Execution speed on the NYSE hybrid

Discussion in 'Order Execution' started by traitor786, Mar 15, 2013.

  1. Bumped in to this

    http://faculty.haas.berkeley.edu/hender/Hybrid.pdf


    A couple of things that stand out while I skim through it

    "Market and marketable limit orders are now automatically executed by default,
    rather than requiring a special code"

    "In Hybrid, the NYSE also introduced Liquidity Replenishment Points (LRPs), which
    are stock-specific price ranges intended to defend against erroneous trades and dampen
    volatility by converting the market from fast (automatic execution available) to ‘‘auction
    only’’ (auction mechanism, no automatic execution available) when prices move quickly
    in either direction. Immediately following Hybrid introduction, the market was fast 98.9%"


    "In addition to reducing execution time, the expansion of automatic execution reduces
    the opportunities for specialists and floor brokers to participate manually in executions.
    Another important set of changes in Hybrid gives the specialist and floor brokers ways to
    participate electronically that correspond to their prior trading capabilities: placing
    undisplayed as well as displayed orders on the limit order book. In addition, the specialist
    for each stock can use a proprietary algorithm to interact electronically with customer order flow, subject to a set of rules intended to replicate in an electronic framework what
    the specialist is allowed to do manually in the auction market"


    "decline in the floor’s last-mover advantage should
    decrease the competition faced by limit order submitters, which would make them more
    patient."

    "t the Hybrid Market raises the cost
    of immediacy (the effective spread) by about 10%"


    ", but the move to faster
    electronic trading raises the cost of immediacy via adverse selection"


    Hasbrouck (1991a, b) introduces a Vector Autoregression (VAR) based model that
    makes almost no structural assumptions about the nature of information or order flow,
    but instead infers the nature of information and trading from the observed sequence of
    prices and orders. In this framework, all stock price moves are assigned to one of two
    categories: They are either associated or unassociated with a recent trade. Although the
    model does not make any structural assumptions about the nature of information, we
    usually refer to price moves as private-information-based if they are associated with a
    recent trade. Price moves that are orthogonal to recent trade arrivals are sometimes


    considered to be based on public information [examples of this interpretation include
    Jones, Kaul, and Lipson (1994) and Barclay and Hendershott (2003)].
    We construct a VAR with two equations to separate price moves into trade-related and
    trade-unrelated components. The first equation describes the trade-by-trade evolution of
    the quote midpoint, while the second equation describes the persistence of order flow.
    Define qj,t to be the buy-sell indicator for trade t in stock j (þ1 for buys, 1 for sells), and
    define rj,t to be the log return based on the quote midpoint of stock j from trade t1 to
    trade t. The VAR picks up order flow dependence out to 10 lags


    "We construct a VAR with two equations to separate price moves into trade-related and
    trade-unrelated components. The first equation describes the trade-by-trade evolution of
    the quote midpoint, while the second equation describes the persistence of order flow.
    Define qj,t to be the buy-sell indicator for trade t in stock j (þ1 for buys, 1 for sells), and
    define rj,t to be the log return based on the quote midpoint of stock j from trade t1 to
    trade t. The VAR picks up order flow dependence out to 10 lags"

    "Using a similar VAR, Hasbrouck (1993) decomposes price changes into their random walk
    (permanent price change) and transitory (pricing error) changes and calculates a lower
    bound on the pricing error. Because the pricing error has zero mean, its volatility is used to
    measure the magnitude of the pricing error. By using midquote prices/return in our VAR,
    we remove the effects of the increase in spreads and focus on the efficiency of quotes.
    Estimating the VAR on a daily basis occasionally results in large outliers (e.g., impulse
    responses that differ from the mean by more than 10 standard deviations). Therefore, we"

    Yes, ORTHOGONAL. It had to pop up !