This lecture: HFT overview

HFT definition:
* Propreitary automated trading that occurs at speeds faster than an unaided human can achieve
* Profits are inherently dependent on speed advantages
* Propreitary as in: trade with one own's money

Origins of HFT:
* Initially, electronic or automated trading was prohibited.
* Strange workarounds like the Clackatron.

Initial electronic trading:
* Automation of trading and exchanges
* Island as an early example of electronic trading
* Island was a machine friendly venue from the outset
* Origin of the term HFT: Citadel in the early 2000s

Speed advantages:
* Initially, just the automation provided a speed advantage.
* Later, needed to colocate within Island's building.
* Next, needed to colocate within Island itself.
* Example of Tradebot located in Kansas

Scale of HFT:
* Half of all trading in 2015 by soime measures: so, large by trading volume
* Dominant in shares and futures, limited presence in treasuries, very little in forex

HFT firms:
* Small in size (150 is quite big)
* Disproportionate share of trading volume relative to size
* Geared towards rapid development of highly specialized software
* More like a startup, than a large bank
* But common to have multiple trading groups within a firm (will have more to say on this)

Where HFT trading actually happens:
* Datacenters in suburban areas (e.g., Chicago, NJ)
* 2 dozen DCs host the vast majority of HFT
* Most US share trading in 4 DCs in NJ

The essence of HFT:
* Simple signals
* Relatively simple algorithms (some "taking" algorithms are more sophisticated and employ ML for predictions of stock prices)
* When signals are common knowledge, speed is of essence.

The book's central thesis:
* Material political economy
* Which I roughly read as the intersection of
* technology
* politics
* economics

The technology of HFT (we'll have a lot to say about this):
* Co-location
* Communication: communication vendors have now made it easier for startups to compete in the HFT space
* Hardware
* Breaking down abstraction boundaries: algorithms are super concrete now, no longer abstract
* Distributed algorithms interacting with each other AND emergent behavior
(Expand on these)
* Sensing-to-actuation (or tick-to-trade) delay
* It's all about low latency and every little detail matters

HFT from a networked systems lens:
* Interaction of many distributed algorithms
* Latency of message passing is critical to mediating these interactions
* At a very high level, similar to how distributed congestion control or route selection algorithms interact with each other
* Many interesting examples of emergent behavior (Chapter 6)

The politics of HFT:
* Two notions of politics per Mackenzie
  --> Broad: Various actors in the ecosystem and their incentives; incumbent-challenger conflicts
  --> Narrow: political parties, members of Congress, hearings, govt. regulations
* "Last look" as a political outcome
* Roads not taken
* RegNMS
* The national best bid and offer
* Intermarket sweep orders
* Engineering vs. quantitative approaches to trading: making vs. taking
* Political leanings of the government play a role as well: deregulation vs. regulation

Economics of HFT:
* Order books are visible to everyone
* Was not always the case: historically, NASDAQ order books were visible only to specialists who could then talk about it informally.
* Still not the case in dark pools
* Probably 2B total in annual revenue across all HFT firms: so, small significant to the rest of finance
* About 0.27 cents per trade is quite good, but this adds up across multiple trades
* Revenue net of expenses of around 0.05 to 0.1 cents per trade is considered quite good.
* Exchanges benefit from HFT (fees, etc.); HFT benefits from exchanges (colocation, better data etc.)
* Exchanges often lease out communication links 
* Concept of rents and pinch points where people can charge unduly high costs

Futures as an example of how politics affects HFT:
* Futures initially restricted to agricultural commodities
* Need to contemplate delivery in order to distinguish futures from gambling
* Regulation and lobbying allowed futures to break the connection to contemplating delivery; head-on challenge may not have worked.
* Some of this was connected to key changes in the US government.
* Allowed more abstract concepts like futures on share indexes, futures on interest rates, etc.
* Allowed a more expanded role for CME and for futures.
* CME was also one of the early movers in electronic trading, e.g., GLOBEX

Why futures lead:
* Allow high leverage relative to other financial products because of low costs of trading on futures (page 43)
* Stickiness of liquidity: the Chicago futures market has tended to be most liquid over the years

Places where HFT has made limited inroads:
* Good example of political considerations
* HFT is in the inter-dealer market for treasuries
* But clients still send orders directly to dealers
* An all-to-all markets for treasuries in the same style as shares doesn't exist.