Alternative Execution Venues
Liquidity Fragmentation
An aggregator merges quotes from various liquidity providers and market makers
Finds the tightest spread among liquidity providers
Quotes aren't firm
Send requests to trade at best price
Execution styles (sweep vs full amount):
Full amount places entire order with one LP. LP takes on associated risk.
Trade is contained so market impact is reduced.
LPs less likely to aggressively change prices since they control entire trade
Builds trust between client and LP
LP may quote higher prices to account for risk
Capacity limitations
Sweep executions split large trades into smaller parts executed across multiple LPs.
All trades occur nearly simultaneously, leading to a quick market response
LPs must hedge risk quickly which leads to volatile prices and higher transaction costs
Assessing quality of liquidity provider:
We can use the following criteria:
Visible spread determines the initial cost of a transaction. Narrower spread is more competitive pricing
Rejection profile shows the frequency and reasons for trade rejections by the LP
Market impact is the extent to which an LP usually affects overall market price. LPs smoothing flows reduce adverse impacts
Correlation is the similarity of quotes between LPs and the aggregator. Low correlation gives better diversification, high correlation adds redundant liquidity
Dark Pools
Mechanics and order placement strategies:
Mechanics are as follows:
Bid and offers sent without price, just quantity
When opposing interests match, fill is generated at the current mid-price
Can impose a minimum fill quantity or a minimum time before cancellation to avoid arb
Dark pools are used for anonymity, to trade at the mid-price and to trade large blocks without market impact. Tradeoff between this and adverse selection
Strategies are as follows:
HFT arbitrageur can buy deep into primary LOB to raise the price, then sell at the mid in a dark pool which will be greater than purchase cost
We form game theory problem. Toxicity metric is |r-r*| (closer to 0 means more toxic), where r is probability toxic trader submits trade and r* is nash equilibrium
To limit adverse selection:
Set limit prices on dark pool orders that make the strat unprofitable
Use toxicity metric to select execution venue and timing
A genuine trader with access to large data sets is most likely to win (i.e. a large broker)
