Discussion of
Trust in DeFi: An Empirical Study
of the Decentralized Exchange

Paper by Jianlei Han, Shiyang Huang, and Zhuo Zhong
Discussion by Andreas Park

CEX vs DEX?

Research Question

Does "trust" in the exchange operation determine where price discovery happens? Centralized Stinky Binance
vs.
Decentralized Angelic UniSwap

Main question

trust drives price discovery

Answer

Use the venue whose price you trust the most
Can't trust Binance as much because of wash trading

Premise

Some Data: Cryptomarkets and Binance

= corrects for wash trading

since the 2020 DeFi Summer: Automated Market Makers

New institutions!

passive "shared" liquidity provision
new pricing function

What's an AMM? Liquidity Pools

a closer look at UniSwap

zoomed in

Observations

Mechanism

price

$=$

truth?

UniSwap is right

$\Rightarrow$ sell on Binance

$t=1$

$t=2$

$t=3$

Binance is right

$\Rightarrow$ buy on UniSwap

Truth is in the middle?

price

$=$

$t=1$

$t=2$

$t=3$

$\Rightarrow$ sell on Binance and buy on UniSwap
$\Rightarrow$ $-$ve to own and $+$ to other
Can correlations be any other way?

truth

What do they do?

Section 2:

order imbalance react to "other" venue to push own price in direction of other venue

Section 3:

User base on UniSwap increases $\Rightarrow$ Binance prices react more to UniSwap price changes
"address age": pools with older addresses contribute more to price discovery
"fake" volume (mean absolute deviation): more fake volume, more influence of the UniSwap price

Section 4:

Causality through yield farming
$=$ exogenous shock to user base
$\ldots$ I'm not sure I can follow the reasoning $\ldots$

Trading volume evolution

The paper's main motivating thesis: lack of trust in Binance because of wash trading

Observation 1: UniSwap and Binance volumes are not far apart in the sample horizon (although Binance $\approx 5 \times$ UniSwap

Observation 2: Binance volume grew dramatically since 2021 - lack of trust?

Liquidity Provision

new information arrives

centralized exchange

automated market maker

people trade with marketable orders against stale limit orders

people adjust limit orders

people remove valuable tokens from the pool for cheap

liquidity removal changes price slope, not the price level

Liquidity deposits make AMMs much cheaper

total liquidity $\approx$ $167M

fees UniSwap: 5 bps

fees Binance: 10 bps

Price Changes

centralized exchange

automated market maker

trade with limit order to get a better price

professional LP who is at the top of the book whenever it's opportune

liquidity provision is an alternative to make money work in buy-and-hold

high frequency market maker

liquidity depositors rarely changes their positions

Capponi & Jia 2021
Lehar & Parlour (Uniswap V2) 2021
Lehar, Parlour & Zoican (Uniswap V2) 2022

Why does this matter?

you need capital to move AMM price
$\Rightarrow$ price adjustment speed must be slower
suggested research question: who is the leader, who the follower, and what does the liquidity level have to do with it?

The Event (Liquidity Mining)