Impact and Total Cost to Trade
We were recently asked to examine whether large fund managers, or those executing large trades, were disadvantaged by disproportionate trading costs. We reviewed our H2 2020 trading data and explored the use of dynamic scheduling as an alternative means of controlling costs when trading.
What’s impact cost?
Impact cost is a measure of how much an order is expected to move the share price during an execution. It is a proxy for the cost to trade, which for completeness should also consider the expected independent share price move over the life of an order*.
What we observed
- At each level of participation, impact cost rises with order size, although this is not a linear increase.
- Trading with higher urgency will result in higher impact. At each order size, the corresponding impact is higher at 15% POV than at 10%, which, in turn, is higher than at 5%.
- Impact cost rises with volatility and is greater for smaller capitalized stocks.
- Increasing the rate of participation raises the impact cost at every level of market capitalization.
Mitigating impact cost
We examined the strategy of dynamic scheduling with block crossing as a complementary means of lowering impact cost.
- Block crossing
Adverse selection is the primary reason provided by our Members for why they might not trade a block. This concern derives from the perceived risk of being “run over” by an informed trader or soaking up liquidity from a large trade that would otherwise have moved a share price in your favor.
To limit this risk, the maximum size of a block trade may be capped, pauses between block trading may be scheduled, and trades can be limited to only executing after a tick in the opposite direction to the order side. Our algorithms are enabled with these features and our basket of peers is specifically designed as a reference instrument that has a minimum variance with the expected price performance.
- Dynamic scheduling
Dynamic scheduling algos move between participation rates based upon the performance of a share price against a benchmark. Our Dynamic POV algo is differentiated by having the basket of peers as a benchmark option and our Algo Ranking Model** uses a range of parameters to determine the optimal strategy to execute a trade at a point in time.
Our conclusions
Impact is the most significant cost when trading shares. Order size is the most significant determinant of impact, but the relationship between impact and order size is non-linear. Impact cost may be mitigated by slowing down the rate of participation and by using a dynamic schedule that adjusts to what is happening in the market.
For a given level of impact cost, traders may trade orders in larger capitalized names at a higher level of urgency. This is because volatility, which is an important component of impact cost, is generally higher and hence more costly for smaller capitalized names.
The extension of impact cost to total cost, includes considering the expected price move over the life of an order. Traders should be cognizant of the risk aversion of the initiator of an order and adjust the speed of execution accordingly. The greater the risk-aversion, the higher the cost will be tolerated to complete an order quickly. This should be based on empirical knowledge of the expected price moves in a stock. The Liquidnet Schedule application will perform the necessary calculations and recommend a dynamic schedule designed to minimize the total cost to trade.
The impact and total cost to trade, together with the optimal schedule, change over the course of a day depending on market conditions. This should be considered when comparing post-trade cost analysis of child orders and executions performed by different desks. Orders that are targeted to complete when share price volatility is higher will incur higher cost.
* Our Value-of-the-Block analysis assumes the additional saving of half of the spread at the time of execution. Our model calculates the market impact of a VWAP order as a factor of 30-day close-to-close volatility and the % ADV of the order. This is adjusted to the impact of a specific participation rate by multiplying by the % ADV as a function of the target % POV. The model parameters are calculated from market data rather than in-house transactions.
** Pat. No. 10,600,121. The Liquidnet Algo Ranking Model with Real-Time Course Correction is available for US and Canadian equities only.
The report has been compiled by Simon Maughan, Head of Trading Alpha, Naomi Fernbach and Kristina Link, Alpha Analytics Associates