Methods and systems for pricing derivatives at low latency
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 compute updated theoretical fair prices for derivatives at low latency. The automated trading engine can then use such real-time derivative pricing to better drive decision-making by trading strategies implemented by the automated trading engine, such as market making strategies and aggressing strategies.
1. An apparatus for real-time pricing of derivatives, the apparatus comprising:
a processor, wherein the processor comprises a field programmable gate array (FPGA), application-specific integrated circuit (ASIC), or graphics processing unit (GPU), wherein the FPGA, ASIC, or GPU implements a market making engine, wherein the market making engine comprises a pricing engine for a plurality of derivatives;
wherein the processor is configured to (1) receive reference prices over time for the derivatives, (2) receive reference prices over time for the financial instruments underlying the derivatives, (3) receive reference values for at least one Greek over time for the derivatives, and (4) receive a plurality of current market prices over time for the financial instruments underlying the derivatives, wherein the received current market prices for the underlying financial instruments refresh at a rate;
wherein the FPGA, ASIC, or GPU is configured to generate a plurality of theoretical prices for the derivatives using a parallelized and non-iterative extrapolation model that computes the theoretical prices based on (1) the received reference prices for the derivatives, (2) the received reference prices for the financial instruments underlying the derivatives, (3) the received reference values for the at least one Greek, and (4) the received current market prices for the financial instruments underlying the derivatives;
wherein the pricing engine is further configured to refresh the generated theoretical prices for the derivatives using the parallelized and non-iterative extrapolation model in response to changes in the current market prices for the underlying financial instruments and/or changes in the reference prices or values; and
wherein the market making engine is configured to generate quotes for derivatives based on the generated theoretical prices in combination with defined offsets that establish minimum spreads for the generated quotes.
2. The apparatus of claim 1 wherein the pricing engine is further configured to refresh the generated theoretical prices for the derivatives using the parallelized and non-iterative extrapolation model at a computational latency of 5 to 50 nanoseconds.
3. The apparatus of claim 1 wherein a refresh rate for the current market prices for the underlying financial instruments is a first rate, wherein the received current reference prices for the derivatives and the underlying financial instruments refresh at a second rate, wherein the received current Greek values refresh at the second rate or a third rate, and wherein the first rate is faster than the second rate or faster than the second rate and the third rate.
4. The apparatus of claim 3 wherein the first rate is a value in a range between 1 million to 10 million updates per second, and wherein the second rate is a value in a range between 1 to 100 updates per second.
5. The apparatus of claim 4 wherein the received current Greek values refresh at the third rate, and wherein the third rate is a value in a range between 1 to 100 updates per second.
6. The apparatus of claim 1 wherein the at least one Greek comprises a delta that reflects a rate of change between pricing for a subject derivative relative to pricing for the financial instrument underlying the subject derivative.
7. The apparatus of claim 1 wherein the FPGA, ASIC, or GPU comprises a plurality of logic resources that operate in parallel with each other to implement the parallelized and non-iterative extrapolation model, wherein the logic resources implement different portions of the parallelized and non-iterative extrapolation model.
8. An apparatus for real-time pricing of derivatives, the apparatus comprising:
a processor having parallelized computational resources that implement a pricing engine for a plurality of derivatives;
wherein the processor is configured to (1) receive reference prices over time for the derivatives, (2) receive reference prices over time for the financial instruments underlying the derivatives, (3) receive reference values for at least one Greek over time for the derivatives, and (4) receive a plurality of current market prices over time for the financial instruments underlying the derivatives, wherein the received current market prices for the underlying financial instruments refresh at a rate;
wherein the parallelized computational resources are configured to, for each of a plurality of the derivatives, generate a theoretical price for a derivative based on a model that uses the current reference price for that derivative, the current reference price for the financial instrument underlying that derivative, the current reference Greek value for that derivative, and a difference in price for the current reference price for the financial instrument underlying that derivative relative to a new received current market price for the financial instrument underlying that derivative;
wherein the pricing engine is further configured to refresh the generated theoretical prices for the derivatives in response to changes in the current market prices for the underlying financial instruments and/or changes in the reference prices or values;
wherein the parallelized computational resources comprise a plurality of logic resources that operate in parallel with each other to implement the model, wherein the logic resources comprise:
first logic resources configured to compute a difference between (1) the current market price for the financial instrument underlying a subject derivative and (2) the current reference price for the financial instrument underlying the subject derivative;
second logic resources configured to multiply (1) the computed difference for the subject derivative by (2) the current reference Greek value for the subject derivative to generate a product of the computed difference and the current reference Greek value for the subject derivative; and
third logic resources configured to sum (1) the product and (2) the current reference price for the subject derivative to thereby generate a current theoretical price for the subject derivative.
9. The apparatus of claim 1 wherein the generated theoretical prices comprise best bid and offer (BBO) prices for the derivatives.
10. The apparatus of claim 1 wherein the market making engine is further configured to define exclusion boundaries for a plurality of joining and/or bettering quotes based on a plurality of the generated theoretical prices.
11. An apparatus for real-time pricing of derivatives, the apparatus comprising:
a processor, wherein the processor comprises a field programmable gate array (FPGA), application-specific integrated circuit (ASIC), or graphics processing unit (GPU), wherein the FPGA, ASIC, or GPU implements an automated trading engine, wherein the automated trading engine comprises a pricing engine for a plurality of derivatives;
wherein the processor is configured to (1) receive reference prices over time for the derivatives, (2) receive reference prices over time for the financial instruments underlying the derivatives, (3) receive reference values for at least one Greek over time for the derivatives, and (4) receive a plurality of current market prices over time for the financial instruments underlying the derivatives, wherein the received current market prices for the underlying financial instruments refresh at a rate;
wherein the FPGA, ASIC, or GPU is configured to generate a plurality of theoretical prices for the derivatives using a parallelized and non-iterative extrapolation model that computes the theoretical prices based on (1) the received reference prices for the derivatives, (2) the received reference prices for the financial instruments underlying the derivatives, (3) the received reference values for the at least one Greek, and (4) the received current market prices for the financial instruments underlying the derivatives;
wherein the pricing engine is further configured to refresh the generated theoretical prices for the derivatives using the parallelized and non-iterative extrapolation model in response to changes in the current market prices for the underlying financial instruments and/or changes in the reference prices or values; and
wherein the automated trading engine is further configured to define aggressing zones based on the generated theoretical prices, the aggressing zones used by the automated trading engine in generating derivative orders.
12. The apparatus of claim 1 wherein the processor comprises the FPGA, and wherein the market making engine and the pricing engine are implemented on the FPGA.
13. The apparatus of claim 1 wherein the processor comprises the ASIC, and wherein the market making engine and the pricing engine are implemented on the ASIC.
14. The apparatus of claim 1 wherein the processor comprises the GPU, and wherein the market making engine and the pricing engine are implemented on the GPU.
15. The apparatus of claim 1 wherein the FPGA, ASIC, or GPU comprises a plurality of parallel logic instances of the parallelized and non-iterative extrapolation model that generate the theoretical prices for a plurality of derivatives in parallel with each other.
16. A method for real-time pricing of derivatives, the method comprising:
receiving reference prices over time for a plurality of derivatives, wherein the received reference prices for the derivatives exhibit changes over time;
receiving reference prices over time for the financial instruments underlying the derivatives, wherein the received reference prices for the financial instruments exhibit changes over time;
receiving reference values for at least one Greek over time for the derivatives, wherein the received reference values exhibit changes over time;
receiving a plurality of current market prices over time for the financial instruments underlying the derivatives, wherein the received current market prices for the underlying financial instruments exhibit changes over time, and wherein the received current market prices for the underlying financial instruments refresh at a rate;
for each of a plurality of the derivatives, generating a theoretical price for a derivative based on a model that uses the current reference price for that derivative, the current reference price for the financial instrument underlying that derivative, the current reference Greek value for that derivative, and a difference in price for the current reference price for the financial instrument underlying that derivative relative to a new received current market price for the financial instrument underlying that derivative; and
refreshing the generated theoretical prices for the derivatives in response to the changes in the current market prices for the underlying financial instruments and/or the changes in the reference prices or values;
wherein the parallelized computational resources comprise first logic resources, second logic resources, and third logic resources that operate in parallel with each other to implement the model, and wherein the generating step comprises:
first logic resources computing a difference between (1) the current market price for the financial instrument underlying a subject derivative and (2) the current reference price for the financial instrument underlying the subject derivative;
second logic resources multiplying (1) the computed difference for the subject derivative by (2) the current reference Greek value for the subject derivative to generate a product of the computed difference and the current reference Greek value for the subject derivative; and
third logic resources summing (1) the product and (2) the current reference price for the subject derivative to thereby generate a current theoretical price for the subject derivative; and
wherein the generating and refreshing steps are performed by parallelized computational resources on a processor.
17. The method of claim 16 wherein the refreshing step comprises refreshing the generated theoretical prices for the derivatives using the model at a computational latency of 5 to 50 nanoseconds.
18. The method of claim 16 wherein the processor comprises a field programmable gate array (FPGA) on which the parallelized computational resources are resident.
19. The method of claim 16 wherein the processor comprises an application-specific integrated circuit (ASIC) on which the parallelized computational resources are resident.
20. The apparatus of claim 1 wherein the pricing engine is further configured to refresh the generated theoretical prices for the derivatives using the parallelized and non-iterative extrapolation model in response to changes in the current market prices for the underlying financial instruments.
21. The apparatus of claim 20 wherein the processor is further configured to receive the current market prices for the financial instruments underlying the derivatives as a real-time feed of financial market data, and wherein the pricing engine is further configured to refresh the generated theoretical prices for the derivatives using the parallelized and non-iterative extrapolation model at a rate that keeps up with new market data events in the real-time feed.
22. The apparatus of claim 20 wherein the processor is further configured to receive the current market prices for the financial instruments underlying the derivatives as a real-time feed of financial market data, wherein the real-time feed refreshes at a rate of one million to ten million messages per second, and wherein the pricing engine is further configured to refresh the generated theoretical prices for the derivatives using the parallelized and non-iterative extrapolation model at a computational latency of 5 to 50 nanoseconds.
23. The apparatus of claim 20 wherein the pricing engine is further configured to refresh the generated theoretical prices for the derivatives using the parallelized and non-iterative extrapolation model in response to changes in the current market prices for the underlying financial instruments and changes in the reference prices or values.
24. The apparatus of claim 7 wherein the logic resources comprise:
first logic resources configured to compute a difference between (1) the current market price for the financial instrument underlying a subject derivative and (2) the current reference price for the financial instrument underlying the subject derivative;
second logic resources configured to multiply (1) the computed difference for the subject derivative by (2) the current reference Greek value for the subject derivative to generate a product of the computed difference and the current reference Greek value for the subject derivative; and
third logic resources configured to sum (1) the product and (2) the current reference price for the subject derivative to thereby generate a current theoretical price for the subject derivative.
25. The apparatus of claim 12 wherein the pricing engine is implemented as a hardware logic pipeline on the FPGA.
26. The apparatus of claim 12 wherein the FPGA comprises a plurality of FPGAs.
27. The apparatus of claim 13 wherein the pricing engine is implemented as a hardware logic pipeline on the ASIC.
28. The apparatus of claim 13 wherein the ASIC comprises a plurality of ASICs.
29. The apparatus of claim 14 wherein the GPU comprises a plurality of GPUs.
30. The apparatus of claim 8 wherein the processor comprises a field programmable gate array (FPGA), and wherein the first, second, and third logic resources are deployed on the FPGA.
31. The apparatus of claim 8 wherein the processor comprises an application-specific integrated circuit (ASIC), and wherein the first, second, and third logic resources are deployed on the ASIC.
32. The apparatus of claim 11 wherein the pricing engine is further configured to refresh the generated theoretical prices for the derivatives using the parallelized and non-iterative extrapolation model in response to changes in the current market prices for the underlying financial instruments.
33. The apparatus of claim 32 wherein the processor is further configured to receive the current market prices for the financial instruments underlying the derivatives as a real-time feed of financial market data, and wherein the pricing engine is further configured to refresh the generated theoretical prices for the derivatives using the parallelized and non-iterative extrapolation model at a rate that keeps up with new market data events in the real-time feed.
34. The apparatus of claim 32 wherein the processor is further configured to receive the current market prices for the financial instruments underlying the derivatives as a real-time feed of financial market data, wherein the real-time feed refreshes at a rate of one million to ten million messages per second, and wherein the pricing engine is further configured to refresh the generated theoretical prices for the derivatives using the parallelized and non-iterative extrapolation model at a computational latency of five to fifty nanoseconds.
35. The apparatus of claim 32 wherein the pricing engine is further configured to refresh the generated theoretical prices for the derivatives using the parallelized and non-iterative extrapolation model in response to changes in the current market prices for the underlying financial instruments and changes in the reference prices or values.
36. The apparatus of claim 11 wherein the pricing engine is further configured to refresh the generated theoretical prices for the derivatives using the parallelized and non-iterative extrapolation model at a computational latency of 5 to 50 nanoseconds.
37. The apparatus of claim 11 wherein the generated theoretical prices comprise best bid and offer (BBO) prices for the derivatives.
38. The apparatus of claim 11 wherein the processor comprises the FPGA, and wherein the automated trading engine and the pricing engine are implemented on the FPGA.
39. The apparatus of claim 38 wherein the FPGA comprises a plurality of FPGAs.
40. The apparatus of claim 11 wherein the processor comprises the ASIC, and wherein the automated trading engine and the pricing engine are implemented on the ASIC.
41. The apparatus of claim 40 wherein the ASIC comprises a plurality of ASICs.
42. The apparatus of claim 11 wherein the processor comprises the GPU, and wherein the automated trading engine and the pricing engine are implemented on the GPU.
43. The apparatus of claim 11 wherein the FPGA, ASIC, or GPU comprises a plurality of logic resources that operate in parallel with each other to implement the parallelized and non-iterative extrapolation model, wherein the logic resources implement different portions of the parallelized and non-iterative extrapolation model.
44. The apparatus of claim 43 wherein the logic resources comprise:
first logic resources configured to compute a difference between (1) the current market price for the financial instrument underlying a subject derivative and (2) the current reference price for the financial instrument underlying the subject derivative;
second logic resources configured to multiply (1) the computed difference for the subject derivative by (2) the current reference Greek value for the subject derivative to generate a product of the computed difference and the current reference Greek value for the subject derivative; and
third logic resources configured to sum (1) the product and (2) the current reference price for the subject derivative to thereby generate a current theoretical price for the subject derivative.
45. The apparatus of claim 11 wherein the FPGA, ASIC, or GPU comprises a plurality of parallel logic instances of the parallelized and non-iterative extrapolation model that generate the theoretical prices for a plurality of derivatives in parallel with each other.