|
on Market Microstructure |
| By: | International Monetary Fund |
| Abstract: | The Unites States financial system includes several systemically important financial market infrastructures (FMIs); they are regulated, supervised, and overseen by multiple authorities. The U.S. FMIs are crucial to U.S. dollar clearing, i.e. the payment systems Fedwire Funds Service and The Clearing House Interbank Payments System (CHIPS), and for the clearing and settlement of U.S. Treasuries, i.e., the Fedwire Securities Service and the Fixed Income Clearing Corporation (FICC). Central counterparties (CCPs) that clear exchange-traded or over-the-counter (OTC) corporate securities or derivatives are of key importance to the safe and efficient functioning of these (global) markets. Disruption of critical operations at one of the large U.S. FMIs may spread to its participants, other FMIs, markets, and throughout the U.S. and global financial systems. The Financial Stability Oversight Council (FSOC) designated eight financial market utilities (FMUs) to be systemically important.1 These designated FMUs are regulated, supervised and overseen by the Federal Reserve Board (FRB), the Securities and Exchange Commission (SEC), or the Commodity Futures Trading Commission (CFTC), depending on their activities. In addition, the Dodd-Frank Wall Street Reform and Consumer Protection Act (DFA) authorized the FRB to promote uniform standards for the management of risks by systemically important FMUs. |
| Keywords: | Central counterparty clearing house;Payment systems;Stress testing;Banking;PFM information systems;ISCR,CR,default fund,liquidity needs,loss waterfall,market participant,recovery plan,securities CCP,stock option |
| Date: | 2020–08–10 |
| URL: | http://d.repec.org/n?u=RePEc:imf:imfscr:2020/249&r=all |
| By: | Aaron Wray; Matthew Meades; Dave Cliff |
| Abstract: | We present results demonstrating that an appropriately configured deep learning neural network (DLNN) can automatically learn to be a high-performing algorithmic trading system, operating purely from training-data inputs generated by passive observation of an existing successful trader T. That is, we can point our black-box DLNN system at trader T and successfully have it learn from T's trading activity, such that it trades at least as well as T. Our system, called DeepTrader, takes inputs derived from Level-2 market data, i.e. the market's Limit Order Book (LOB) or Ladder for a tradeable asset. Unusually, DeepTrader makes no explicit prediction of future prices. Instead, we train it purely on input-output pairs where in each pair the input is a snapshot S of Level-2 LOB data taken at the time when T issued a quote Q (i.e. a bid or an ask order) to the market; and DeepTrader's desired output is to produce Q when it is shown S. That is, we train our DLNN by showing it the LOB data S that T saw at the time when T issued quote Q, and in doing so our system comes to behave like T, acting as an algorithmic trader issuing specific quotes in response to specific LOB conditions. We train DeepTrader on large numbers of these S/Q snapshot/quote pairs, and then test it in a variety of market scenarios, evaluating it against other algorithmic trading systems in the public-domain literature, including two that have repeatedly been shown to outperform human traders. Our results demonstrate that DeepTrader learns to match or outperform such existing algorithmic trading systems. We analyse the successful DeepTrader network to identify what features it is relying on, and which features can be ignored. We propose that our methods can in principle create an explainable copy of an arbitrary trader T via "black-box" deep learning methods. |
| Date: | 2020–11 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2012.00821&r=all |
| By: | Jurkatis, Simon (Bank of England) |
| Abstract: | The reliability of established trade classification algorithms that identify the liquidity demander in financial markets transaction data has been questioned due to an increase in the frequency of quote changes. Hence, this paper proposes a new method. While established algorithms rely on an ad hoc assignment of trades to quotes, the new algorithm actively searches for the quote that matches a trade. Using an ideal data set that identities the liquidity demander I impose various deficiencies to simulate more typical data sets and find that the new method considerably outperforms the existing ones, particularly at lower timestamp precisions: at data timestamped to seconds the misclassification rate is reduced by half. These improvements also carry over into empirical applications. A risk-averse investor would pay up to 33 basis points per annum to base her portfolio allocation on transaction cost estimates obtained from the new method instead of the popular Lee-Ready algorithm. The recently proposed interpolation method (Holden and Jacobsen, 2014) and the bulk volume classification algorithm (Easley, de Prado and O’Hara, 2012), on the other hand, do not offer improvements. |
| Keywords: | Trade classification algorithm; trade initiator; transaction costs; portfolio optimization; limit order book; market microstructure |
| JEL: | G14 G17 G19 |
| Date: | 2020–12–11 |
| URL: | http://d.repec.org/n?u=RePEc:boe:boeewp:0896&r=all |