Jordi Alexander is an investor, speculator, and writer in the Crypto space. He has a background in poker and fixed-income trading and is now currently CIO at Selini Capital.
In the 5th episode of the Depth and Spread podcast by dYdX, host Ivo Crnkovic-Rubsamen is joined by Jordi Alexander to discuss his views on trading crypto markets, his game theory approach, and what DEXs can do to attract users.
Read our notes below to learn more.
How Did You Become a Game Theory Expert?
Jordi has a background in strategy games, playing chess, bridge, and eventually making a living playing poker.
Did an MBA and transitioned to TradFi futures HFT (High Frequency Trading) before crypto.
Jordi doesn’t have time to memorize task-specific information for each game or task he tries.
Instead, he creates general game theory infrastructure that can be applied to many games, trades, etc.
People who excel in crypto often have a skill set that could also apply to other fields.
Intro to Crypto
Jordi enjoys becoming an expert in a field after applying his general game theory skillset.
Became interested in crypto in 2017.
Had too much success in TradFi markets which prolonged his move to crypto, didn’t fully transition to starting Selini Capital until late 2020.
Game Theory in Crypto
Chess is a game of perfect information, poker is a game of imperfect information.
Crypto market is like poker.
Thinking in probabilistic terms is valuable in crypto.
Need to be consistent in crypto, don’t let worst days negate best days of trading.
Takers have simpler prediction periods than market makers.
Can build rules from intuition, use this to adapt to different market scenarios.
Firms use simulations to create potential market scenarios.
Jordi is very opposed to efficient market hypothesis: there are many inefficiencies in all markets.
There is always edge to be found on all time horizons.
Crypto vs TradFi
TradFi HFT is very secretive, people are cautious of alpha leakage.
Crypto may end up similar to TradFi in this sense.
Selini collaborates with other crypto firms when opportunities present themselves.
Many are initially interested in crypto for trading opportunities but stay for the innovations.
Healthy low-fee exchanges help everyone.
dYdX UX is similar to CEX.
Jordi thinks dYdX incentives for users are interesting, helps users own the protocol which is the thesis of Web3.
Better to have thick order books in a handful of exchanges, than liquidity being spread out across many exchanges.
Selini Capital Structure
Selini prioritizes systematic, scalable trading over discretionary trading.
Initially had same people working on trading and venture deals.
Launching a VC fund with researchers who aren’t involved in markets.
Exchange Design
When leading limit order on passive side, you’re writing an option against yourself.
Market makers start widening bids to counter this.
Maker rebate exists to incentivize people to sell options for 0 by paying them for every trade.
If there’s too much toxic (institutional) flow, benign (retail) flow is negatively impacted.
Some CEXs charge higher rates for institutions to counter this and defend retail.
Good for DEXs to experiment.
DEXs are competing with CEXs for flow.
Some value their privacy, in this case, DEXs are competing with oracle-based DEXs such as GMX or orderbook-based DEXs such as Serum.
With tokenization, users can own a piece of the DEX and be a part of the protocol’s growth.
Important to harness power of community, rather than just creating an efficient orderbook.
dYdX Incentive Plans
Ivo
dYdX allows small traders to trade for free.
Exchanges should prioritize UX for retail, institutions can afford to pay fees.
Some users actually like paying fees, as it makes them eligible to receive DYDX token.
Important to consider twitter sentiment to attract retail to your exchange.
People not experienced with HFT are at disadvantage to design product incentive programs.
While developing dYdX V4, team is considering the issues unique to DEXs including MEV.
Instead of potential trades going through validator network, they could go only to the validator running the matching engine as the order of validators is known ahead of time.
This restricts the amount of people that could potentially view mempool and extract value.