High Frequency Trading + Stochastic Latency and Regulation 2.0. Andrei Kirilenko MIT Sloan

High Frequency Trading + Stochastic Latency and Regulation 2.0 Andrei Kirilenko MIT Sloan

High Frequency Trading: Good or Evil? Good Bryan Durkin, Chief Operating Officer, CME Group: "There is considerable evidence that high- frequency traders increase liquidity, narrow spreads and enhance the efficiency of markets. Evil Charlie Munger, Vice Chairman, Berkshire Hathaway: It's legalized front- running. I think it is basically evil and I don't think it should have ever been allowed to reach the size that it did. Why should all of us pay a little group of people to engage in legalized front- running of our orders?

High Frequency Trading: An Asset Manager s Perspective From an e- mal to me: I manage a 40 Act fund inside a major insurance company and see nothing but peril in HFT. The narrowing of spreads that the HFT apologists claim to provide for the rest of us redounds to their bank accounts, not ours. The other side increased volatility, false signaling of volume and investor preference, market dislocations, exchanges divided loyalties, and market stresses are not worth the risk. We are definitely paying for something we do not want.

HFTs and Market Dislocations: The Flash Crash How did High Frequency Traders trade on May 6, 2010? What may have triggered the Flash Crash? What role did HFTs play in the Flash Crash? increased volatility, false signaling of volume and investor preference, market dislocations,

Classifying Traders HFTs: High volume, low inventory, end the day flat Non- HFT Market Maker: Provide liquidity Fundamental (Institutional): Take directional positions Small (Retail): Trade very few contracts Opportunistic: Trade across multiple markets, against a model, during events

Trading Categories Trader Volume 700000 600000 500000 400000 300000 200000 100000 700000 600000 500000 Trader Volume May 3 High Frequency Traders Opportunistic Traders and Intermediaries Trader Volume 700000 600000 500000 400000 300000 200000 100000 May 4 High Frequency Traders Opportunistic Traders and Intermediaries 0 Fundamental Sellers Fundamental Buyers 0 Fundamental Sellers Fundamental Buyers -0.00736-0.00536-0.00336-0.00136 0.00064 0.00264 0.00464 0.00664-0.00736-0.00536-0.00336-0.00136 0.00064 0.00264 0.00464 0.00664 May 5 600000 High Frequency Traders 500000 High Frequency Traders 400000 300000 300000 200000 200000 Opportunistic Traders and Intermediaries 100000 Opportunistic Traders and Intermediaries 100000 Fundamental Sellers Fundamental Buyers 0 Fundamental Sellers Fundamental Buyers 0-0.00736-0.00536-0.00336-0.00136 0.00064 0.00264 0.00464 0.00664-0.00736-0.00536-0.00336-0.00136 0.00064 0.00264 0.00464 0.00664 Net Position Scaled by Market Trading Volume Trader Volume 700000 400000 May 6

HFTs and Market Dislocations: Net Holdings Net Position 2500 1205 2000 May 3 1500 1200 1000 500 0 1195-500 - 1000 1190-1500 - 2000 1185-2500 HFT NP - 3000 Price 1180 8:31 9:21 10:11 11:01 11:51 12:41 13:31 14:21 15:11 Time 5000 4000 3000 2000 1000 0-1000 - 2000-3000 - 4000 May 5 1175 1170 1165 1160 1155 1150-5000 1145 8:31 9:21 10:11 11:01 11:51 12:41 13:31 14:21 15:11 4000 3000 2000 1000 0-1000 - 2000-3000 May 4 1185 1180 1175 1170 1165 1160-4000 1155 8:31 9:21 10:11 11:01 11:51 12:41 13:31 14:21 15:11 4000 3000 2000 1000 0-1000 - 2000-3000 - 4000 May 6 1180 1160 1140 1120 1100 1080 1060-5000 1040 mtime 9:20 10:10 11:00 11:50 12:40 13:30 14:20 15:10

The Flash Crash Large Fundamental Seller hedges exposure in equities Sell Algorithm sell 75,000 E- mini s with 9% volume participation target Size Largest net position of the year executed in about 20 minutes Price Decline sells 35,000 ($1.9 billion) contracts in 13 minutes Cross- Market Arbitrage buy E- mini/sell SPY or basket of equities Across the Board Price Declines trigger automated pauses Lack of Liquidity in Individual Equities systems reset to reflect higher risk Broken Trades in Equities retail stop loss orders executed against stub quotes Source: CFTC- SEC Report on the Events of May 6, 2010

HFTs and Market Dislocations: The Flash Crash On May 6, 2010, HFTs traded the same way as they did on May 3-5: Small inventory, high trading volume, take more liquidity than provide. High Frequency Traders did not cause the Flash Crash. A large, but short lived imbalance between Fundamental Sellers and Fundamental Buyers appeared. Opportunistic Traders held it, but for a massive price concession. Source: CFTC- SEC Report on the Events of May 6, 2010

Number of Contracts 600 400 200 0 200 400 600 Number of Contracts 600 400 200 0 200 400 600 1001.25 1001.00 1000.75 1000.50 ASKS 1001.25 1001.00 1000.75 1000.50 Price 1000.25 1000.00 999.75 999.50 999.25 999.00 BIDS Price 1000.25 1000.00 999.75 999.50 999.25 Number of Contracts Number of Contracts 600 400 200 0 200 400 600 600 400 200 0 200 400 600 999.00 paying for something we do not want. 1001.25 1001.00 1000.75 Selling Pressure Continues: Execute Passively 1001.25 1001.00 1000.75 Selling Pressure Stops: Scratch Trade 1000.50 1000.50 Price 1000.25 1000.00 Price 1000.25 1000.00 999.75 999.75 999.50 999.50 999.25 999.25 999.00 999.00

HFTs under regular market conditions (1) Are HFTs profitable? (2) Do HFTs provide liquidity? (3) Do HFTs bear commensurate risk? The narrowing of spreads that the HFT apologists claim to provide for the rest of us redounds to their bank accounts, not ours.

HFTs: a. high volume b. low inventory Classifying HFTs c. end the day with near zero positions Not all HFTs are the same: a. Aggressive HFT + mostly take liquidity b. Mixed HFT + both take and provide liquidity c. Passive HFT + provide liquidity

Sharpe ratios Sharpe Ra6o = E[ r ] r i SD[ r ] i f * 252 E[ π i ] * SD[ π ] i 252 [Assuming constant capitaliza6on over 6me and r f = 0] 25 HFT- Aggressive Sharpe ra1os 20 15 10 5 0-5 - 10 90% 75% 50% 25% 10% 25 HFT- Passive Sharpe ra1os 20 15 10 5 0-5 - 10 90% 75% 50% 25% 10%

Profit consistency Avg Daily Profits $250,000 $200,000 $150,000 $100,000 $50,000 $0 - $50,000 HFT- Aggressive profit consistency $60,000 HFT- Passive profit consistency $40,000 $20,000 $0 - $20,000 Top 3rd All BoFom 3rd

Providing or Taking Liquidity? $800,000 $600,000 $400,000 $200,000 $0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 p10 p25 p50 p75 p90 - $200,000 - $400,000

HFTs under regular market conditions (1) HFT Profitability: - High profitability, very persistent. - Very high Sharpe ratios and very low inventory. - Large variations in profitability across firms. (2) Sources of HFT Profits: - Over short time horizons. - Aggressive HFTs make money on momentum. - Mixed and Passive HFTs make money on the bid- ask spread. (3) HFT Liquidity Provision: - Large heterogeneity in liquidity provision. - Most profitable HFTs are liquidity takers.

HFTs at times of market stress 1. HFTs trade the same as under regular market conditions. 2. HFTs hot potato trading leads to a spike in trading volume. 3. HFTs exacerbate volatility by aggressively unwinding inventory. HFTs under regular market conditions 1. HFTs earn large, persistent profits, take little risk. 2. HFTs exhibit wide heterogeneity in liquidity provision. 3. HFT profits increase in aggressiveness. Charlie Munger: I think the long term investor is not too much affected by things like the flash crash. That said, I think it is very stupid to allow a system to evolve where half of the trading is a bunch of short term people trying to get information one millionth of a nanosecond ahead of somebody else.

What about price discovery? people trying to get information one millionth of a nanosecond ahead of somebody else can make prices more informative one millionth of a nanosecond sooner. What does it mean sooner? How do we measure the speed of information transmission? What role do HFTs play in price discovery?

Latency Latency is the delay between the occurrence of an event and its manifestation or recording. A standard way to measure latency is by determining the time it takes a given data packet to travel from source to destination and back, the so- called round- trip time or RTT. The data packet we will use is the so- called message. A message is a standardized packet of data that enables a trader and a trading venue to communicate with each other.

Latency of Automated Trading System Three different types of latency. Communication latency is the time it takes for a message to travel between an individual trader s computer and an automated trading venue. Market feed latency is the time it takes for an automated trading venue to disseminate market data out to all market participants. Trading system latency is the time it takes for a message to travel within an automated trading venue from the initial entry to the eventual confirmation going back to the trader.

Measuring Trading System Latency

Trading System Latency: It s Random Variable!

Stochastic Latency: Power Law alpha = 4.28 (0.028) X_min = 1,588 N Tail = 16,803 (18%)

Stochastic Latency: Power Law

Stochastic Latency: Lognormal Mean: 1220.06 Variance: 129.7

Stochastic Latency: A Risk Factor in Automated Markets Suppose that there is a true price process with constant or stochastic volatility. Suppose also that the true price process is observed with a stochastic delay (latency). Stochastic latency (i.e., it s a random variable) increases the volatility of volatility.

Financial Regulation 1.0 Topic from Final Rules insurance Topic from Proposed Rules interpretation index securities agreement vii security commissions special title based commenters definition transactions final relief bankingcftc major advisers entity activities dealers hedge investment frankletter exempt cpos private exchange participant agricultural assets pf act entities dealer order form fund option exemption cea options counterparty identity covered person electric dodd participants institution notice rulesection theftaction swap public whistleblower opt information part counterparties foreign rules consumer swaps reporting asset contract notional funds regulations required market time transaction retail proposed sdr trading registration registered dataclass size commodity block rate affiliate sec compliance trade minimum financial commission risk msps comment merchant sef limits position regulation dcm requirements execution model positions core cl segregation account futures designated fcm customer msp facility customers requirement sds capital board believes fcms contracts cleared management collateral referenced clearing spot note margin member dco derivatives cash physical members dcos supra bona fide month organization price resources ii procedures subpart

Financial Regulation 2.0 Systems- Engineered. Regulate automated markets as complex systems composed of software, hardware, and human personnel; promote best practices in systems design and complexity management. Safeguards- Heavy. Make risk safeguards consistent with the machine- readable communication protocols and operational speeds. Transparency- Rich. Mandate that versions and modifications of the source code that implement each rule are made available to the regulators and potentially the public. Cyber- centric. Change regulatory surveillance and enforcement practices to be more cyber- centric rather than human- centric. Platform- Neutral. Make regulations neutral with respect to computing technologies.