PT  - JOURNAL ARTICLE
AU  - Milan Borkovec
AU  - Konstantin Tyurin
TI  - Are Expected Costs and Returns Identical Twins? <em>Decoupling Slippage from Momentum over Shorter Horizons</em>
AID  - 10.3905/joi.2020.1.136
DP  - 2020 Jul 31
TA  - The Journal of Investing
PG  - 97--116
VI  - 29
IP  - 5
4099  - https://pm-research.com/content/29/5/97.short
4100  - https://pm-research.com/content/29/5/97.full
AB  - Does the quest for best execution on behalf of large institutional orders boil down to optimizing the short-term momentum during their trading? We clip the trades originating from buy-side clients’ orders into 5- and 30-minute clusters and find that (1) mostly favorable short-term returns experienced by clients’ fills are counterbalanced by adverse returns when executions of parent orders in progress are postponed; (2) although cost curve heuristics for short and longer horizon clusters are mostly alive and well, the “identical twins” positive cost-return relation over longer horizons tends to break down because of adverse selection dominating executions in shorter intervals; and (3) the nature and strength of the cost-return link vary with liquidity, duration, and type of trading intervals. The distinct “signature” patterns behind price trajectories around child order executions vary systematically by spread capture and the algorithmic strategy in place. Multiresolution time frames, trading/nontrading interval taxonomy, and other methodological tools have practical implications for posttrade modeling and transaction cost analytics.TOPICS: Fundamental equity analysis, accounting and ratio analysis, technical analysisKey Findings• Significant price impact occurs in intervals without own trading.• In contrast to traditional transaction cost analytics at parent order level, no “identical twins” relation between slippage cost and signed returns is found in granular fixed duration (5-minute) intervals.• Although traditional “cost curve” heuristics are (mostly) alive and well, multiple timeframe analytics (full-day, 30-minute, 5-minute, tick-level) and signature price profiles bring new insights into algorithmic trading behavior.