1
MEV Reduction�via Batch Auctions
MEV vs HFT
2
MEV and HFT similarity:
They represent a race to exploit market opportunities
| MEV | HFT |
Venue | Blockchain | Traditional exchanges |
How to win the race | Tip/bribe miners | Be the fastest / buying speed tech. from exchanges |
Who suffers from negative externalities? | Front-run traders, worse prices-> retail & institutional investors | Higher spread in markets -> retail & institutional investors |
Batch Auctions as response to HFT
in traditional finance
3
Paper 2015:
“The High-Frequency Trading Arms Race: Frequent Batch Auctions as a Market Design Response”
Eric Budish, Peter Cramton, John Shim
Thesis
4
Frequent Batch Auctions (FBA) are able to reduce the MEV on the application layer
Agenda
5
Agenda
6
Frequent Batch Auctions (FBA)
7
Discretize trading in time
Collect orders into batches
Calculate 1 uniform clearing price
per batch
clearing
price
amount
price
Frequent Batch Auctions (FBA)
8
Discretize trading in time
Collect orders into batches
Calculate 1 uniform clearing price
per batch
clearing
price
amount
price
given in web3
Agenda
9
The inefficiency of sequential trading
in web3
10
Sequential Execution in 1 block
AMM
Efficiency of FBA trading in web3
11
AMM
Leftover Order
Batch Execution
=> Less AMM fees, less slippage
clearing price
Price spreads �example block: 1417421
12
ETH-USDT price occurrences
price spread of 1.4%
Price spreads �example block: 1417421
13
ETH-USDT price occurrences
ETH-USDT price occurrences
Auction execution: Additional price improvement 2.5K$
CoW study: Quantitative impact of FBA
in web3 today
14
Surplus yearly:
~27M$ on ETH-USDT
~28M$ on ETH-USDC
…
if FBA would be used per 1 block
Agenda
15
MEV surface
16
AMM
All orders are
prone to MEV-bots
Buy-Order 1
Buy-Order 2
Buy-Order 3
Sell-Order 4
Leftover Order
Only 1 order
is prone to MEV
Sequential Execution
Batch Execution
Additional MEV protection
17
Less Back-Running:
Optimal trade path is chosen
in the first place
Protocols like CoW Protocol offer also
RFQ trading
RFQ-trading has no MEV surface
Private Submission
Prevent front-running
CoW study: Quantitative impact of FBA
in web3 today
18
~13% CoW volume
13% of the ETH-USDC volume could be fully settled internal in a batch auction without AMM liquidity
Waiting for CoW
&
Market makers integration
will boost the percentage a lot
Agenda
19
Latency
20
Block n-1
Block n
Price
calculation
Order collection
Not accepting new trades
Submission
13 secs
Challenges with auctions
21
Information leakage:
Incentive compatibility
Auction liquidity & AMM price shifts
22
Block n-1
Auction price calculation won’t know AMM price shifts within block n
AMM
AMM
AMM
AMM price shifts cause deficits or surpluses of the auction
Block n (1. tx)
Block n (2. tx auction)
Agenda
23
Winners of FBA model
24
| Should favor batch auctions | Reason/Examples |
Retail investor | ✓ | Better MEV protection, better price (if not time critical) |
Arbitrageurs | x | They need the speed/bribe to extract value |
Market makers (with inventory risk) | x/✓ | AMMs would earn less fee Less risk from order snipping |
Liquidations | ✓ | Example: MakerDAO / English Auction |
DAO investments | ✓ | Example: Yearn is using CowSwap |
Dapp layer MEV contribution
25
If we get adoption in green area, => big MEV reduction on Dapp layer
| Should favor batch auctions | Example/Comment |
Retail investor: | ✓ | Better MEV protection, better price |
Arbitrageurs | x | They need the speed/bribe to extract value |
Market makers | x/✓ | AMMs would earn less fee Less risk from speed snipping |
Liquidations | ✓ | MakerDAO / English Auction |
DAO investments | ✓ | Yearn is using CowSwap |
Thank you for listening
26
This was a lot of theory,� feel free to give it a practical try at
�
27
Appendix
Abstract:
28
MEV reduction on the application layer is important. Batch auctions are a promising venue to reduce the MEV of trading activity. This talk will investigate the benefits, but also all the challenges and shortcomings of auctions usage.
Main criticism for FBT
In traditional
29
Markets are faster than batch time
Incentive compatibility
Challenges with auctions
30
Information leakage:
Incentive compatibility
Batch Auctions as response to HFT
in traditional finance
31
Paper 2015:
“The High-Frequency Trading Arms Race: Frequent Batch Auctions as a Market Design Response”
Eric Budish, Peter Cramton, John Shim
300M$ Cable for 3 milli sec
Incentive compatible
32
1. Batch
2. Batch
Revealing willingness to buy
In on batch
Not revealing willingness to buy
In one batch
p
p_1
p_2
p> (p_1+p_1)/2
Price spreads
example block: 1417421
33
Yearly savings (estimated): 30M$ on ETH-USDT only
Auction execution:
price
Clearing price
Additional surplus 2.5K$
Sequential vs. auction-trading
in web3
34
Price ETH/USDC
Price ETH/USDC
# trades
# trades
Prices in a block spread
over a spectrum.
=> MEV opportunities
With auctions: 1 price per block
=> less MEV
Sequential trade execution
FBA trade execution
Challenges with auctions
35
Auction liquidity
&
AMM price shifts:
Block n-1
Block n
Auction price calculation won’t know AMM price shifts within block n
AMM
AMM
Auctions will suffer deficits or surplus depending on the price changes AMMs