Method and system for forecasting commodity prices using capacity utilization data

• US 20050240539A1
• Filed: 04/22/2004
• Published: 10/27/2005
• Est. Priority Date: 04/22/2004
• Status: Active Grant

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Olavson, Thomas
• Granted Patent: US 7,693,801 B2
• Time in Patent Office: Days
• US Class Current: 705/400
• CPC Class Codes: G06Q 30/0283 Price estimation or determination
• CPC Class Codes: G06Q 40/04 Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange

Abstract

The present invention includes a method and system for creating a price-forecasting tool for a commodity. The method and system comprise providing price data and industry capacity utilization data related to the commodity, and establishing the price-forecasting tool based on a statistical relationship between the price data and the industry capacity utilization data. A method of doing business is disclosed comprising creating a price forecasting tool based on a statistical relationship between price data and industry capacity utilization data of the commodity, and utilizing the price forecasting tool to generate a plurality of scenario prices for the commodity for a plurality of forecast horizons. A computer program product is disclosed for creating a price forecasting tool comprising a computer usable medium having computer readable program means for causing a computer to perform the steps of providing price data and industry capacity utilization data related to the commodity, and establishing the price forecasting tool based on a statistical relationship between the price data and the industry capacity utilization data.

Claims

1. A method of creating a price forecasting tool for a commodity, comprising:

(a) providing price data related to the commodity;

(b) providing industry capacity utilization data related to the commodity; and

(c) establishing the price forecasting tool based on a statistical relationship between the price data and the industry capacity utilization data for a given time horizon.

2. The method of claim 1, wherein the step of providing the price data includes providing historical and current price data and wherein the step of providing the industry capacity utilization data includes providing historical and current industry capacity utilization data.

3. The method of claim 2, wherein the step of establishing the price forecasting tool comprises defining a long-run price trend based on selected price data.

4. The method of claim 3, wherein the selected price data is historical price data.

5. The method of claim 3, wherein the selected price data is future price data based on industry price forecasts.

6. The method of claim 3, wherein the step of establishing the price forecasting tool comprises defining a first relationship between a future price and the long-run price trend.

7. The method of claim 6, wherein the step of establishing the price forecasting tool further comprises defining a second relationship as a pair of relationships between the long-run price trend and a current price, and between the long-run price trend and a previous period price.

8. The method of claim 7, wherein the step of establishing the price forecasting tool comprises defining a long-run industry capacity utilization trend based on the historical industry capacity utilization data.

9. The method of claim 8, wherein the step of establishing the price forecasting tool comprises defining a third relationship between the long-run industry capacity utilization trend and the current industry capacity utilization.

10. The method of claim 9, wherein the step of establishing the price forecasting tool comprises basing the price forecasting tool on a statistical relationship between the first relationship and second and third relationships.

11. The method of claim 10, wherein the step of defining the third relationship comprises running the price forecasting tool through a linear regression process using the historical price data and historical industry capacity utilization data.

12. The method of claim 11, wherein the step of running the price forecasting tool through a linear regression process comprises deriving at least one coefficient for the second and/or third relationships to balance the first relationship with the second and third relationship for a selected time horizon.

13. A method of creating a price forecasting tool for a commodity comprising:

(a) providing historical and current price data and historical and current industry capacity utilization data related to the commodity;

(b) defining a first relationship between a future price and a long-run price trend;

(c) defining a second relationship as a pair of relationships between the historical price data and the current price data, and between the historical price data and a previous period price data;

(d) defining a third relationship between the historical industry capacity utilization data and the current industry capacity utilization data; and

(e) establishing the price forecasting tool as a statistical relationship between the first relationship and the second and third relationships.

14. The method of claim 13, wherein the first relationship is defined by the equation:

z(t+k)=LN[price(t+k)]−

LN[trend price(t+k)], where LN denotes the natural log function, price(t+k) is the future price at a time horizon (t+k) and trend price(t+k) is the trend price at the future time horizon (t+k).

15. The method of claim 14, wherein the second relationship is defined by the equations:

z(t−

1)=LN[price(t−

1)]−

LN[trend price(t−

1)], and z(t)=LN[price(t)]−

LN[trend price(t)] where LN denotes the natural log function, price(t−

1) is the price at a previous time period (t−

1) and trend price(t−

1) is the trend price at the previous time period (t−

1), and where price(t) is the price at a current time period (t) and trend price(t) is the trend price at the current time period (t).

16. The method of claim 13 wherein the third relationship is defined by the following equations:

u_high(t)=max[0, u(t)−

0.5], and u_low(t)=max[0,0.5−

u(t)], where u=N_L/N_T, N_L is the number of months in which utilization has been lower than the current level of utilization and N_T is the total number of months in the sample history.

17. A method of creating a price forecasting tool for a commodity comprising:

(a) providing forecast price data, previous period price data and current price data relating to the commodity;

(b) providing historical and current industry capacity utilization data related to the commodity;

(c) defining a first relationship between a future price and a long-run price trend;

(d) defining a second relationship as a pair of relationships between the forecast price data and the current price data, and between the forecast price data and a previous period price data;

(e) defining a third relationship between the historical industry capacity utilization data and the current industry capacity utilization data; and

(f) establishing the price forecasting tool as a statistical relationship between the first relationship and the second and third relationships.

18. A method of creating a price forecasting tool for a commodity, comprising utilizing the equation:

z(t+k)=a_k*z(t−

1)+b_k*z(t)+c_k*u_high(t)+d_k*u_low(t)

where t is the current time period, t+k is a forecast time period and t−

1 is a prior time period, a_k and b_k are regression coefficients relating to price in the prior time period and the current time period, respectively for a given time horizon (t+k), where c_k and d_k are regression coefficients relating to high and low industry capacity utilization in the current time period with respect to a given time horizon (t+k), where z(t+k)=LN [(price(t+k)]−

LN[trend price(t+k)], where LN denotes the natural log function, price(t+k) is the future price at the selected horizon (t+k) and trend price(t+k) is the future trend price at horizon (t+k), z(t−

1)=LN [(price(t−

1)]−

LN[trend price(t−

1)], price (t−

1) is the price in a previous time period and trend price (t−

1) is the trend price in a previous time period, where z(t)=LN [(price(t)]−

LN[trend price(t)], price (t) is the price in a current time period and trend price (t) is the trend price in a current time period, where u is defined as;

u=N_L/N_T, where N_L is the number of months in which utilization has been lower than the current level of utilization and NT is the total number of months in the sample history, and where u_high(t)=max[0, u(t)−

0.5], and u_low(t)=max[0,0.5−

u(t)].

19. The method of claim 18, wherein a_k, b_k, c_k and d_k are estimated by utilizing linear regression techniques and historical price and industry capacity utilization data.

20. A system for creating a price forecasting tool for a commodity comprising:

(a) first means for providing price data related to the commodity;

(b) second means for providing industry capacity utilization data related to the commodity; and

(c) third means for establishing the price forecasting tool based on a statistical relationship between the price data and the industry capacity utilization data.

21. The system according to claim 20, wherein the first means provides current and historical price data related to the commodity.

22. The system according to claim 21, wherein a long-term price trend is established using the historical price data.

23. The system according to claim 22, wherein a first relationship is established between a future price and a long-run price trend.

24. The system according to claim 23, wherein a second relationship is established by a pair of relationships between the current price data and the long-term price trend, and between a previous period price data and the long-term price trend.

25. The system according to claim 24, wherein the industry capacity utilization data comprises historical industry capacity utilization data and current industry capacity utilization data.

26. The system according to claim 25, wherein a third relationship is established between the historical industry capacity utilization data and the current industry capacity utilization data.

27. The system according to claim 26, wherein a statistical relationship is established between the first relationship and the second and third relationships.

28. The system according to claim 20, wherein the first means provides current price data, previous period price data, and price forecast data related to the commodity.

29. A system for creating a price forecasting tool for a commodity, comprising calculating means for utilizing the equation:

z(t+k)=a_k*z(t−

1)+b_k*z(t)+c_k*u_high(t)+d_k*u_low(t)

where t is the current time period, t+k is a forecast time period and t−

1 is a prior time period, a_k and b_k are regression coefficients relating to price in the prior time period and the current time period, respectively, for a selected time horizon (t+k), where c_k and d_k are regression coefficients relating to high and low industry capacity utilization in the current time period with respect to a selected time horizon (t+k), where z(t+k)=LN [(price(t+k)]−

LN[trend price(t+k)], where LN denotes the natural log function, price(t+k) is the future price at the selected horizon (t+k) and trend price(t+k) is the future trend price at horizon (t+k), z(t−

1)=LN [(price(t−

1)]−

LN[trend price(t−

1)], price (t−

1) is the price in a previous time period and trend price (t−

1) is the trend price in a previous time period, where z(t)=LN [(price(t)]−

LN[trend price(t)], price (t) is the price in a current time period and trend price (t) is the trend price in a current time period, and where u is defined as;

u=N_L/N_T, where N_L is the number of months in which utilization has been lower than the current level of utilization and N_T is the total number of months in the sample history, and where u_high(t)=max[0, u(t)−

0.5], and u_low(t)=max[0,0.5−

u(t)].

30. A method of doing business comprising:

creating a price forecasting tool based on statistical relationship between price data and industry capacity utilization data of the commodity; and

utilizing the price forecasting tool to generate a plurality of scenario prices for the commodity for a plurality of forecast horizons.

31. The method of claim 30, wherein the price forecasting tool includes a long-run average price trend.

32. The method of claim 30, wherein the price forecasting tool includes a market momentum of the commodity.

33. The method of claim 30, wherein the price forecasting tool includes at least one coefficient establishing a statistical relationship between the long-run average price trend and the industry capacity utilization data for the commodity for a selected time horizon.

34. The method of claim 33, wherein the coefficient is determined by a linear regression technique using the price data and the industry capacity utilization data.

35. A computer program product for creating a price forecasting tool, comprising a computer usable medium having computer readable program means for causing a computer to perform the steps of providing price data and industry capacity utilization data related to the commodity and establishing the price forecasting tool based on a statistical relationship between the price data and the industry capacity utilization data.

36. The computer program product of claim 35 further comprising means for utilizing a linear regression technique on the price data to estimate at least one forecasting coefficient related to the price data for a selected time horizon.

37. The computer program product of claim 35 further comprising means for utilizing a linear regression technique on the industry capacity utilization data to estimate at least one forecasting coefficient related to the industry capacity utilization data for a selected time horizon.

38. The computer program product of claim 37, characterized by an algorithm in the form of the following equation:

z(t+k)=a_k*z(t−

1)+b_k*z(t)+c_k*u_high(t)+d_k*u_low(t)

where t is the current time period, t+k is a forecast time period and t−

1 is a prior time period, a_k and b_k are regression coefficients relating to price in the prior time period and the current time period, respectively, for a selected time horizon (t+k), where z(t+k)=LN [(price(t+k)]−

LN[trend price(t+k)], where LN denotes the natural log function, price(t+k) is the future price at the selected horizon (t+k) and trend price(t+k) is the future trend price at horizon (t+k), z(t−

1)=LN [(price(t−

1)]−

LN[trend price(t−

1)], price (t−

1) is the price in a previous time period and trend price (t−

1) is the trend price in a previous time period, where z(t)=LN [(price(t)]−

LN[trend price(t)], price (t) is the price in a current time period and trend price (t) is the trend price in a current time period, where c_k and d_k are regression coefficients relating to high and low industry capacity utilization in the current time period for a selected time horizon (t+k), and where u is defined as;

u=N_L/N_T, where N_L is the number of months in which utilization has been lower than the current level of utilization and NT is the total number of months in the sample history, and where u_high(t)=max[0, u(t)−

0.5], and u_low(t)=max[0,0.5−

u(t)].