Market impact (price decrease when selling a large amount) increases roughly as the square root of order size.
In the first quarter of 2018, 11% of stock trading in the US executed in Dark Pools
There are many cases of dark pool operators abusing their knowledge of the trades in the pool. One solution is Multiparty Computation
The work introduces a proof of concept trading system using three algorithms:
All experiments conducted on SCALE-MAMBA MPC System.
They found that while Continuous Double Auction is the algorithm of choice in normal markets, it performs poorly in an MPC context, but were still able to complete 10-50 orders per second.
Periodic auction mechanisms are algorithmically simpler in MPC than CDA.
During the open auction period, orders are collected and sorted. Then, on auction close, a price discovery phase determines the clearing price at which the maximum volume is traded. All trades execute at this clearing price.
The simplest of the three algorithms, volume is matched but the price is set by some external lit market.