Methods and Systems for Low Latency Automated Trading Using a Hedging Strategy

Abstract

Disclosed herein are automated trading engine embodiments that operate on market data and re-engineer trading logic to operate on computational resources that are capable of providing highly parallelized and pipelined processing operations to improve tick to trade latency. As an example, logic resources for the automated trading engine can implement hedging strategies to place hedging orders on markets when defined conditions are met. Such hedging strategies can be combined with aggressing strategies to hedge against aggressing orders placed by the automated trading engine.

Description

CROSS-REFERENCE AND PRIORITY CLAIM TO RELATED PATENT APPLICATIONS[0001]This patent application claims priority to U.S. provisional patent application Ser. No. 63 / 149,904, filed Feb. 16, 2021, and entitled “Methods and Systems for Low Latency Automated Trading”, the entire disclosure of which is incorporated herein by reference.[0002]This patent application is related to (1) U.S. patent application Ser. No. ______, filed this same day, and entitled “Methods and Systems for Market Making at Low Latency” (said patent application being identified by Thompson Coburn Attorney Docket Number 44826-215369), (2) U.S. patent application Ser. No. ______, filed this same day, and entitled “Methods and Systems for Joining Market Making at Low Latency” (said patent application being identified by Thompson Coburn Attorney Docket Number 44826-217124), (3) U.S. patent application Ser. No. ______, filed this same day, and entitled “Methods and Systems for Bettering Market Making at Low Latency” (said ...

AI Technical Summary

Benefits of technology

[0007]Market making is a trading strategy that continuously posts offers to both buy and sell a financial instrument with the goal of achieving an aggregate sale price higher than an aggregate buy price. This is commonly referred to as “capturing the spread” between the best offer and best bid price. Competition among market makers for these opportunities to capture the spread generally drives offer and bid prices closer together, thus narrowing the spread and reducing the profit on each market making trade. Generally, maximizing market making profits then depends upon maximizing the number of trades performed. It is largely for this reason that market making strategies were some of the first to be automated. Market making strategies that can quickly respond to changes in financial instrument prices and post new best bid and offer (BBO) prices maximize their opportunity to capture the spread on the next transaction.

[0050]To reduce the latency of computing theoretical fair market prices for derivatives, the inventors disclose a computational approach where the theoretical fair market prices for a derivative (which in turn can drive the trading strategies, BBO prices for quotes, exclusion boundaries for market making quotes, and / or aggressing boundaries for aggressing orders) are extrapolated through parallelized and pipelined computations from a feed of “coarse” theoretical fair market prices for the derivative (where these coarse theoretical fair market prices serve as reference prices for the derivative), a feed of “coarse” prices for the financial instrument underlying the derivative (where these coarse underlying prices serve as the reference prices for the underlying that are linked to the derivative's reference price), a feed of “coarse” Greek values for the derivative, and a market data feed that includes real-time pricing for the financial instrument underlying the derivative. The terminology “coarse” in this context refers to a relatively slow refresh rate of these feeds for the reference prices and Greek values relative to the refresh rate of the underlying financial instrument feed. That is, the reference prices for the derivative and the underlying and associated Greek values for the derivative are expected to refresh at a relatively low rate of around one to 100 times per second. By contrast, the market data feed of underlying financial instrument prices is expected to refresh at a much higher rate of around one million to ten million messages per second For highly traded underlying financial instruments, the refresh rate of pricing for a particular underlying financial instrument may be of similar rates to the refresh rate of the underlying market data feed itself. As such, the refresh rate for the market data feed of underlying financial instrument prices will need to be the driver for refreshing the computed theoretical fair market prices of the derivative if real-time theoretical pricing is desired for derivatives. To accomplish this, the extrapolation model for computing theoretical fair market prices of a subject derivative is designed to use a significantly smaller sequence of arithmetic operations than those used by the Black-Scholes or CRR models. In this fashion, the computational latency for computing theoretical fair market prices of a derivative will be on the order of five to fifty nanoseconds, which is sufficiently fast to keep up with the fast pace of new events in the market data feed of underlying financial instrument prices (and significantly faster than the conventional computational approaches discussed above). By keeping such a close watch over the changing theoretical fair pricing of derivatives, this inventive approach helps traders greatly reduce the uncertainty and risk of their automated trading strategies for derivatives.

Problems solved by technology

Competition among market makers for these opportunities to capture the spread generally drives offer and bid prices closer together, thus narrowing the spread and reducing the profit on each market making trade.

Firstly, the price of a derivative instrument generally depends on the price of its underlying instrument.

How to arrive at the fair value of an option contract is complex and computationally intensive.

While the Black-Scholes model is computationally far simpler than CRR, it makes simplistic assumptions, lacks support for American exercise, and therefore lacks of precision for many strategies in modern electronic markets.

In addition to the complexity of computing fair values and the large number of financial instruments, automated trading in derivatives markets presents additional challenges when the underlying financial instruments are traded on a different market (e.g. stocks on the New York Stock Exchange) and in a different physical location than the derivative financial instruments (e.g. stock options on the Chicago Board Options Exchange).

Especially in the case of derivatives market making strategies, it may not be possible for an automated trading strategy to respond to a change in underlying prices by computing new fair prices and sending a message with new mass quotes for thousands of derivative instruments before another market participant is able to send a single message that successfully aggresses against one of those derivatives instruments.

For this reason, automated trading strategies frequently choose to cancel mass quotes or orders when they determine that an undue risk of adverse trades exists.

There may be other conditions, such as a technical outage on the trading venue for underlying instruments, that cause the strategy to cancel all mass quotes and orders for all contracts.

Volatility trading represents one of the most computationally intensive forms of automated trading due to large number of financial instruments, the need to perform fair value calculations for each derivative instrument in order to drive trading logic, and the complexity of each of those fair value calculations.

Furthermore, volatility trading strategies face more risk of adverse trades from aggressing orders of other market participants.

Accordingly, these conventional software-based approaches to automated trading are not adequate for automated trading of derivatives because these conventional software-based approaches are not sufficiently fast.

That is to say that these conventional software-based approaches to automated trading suffer from a technical shortcoming in that they are not able to adequately take into consideration up-to-date pricing information for swiftly changing market conditions in derivatives markets.

While the use of FPGA-based Feed Handlers constitutes a significant improvement over systems that implement the full cycle in software, such alternate approaches still suffer from relatively high tick to trade latencies due to the latency present in the software-based Trading Logic.

All of these steps undesirably result in reduced opportunities to trade and may eliminate incentives to trade in certain contracts or markets.

Furthermore, given the limitations imposed by trading venues on the number of messages per second that may be transmitted by a market participant, conventional GPP-based trading strategies must do one of the following, all of which increase tick-to-trade latencies and thus reduce trading opportunities and increase risk:(1) perform trade decision-making and message scheduling serially on a single GPP platform(2) perform trade decision-making in parallel on multiple GPP platforms and then multiplex messages to the trading venues through another platform that performs scheduling and aggregate risk checks

The inventors believe that such levels of responsiveness for automated trading strategies are simply not something that conventional automated trading systems are technically capable of achieving.

As explained above, techniques such as Black-Scholes or CRR for computing the fair market price of an option can be unduly slow due to the large numbers of iterations that are needed by these techniques to arrive at theoretical fair market prices.

Accordingly, by the time these conventional techniques are able to arrive at a theoretical fair market price, it is almost certain that the pricing for the financial instrument underlying the derivative has changed, in which case the computed theoretical fair market price will be stale and may not accurately reflect market conditions (and a trading strategy that relies on such theoretical fair market price computations will be carrying risks that arise from potential mispricing).

Images