High Frequency Trading & Microstructural Cost Effects For Institutional Algorithms

Size: px

Start display at page:

Download "High Frequency Trading & Microstructural Cost Effects For Institutional Algorithms"

Vivien George
6 years ago
Views:

1 High Frequency Trading & Microstructural Cost Effects For Institutional Algorithms

2 Agenda HFT Positives & Negatives Studying the Negatives Analyzing an Institutional Order: Separating Impact & Timing Costs VWAP Algorithms Dark Aggregating Algorithms Anti-Algorithms ETF Cost Attribution Conclusion

3 What is High-Frequency Trading? Generally fall into two broad categories: (1) Electronic Market Making Try to profit from the spread between bid and ask prices Unlike the market makers of old, rarely hold positions over night Also benefit from liquidity rebates paid by exchanges and market centers (2) Statistical Arbitrage Identify and capitalize on inefficient pricing of financial instruments Market efficiency Information arbitrage Inter-venue latency arbitrage Trading ahead of significant order flow

4 Speed is Everything Need ultrafast, high capacity systems to thrive Freshest market data on every tradable security Nasdaq ITCH BATS FASTPITCH vs SIP feed (Securities Information Processor) Absolute most minimum time delay, or latency EMM s send and cancel thousands of orders simultaneously to effectively manage risk

5 Impact of HFTs POSITIVES Liquidity provision - post liquidity when it wouldn t otherwise be available Decreased bid/ask spreads & lower fees from increased competition NEGATIVES HFTs almost always beat traditional market participants to posting liquidity Shift of exchange rebates from tradition B/Ds to HFTs Information Arbitrage Interpositioning Effect beating two natural counterparties to the punch

6 Inter-venue Latency Arbitrage Interpositioning Effect: Case 1 BATS $30.21 Bid $30.25 Offer Nasdaq $30.22 Bid $30.24 Offer $30.24 Bid $30.23 Bid $30.25 Offer $30.23 Bid $30.24 Offer Broker enters $30.23 Bid on Nasdaq, decresaing spread to $0.01 An HFT sees this on his Nasdaq ITCH feed before others relying on the consolidated feed and INSTANTANEOUSLY bids $30.23 on another venue such as BATS, NYSE Arca or Direct Edge Price-Time Priority DOES NOT apply across venues

7 Trading Ahead with Speed Advantage Interpositioning Effect: Case 2 Nasdaq $30.23 Bid $30.26 Offer Broker representing Institution B may want to Buy 300 shares at the market Broker representing Institution A may want to sell 300 shares at $30.25 An HFT might step in and bid $30.24 for 300 shares Then instantaneously turn around and sell at $30.25 to Institution B, beating Institution A s broker to the best offer $30.24 Bid $30.25 Offer

8 QSG Research Helping institutional investors deal with the NEGATIVES Analyze broker placements with execution-level cost methodology Separate the cumulative price impact from timing costs within the Implementation Shortfall framework. Liquidity Charge (a) Cumulative price concessions for liquidity Timing Consequence (b) Price impact of the market during the execution horizon Execution Differential (Implementation Shortfall), = (a + b) The Total Shortfall of the trade from Arrival Price?

9 Liquidity Charge Vs. Execution Differential

10 Impact of HFT on VWAP Algos Do VWAP algorithms cost more to implement trades? Specifically, is this because they must keep up with inflated, interpositioned volume and because of that suffer poor performance in HFT environments? adverse selection

11 VWAP Study - Sample Set QSG client data with algorithm flags 1/1/2009 through 10/23/2009 Over 50 thousand orders More than $18B executed value VWAP and Arrival price algos only Bifurcated sample set by stock price >$10 stocks <$10 stocks Only liquid, easily absorbed orders (0-5% adv)

12 Results: 9/10/ /23/2009

13 Results: 2009 YTD

14 Dark Aggregators Same methodology can be applied across execution channels to measure signaling risk and quality of execution 3 B/D Dark Aggregators 1B 10B cap stocks 5% - 25% ADV 658 observations

15 Anti-Algorithms

16 Derivative instrument ETF Cost Evaluation Calculate NAV of underlying at each client execution

17 Conclusions VWAP algorithms show increased parceling activity and increased adverse tick execution in stocks below $10 when compared to Arrival Price algorithms Costs attributed to Liquidity Charge and Execution Differential also increased, but Liquidity Charge is more pronounced Increased adverse ticks and Liquidity Charges suggest disadvantages at the micro structural level Methodology can be applied to next-generation algorithms to understand liquidity sourcing versus timing expenses, as well as exposure to adverse selection

18 These materials are confidential. Distribution is Prohibited Quantitative Services Group LLC. All Rights Reserved. The examples in this presentation are for information purposes only and do not constitute investment advice or any recommendation to transact. Investors should consider the appropriate professional advice before making an any investment decision.

Interactive Brokers Rule 606 Quarterly Order Routing Report Quarter Ending December 31, 2016

Interactive Brokers Rule 606 Quarterly Order Routing Report Quarter Ending December 31, 2016 I. Introduction Interactive Brokers ( IB ) has prepared this report pursuant to a U.S. Securities and Exchange