What's Your NFT Really Worth? This Man Is Using AI to Find Out

Nikolai Yakovenko has harnessed the power of machine learning for everything from professional baseball to human genomics. Now he’s coming for NFTs.

AccessTimeIconJul 7, 2022 at 6:11 p.m. UTC
Updated Jul 10, 2022 at 5:06 p.m. UTC
Layer 2

Eli is a news reporter for CoinDesk. He holds ETH, SOL and AVAX.

The year is 2006. The stage: an underground cash poker game in New York City. Nikolai Yakovenko takes a peek at his hand, his facial expression paused, posture slightly slouched. A pair of kings glance back at him.

With $30,000 on the table, the flop is revealed – ten of hearts, seven of spades, six of hearts. “All in,” his opponent gestures from across the table. An $80,000 bet is placed.

Yakovenko begins to crunch the numbers in his head, conjuring up the hand’s possible outcomes and their likelihoods. Moments later, he has his answer – 42%, his chance of victory. With more money on the table than Yakovenko is willing to lose, he folds.

“Maybe a bot would have played it better,” he later said of the hand during a talk at MIT’s sports analytics conference in 2018.

The sentiment – that technology can inform and even outperform human capabilities – is at the heart of Yakovenko’s life work, and has taken him around the world for poker tournaments, chess competitions and now, non-fungible tokens (NFTs).

His latest mission is to tame the Wild West of blue-chip NFT prices with an artificial-intelligence startup he founded called DeepNFTValue, which uses a pricing algorithm to assess the market value of high-priced digital collectibles like CryptoPunks and Bored Apes. (Both are collections of 10,000 computer-generated profile pictures, each with their own sets of traits and rarities.)

While NFTs from these collections have fetched millions of dollars in individual sales on NFT marketplaces (the cheapest price for a CryptoPunk with the “alien” trait is listed at over $12 million), buyers remain on their own in determining a fair price for their treasures. Even with the NFT market cooling off in recent months as part of a broader crypto downturn, Yakovenko sees this as a problem worth solving.

His expertise in doing so, melding the world of AI and statistical modeling with the unpredictable force that is human nature, dates back long before his days in crypto.

Early life

Yakovenko was born in a small town outside Moscow in 1984 to two young scientists who met on the Ukrainian national physics team. His early math talents were prodigious.

Immigrating to the U.S. with his family at the age of seven by way of Italy, Yakovenko began to code at the age of 10, enrolling in college classes at the University of Maryland, where his father was a professor, by the age of 14.

At 16, he became a full-time student at the university, taking graduate-level math courses and discovering the game of poker in the dorm rooms, later becoming a regular at higher-stakes games at an off-campus fraternity.

“I got very lucky in that the guys who played at my friend's frat were better than the players you’d meet in Atlantic City at the time,” Yakovenko told CoinDesk. “Poker was just beginning to boom, and no one knew what they were doing, including me.”

Yakovenko’s poker career began to blossom years later in New York City, where he lived during and after graduate school at Columbia University’s school of engineering and applied science, taking a full-time job at Google as an engineer on the company’s search engine team in 2006.

At 20 years old, he frequented an underground poker ring in Times Square, winning and losing more money than he ever had in college.

“I remember getting off work at Google and then going to the clubs and playing until 7 a.m.,” Yakovenko said. “You’d run up against all sorts of characters in those games. After a few months I think I had won $20,000 playing $300 buy-ins.”

Yakovenko’s poker career eventually took him to more esteemed settings like the World Series of Poker, but his most interesting games came in the underground contexts, including a stint at an infamous table run by Molly Bloom (whose story was turned into the film “Molly’s Game”) that regularly hosted high-profile celebrities, most notably Tobey Maguire.

“Tobey's actually a good player,” Yakovenko said of the actor’s time at the famed table. “Getting peddled by Spider-Man was an interesting experience.”

Like with his passion for chess at an early age, Yakovenko was obsessive about the game’s details, fixated on how deep learning and AI – two topics he had taken both a personal and professional interest in – could improve his game.

Moneyball

After leaving Google in 2008, Yakovenko found himself tinkering with the analytics of a different game: professional baseball.

What began as a passion project, blogging about statistical models and player projections, eventually caught the eye of pitch development pioneer Kyle Boddy, who was building his own baseball research brainchild, known as Driveline Baseball, across the country in Kent, Wash.

The findings of his studies, the most consequential being that it was beneficial for pitchers to throw harder (a seemingly obvious observation that was still contested at the time), led to ongoing consulting gigs with Driveline through the years, contributing in small part to the early days of a larger analytical revolution that has since changed the game of baseball significantly.

CoinDesk - Unknown

Yakovenko stretches atop the line-out at a Columbia University rugby match.

In 2012, Yakovenko suffered a traumatic brain injury during a Columbia alumni rugby match in which he was knocked unconscious and put into a medically induced coma.

After checking himself out of the hospital a week later, still unable to fully feel the right side of his body, Yakovenko waved off his doctor-prescribed therapy plan, instead opting for his own trial-and-error-tested cocktail of resistance bands, ping-pong balls, weight lifting and biking. He eventually made a full recovery.

His professional career from 2015 onward has included stints at Twitter, chip maker Nvidia and the hedge fund Point72 Asset Management, all in positions closely tied to deep learning. His projects ranged from finely tuning recommendations on user feeds at Twitter to genomics and DNA research at Nvidia; his time at Point72 was focused on algorithmic trading.

Punk revolution

When Larva Labs released its experimental on-chain “proof of concept” project CryptoPunks in 2017, Yakovenko was no stranger to cryptocurrency. He had been following bitcoin casually since 2011 and had published theoretical cryptography research of his own during his time at Columbia.

Yakovenko took interest in the collection in 2020 after noticing his old poker buddies talking about the project on Twitter, finding himself once again in the early days of a movement that would grow larger than he could have foreseen.

In the spring of 2021, Yakovenko became obsessed with CryptoPunks Bot, a Twitter account that served as a live feed for CryptoPunk sales.

He recalled a moment during Tech Week Miami where he would pull up the sale bot while riding in Ubers, asking people how much they thought each Punk was worth, trying to make sense of the discrepancies between “floor” Punks with common traits and rarer editions.

The experience would ultimately lead to an “aha” moment for Yakovenko, who figured he could create his own pricing algorithm that would be more accurate than any publicly available information.

“We were going around party to party and I'm like, ‘dude, I have to build a model,’” Yakovenko said. “I wanted to do something in crypto machine learning, but as an engineer, you have to be very careful not to be the hammer looking for the nail, right? I wanted it to be useful.”

After playing around with the model for a few months in his spare time, he founded DeepNFTValue, a company that offered just that service. The model’s current specialty is CryptoPunk prices, but Yakovenko and his team members plan to roll out a Bored Ape Yacht Club price predictor in the coming weeks. The company just raised a $4 million funding round announced on Thursday.

Users of the website can search for any individual NFT in the available collections, see their valuation history alongside its bids, offers and sales. The company also has a Twitter bot that sends out alerts for notable listings alongside real-time price estimates.

The model’s strength is that it looks at data beyond just sale prices, which on their own are a poor measure of an NFT’s value. More important than sales are the prices for active bids and listings, Yakovenko says. While a sale price determines how much a buyer was willing to pay for an NFT, listings that go untouched are equally telling for how much they aren’t willing to pay.

Yakovenko’s journey into NFTs is in many ways emblematic of the industry’s eclectic nature. NFTs, now just a few years old, aren’t a subject you study in school, and their attraction has taken most market participants by surprise. The type of people NFTs attract, too, have some commonalities – they’re comfortable with taking risks, albeit calculated ones.

“Between NFT Twitter and NFT Telegram groups, half the people you meet are former poker players,” Yakovenko said. “I’ve crossed paths with a good number of them.”

Just weeks ago, Yakovenko, who is now based in crypto hub city Miami, found himself once again back at a New York City poker table, this time filled with crypto friends in town for NFT.NYC, the industry’s premier conference.

“They were good games, even though I lost,” Yakovenko said with a grin. “I guess I got a little unlucky.”


Read more about

DISCLOSURE

Please note that our privacy policy, terms of use, cookies, and do not sell my personal information has been updated.

The leader in news and information on cryptocurrency, digital assets and the future of money, CoinDesk is a media outlet that strives for the highest journalistic standards and abides by a strict set of editorial policies. CoinDesk is an independent operating subsidiary of Digital Currency Group, which invests in cryptocurrencies and blockchain startups. As part of their compensation, certain CoinDesk employees, including editorial employees, may receive exposure to DCG equity in the form of stock appreciation rights, which vest over a multi-year period. CoinDesk journalists are not allowed to purchase stock outright in DCG.

CoinDesk - Unknown

Eli is a news reporter for CoinDesk. He holds ETH, SOL and AVAX.

CoinDesk - Unknown

Eli is a news reporter for CoinDesk. He holds ETH, SOL and AVAX.

Investing in the Future of the Digital Economy
October 18-19 | Spring Studio, NYC