Ollie (Oliver) Smith

Wash Block - Investigating Self Match Prevention Abuse on the Chicago Mercantile Exchange (DOI: 10.6084/m9.figshare.30423055)

• Abstract: This paper investigates the misuse of Self- Match Prevention (SMP) mechanisms on the Chicago Mercantile Exchange (CME), originally intended to prohibit wash trades. Drawing on high frequency, tick-level Market-by-Price (MBP) data from the E-mini S&P 500 futures contract, we classify SMP activations and examine their role in facilitating manipulative practices such as momentum ignition and spoofing We formalize detection criteria for SMP events, distinguishing between momentum ignition and position reversals against resting spoof orders. Results from a full Globex session on January 27-28, 2020 reveal over 2,800 SMP activations consistent with momentum ignition, compared with only 10 instances of spoof-related SMP use. While spoofing via SMP appears rare, momentum ignition using
  SMP is prevalent and may distort perceptions of market activity. Our findings suggest that although SMP reduces direct wash trading, its existence provides opportunity for subtle manipulation which may compound to create significant impacts on price discovery, raising important considerations for both regulators and exchange design.

Applied Queuing Dynamics in a Limit Order Book (DOI: 10.6084/m9.figshare.30423196)

• Abstract: This thesis Investigates (LOB) imbalance and queueing dynamics in derivatives markets and proposes a model for forecasting short-term mid-price movements and their extent as a response to queue state, the response of order arrivals in response the queue state and the resulting application of the latter to improving order execution efficiency by reducing implementation shortfall and execution latency. It is found that the queue state, particularly the imbalance of the bid and ask queues is a strong determinant of the direction of limit and market order arrivals. Using the queue imbalance as a condition of price selection when executing a given quantity improves order latency and execution costs whether due to the bid-ask spread or implementation shortfall over primarily market and limit order execution strategies.

If Your Friend Placed an Order Would You? - A Multivariate Hawkes Model in a Limit Order Book Market (DOI: 10.6084/m9.figshare.30511361)

• Abstract: Hawkes processes are self-exciting point processes: after an event occurs, the future rate of occurrence increases by an excitation amount and then decays toward a baseline. The baseline rate is the constant base intensity μ and plays the same role as the constant proportional rate in a Poisson process, while each past event contributes an additional time-dependent excitation to the conditional intensity. Mα shows that bid and ask arrivals are strongly self-exciting, with αAb,Ab ≈ 0.23 and αAa,Aa ≈ 0.21. Cross-excitation between bid and ask is weaker (αAb,Aa ≈ 0.04, αAa,Ab ≈ 0.05), while cancellations influence trades significantly where buy trades respond to bid cancellations with αT b,Cb ≈ 0.18, and sell trades respond to ask cancellations with αT s,Ca ≈ 0.17. The highest excitation occurs in self-events for arrivals and cancellations (αAb,Ab, αAa,Aa, αCb,Cb, αCa,Ca), while trades are more sensitive to cancellation events than to other trades. Trades are moderately excited by arrivals (αT b,Ab ≈ 0.07, αT s,Aa ≈ 0.08), reinforcing the idea that liquidity-taking actions are often triggered by liquidity provision or removal.

Markov Transition in a Limit Order Book Market (DOI: 10.6084/m9.figshare.30423673)

• Abstract: Let the limit order book (LOB) be represented by its bid and ask price levels Bp and Ap and their relevant order queue sizes VB and VA. We do not define indices of bid ask prices due to the constraints of using the highest and lowest (best) bid and offer prices and their sizes respectively. We define 4 possible states of a discrete time markov process, these are Bid Imbalance Bi, Ask Imbalance Ai, Trade at Ask (buy) Bt and trade at bid (sell) St.