Multi-asset portfolio simulation (MAPS)

Patent metadata

Field	Value
Publication number	US-10102581-B2
Application number	US-201414303941-A
Country	US
Kind code	B2
Filing date	Jun 13, 2014
Priority date	Jun 17, 2013
Publication date	Oct 16, 2018
Grant date	Oct 16, 2018

Abstract

An exemplary system according to the present disclosure comprises a computing device that in operation, causes the system to receive financial product or financial portfolio data, map the financial product to a risk factor, execute a risk factor simulation process involving the risk factor, generate product profit and loss values for the financial product or portfolio profit and loss values for the financial portfolio based on the risk factor simulation process, and determine an initial margin for the financial product. The risk factor simulation process can be a filtered historical simulation process.

First claim

The invention claimed is:

1. A computer implemented method of efficiently and accurately modeling linear and non-linear data sets, the method comprising:
receiving as input, by at least one computing device, data defining at least one first financial product belonging to a first data class defining non-linear financial products (“a non-linear data set”) and at least one second financial product belonging to a second data class defining linear financial products (“a linear data set”), said computing device comprising memory and at least one processor executing computer-readable instructions;
empirically modeling, by the at least one computing device, the linear and non-linear data sets by a same empirical modeling process that comprises:
decomposing the at least one first and second financial products into their respective components;
selectively identifying, for each of the at least one first and second financial products, at least one of the respective components that drives profitability, the identified respective components each representing at least one risk factor;
executing a risk factor simulation process involving the at least one risk factor, said risk factor simulation process comprising a filtered historical simulation process;
generating product profit and loss values for the at least one first and second financial products based on output from the risk factor simulation process; and
determining an initial margin for the at least one first and second financial products based on the product profit and loss values.

2. The method of claim 1 , wherein said decomposing comprises mapping each of the at least one first and second financial products to the at least one risk factor identified as driving profitability.

3. The method of claim 1 , wherein the risk factor simulation process further comprises:
retrieving historical pricing data for the at least one risk factor;
determining statistical properties of the historical pricing data;
identifying any co-dependencies between prices that exist within said historical pricing data; and
generating normalized historical pricing data based on said statistical properties and said co-dependencies.

4. The method of claim 3 , wherein the filtered historical simulation process comprises:
executing a co-variance scaled filtered historical simulation that includes:
normalizing the historical pricing data to resemble current market volatility by applying a scaling factor to said historical pricing data, said scaling factor reflecting the statistical properties and co-dependencies of said historical pricing data.

5. The method of claim 3 , wherein generating the product profit and loss values comprises:
calculating, via a pricing model, one or more forecasted prices for the at least one first and second financial products based on the normalized historical pricing data input into said pricing model; and
comparing each of said forecasted prices to a current settlement price of each of the at least one first and second financial products to determine a product profit or loss value associated with each of said forecasted prices.

6. The method of claim 5 , wherein determining the initial margin comprises:
sorting the product profit and loss values, most profitable to least profitable or vice versa; and
selecting the product profit or loss value among the sorted values according to a predetermined confidence level,
wherein the selected product profit or loss value represents said initial margin.

7. The method of claim 6 , wherein the historical pricing data comprises pricing data of the at least one risk factor over a period of at least one-thousand (1,000) days, the method further comprising:
calculating, via said pricing model, one-thousand forecasted prices, each based on the normalized pricing data pertaining to a respective one of the one-thousand days;
determining a product profit or loss value associated with each of the one-thousand forecasted prices by comparing each of the one-thousand forecasted prices to a current settlement price of the at least one first and second financial products;
sorting the product profit and loss values associated with each of the one-thousand forecasted prices from most profitable to least profitable or vice versa; and
identifying a tenth least profitable product profit or loss value,
wherein said tenth least profitable product profit or loss value represents the initial margin, and
wherein said tenth least profitable product profit or loss value represents a ninety-nine percent confidence level.

8. A computer implemented method of efficiently and accurately modeling linear and non-linear data sets, the method comprising:
receiving as input, by at least one computing device, data defining at least one financial portfolio, the at least one financial portfolio comprising at least one first financial product belonging to a first data class defining non-linear financial products (“non-linear data set”) and at least one second financial product belonging to a second data class defining linear financial products (“linear data set”), said computing device comprising memory and at least one processor executing computer-readable instructions;
empirically modeling, by the at least one computing device, the linear and non-linear data sets by a same empirical modeling process that comprises:
decomposing the linear and non-linear data sets and identifying, for each of the at least one first and second financial products, at least one respective component that drives profitability, the identified at least one respective component representing at least one risk factor;
executing a risk factor simulation process involving the at least one risk factor, said risk factor simulation process comprising a filtered historical simulation process;
generating product profit and loss values for the at least one first and second financial products based on output from the risk factor simulation process;
generating portfolio profit and loss values for the at least one financial portfolio based on the product profit and loss values; and
determining an initial margin for the at least one financial portfolio based on the portfolio profit and loss values.

9. The method of claim 8 , wherein said decomposing comprises mapping each of the at least one first and second financial products to the at least one risk factor identified as driving profitability.

10. The method of claim 8 , wherein the risk factor simulation process further comprises:
retrieving historical pricing data for the at least one risk factor;
determining statistical properties of the historical pricing data;
identifying any co-dependencies between prices that exist within said historical pricing data; and
generating normalized historical pricing data based on said statistical properties and said co-dependencies.

11. The method of claim 10 , wherein the filtered historical simulation process comprises a co-variance scaled filtered historical simulation that includes:
normalizing the historical pricing data to resemble current market volatility by applying a scaling factor to said historical data, said scaling factor reflecting the statistical properties and co-dependencies of said historical pricing data.

12. The method of claim 10 , wherein generating product profit and loss values comprises:
calculating, via a pricing model, one or more forecasted prices for the at least one first and second financial products based on the normalized historical pricing data input into said pricing model; and
comparing each of said forecasted prices to a current settlement price of the at least one first and second financial products

Assignees

• Intercontinentalexchange Inc
• Intercontinental Exchange Holdings Inc

Inventors

• Weber Michael G
• Ivanov Stanislav I
• Vice Charles A
• Gibson Michael
• Yang Yunke
• Atre Vikram
• Shapiro Joshua I
• Morrison Judson H
• Shively Bradley D
• Routray Sidhartha
• Hopper Michael A
• Jefson Thomas W
• Cummings R J
• Ensign Jacob S
• O'Shields Jeremy K
• Pounds Stephen R
• Shepherd Eric D J
• Iyigunler Ismail

Classifications

• G06Q40/06 Primary

  Asset management; Financial planning or analysis · CPC title