High Frequency Trading During Flash Crashes: Walk of Fame or Hall of Shame?

Discussion by Katya Malinova

Mario Bellia, Kim Christensen, Aleksey Kolokolov, Loriana Pelizzon, Roberto Reno

Summary

Analyze HFT and non-HFT trading in 37 CAC40 index stocks in 2013 during "flash crashes"

  • granular IDs: "pure" MM HFT, IB HFT, IB-non-HFT
  • Flash crash identification: "drift burst" (Christensen, Oomen, and Reno (2017))

Conclude: "HFTs are not beneficial to the liquidity and efficiency of the stock market during flash crashes";

in particular, when multiple stocks are involved

Comments and Questions: What do MM do during a crash?

Lower their quotes

But: relative to non-MM HFT, MM do lean against the wind!!! (?) 

Comments and Questions: What do MM do during a crash?

MM trading: single-stock flash crashes

IB-HFT MM: buy on average (lean against the wind)!

Pure HFT MM: neutral inventory

IB-HFT MM: sell!

Pure HFT MM: neutral 

MM trading: multi-stock ("systematic") crashes

  • Pure HFT MM: make the market!
  • IB-HFT MM:
    • "lean against the wind" during single-stock crashes
    •  trade in the direction of the crash during multi-stock events
      • "in contrast with their contractual role of liquidity providers" 

Comments and Questions: What do MM do during a crash?

  • Pure HFT MM: make the market!
  • IB-HFT MM:
    • provide liquidity during single-stock crashes
    •  trade in the direction of the crash during multi-stock events
      • "in contrast with their contractual role of liquidity providers" 
  • What are MM contractual obligations? 
    • Quoting at the best?
    • Or (and?) price continuity/liq provision during stress?

Comments and Questions: What do MM do during a crash?

  • What do we expect/want MM to do if trading is informed?
    • Keep quoting at the "wrong" price? (Price efficiency???)
    • Adjust their quotes to reflect new info?
    • Aggressively move prices to the new true/efficient value? 

Comments and Questions: Profitability

Profit in euro

Systematic

Profit in euro

Single-stock

lean against the wind

balance inventory

Comments and Questions: Profitability

Profit in euro

Systematic

Profit in euro

Single-stock

"trade opportunistically"

balance inventory

  • Different strategies pre-event?
  • Trading n different markets?
  • Trading in different assets?

Overlap in some conclusions with Brogaard, Carrion, Moyaert, Riordan, Shkilko, and Sokolov (2018)  \(\Rightarrow\)

  • Must be clearer on the marginal contribution, differences vs. overlaps

Minor point: measuring MM order imbalance in Euros can be misleading:

  • e.g., buying at the "bottom" of the crash and selling later may help recover, yet may appear as "net selling" 

Comments and Questions: MM activity during a crash

What drives differences in behavior?

  • The authors: "easier to recognize information" during multi-stock events(?)
  • Brogaard et al: likely due to capital constraints/reduced cross-stock hedging opportunities
  • Are "systematic crashes" truly systematic?
    • Changes in the index?
    • Stocks within the same industry?
  • Is the IB/HFT behavior the same for "systematic" vs. "multi-stock"?

Comments and Questions: Systematic vs Single-stock

Comments and Questions: "Flash Crashes"

Flash crash: "drift burst" (Christensen, Oomen, and Reno (2017))

  • price process = Brownian motion
  • drift burst = explosive drift coefficient

Pros and cons of zooming onto "drift bursts"?

  • Extreme price movements (Brogaard et al (2018))?
  • "Volatility bursts"? (Christensen et al (2014))? 
  • "Crashes" ("jumps") in lower frequency data?
    • Liquidity may deteriorate even if the price is "smooth."

Why only "crashes" (downward movements)?

Comments and Questions: Flash Crashes

USDJPN rate around 2011 earthquake

5-minute sampling

Christen et al (2014)

But: liquidity deteriorates

No "crash" intraday

@katyamalinova

malinovk@mcmaster.ca

slides.com/kmalinova

https://sites.google.com/site/katyamalinova/