then the bull market comes and this market maker wants to trade PEPE 10 times faster and all of a sudden it is a problem that the rate limit for them is way higher.”
In contrast, the rate limit on decentralized exchanges is set by the protocol since DEXs have no direct relationships with market makers. Rubsamen added that centralized exchanges can be highly reliable during normal trading but less reliable than DEXs during peak bull market loads.
“Centralized matching engines are awesome at performance, they’re super optimized and efficient, but when they go down, that’s it […] There is a reliability trade-off there,” Crnkovic-Rubsamen said.