Method and system for converting order book data into 2D data for machine learning models

1. A method for converting order book data into 2D data for a machine learning model, the method being performed by one or more processors and comprising:

acquiring order book data including data items for a plurality of bid prices and data items for a plurality of ask prices for a stock traded in a first stock exchange according to time from a time point in the past that is a predetermined period of time earlier, to a current time point;

calculating a mid price based on a highest bid price of the plurality of bid prices and a lowest ask price of the plurality of ask prices; and

converting the order book data into data in tensor form having a first axis of times at certain time intervals and a second axis of prices in units of tick, wherein the data in tensor form includes one or more 2D data items having data for a quantity for each of the plurality of ask prices or data for a quantity for each of the plurality of bid prices as values for each of a plurality of coordinates defined according to the times on the first axis and the prices on the second axis, wherein the second axis of prices in units of tick includes the calculated mid price;

inputting the data in tensor form to the machine learning model; and

predicting using the machine learning model, information on a price of the stock at a time point in a future based on the data in tensor form, wherein the machine learning model includes a plurality of successive layers and wherein each of the successive layers configured to reduce the size of a feature map corresponding to an output of a previous layer.

2. The method according to claim 1 , wherein the data in tensor form includes 2D data corresponding to each of a plurality of channels,

the plurality of channels includes a first channel and a second channel, and

the data in tensor form includes:

2D data corresponding to the first channel, which is for a bid price of a stock traded in the first stock exchange according to the time; and

2D data corresponding to the second channel, which is for an ask price of a stock traded in the first stock exchange according to the time.

3. The method according to claim 2 , wherein the data in tensor form further includes a third channel and a fourth channel, and

the data in tensor form further includes:

2D data corresponding to the third channel, which is for the bid price of the stock traded in a second stock exchange according to the time; and

2D data corresponding to the fourth channel, which is for the ask price of the stock traded in the second stock exchange according to the time.

4. The method according to claim 2 , wherein the data in tensor form further includes a fifth channel and a sixth channel, and

the data in tensor form further includes:

2D data including a value indicative of coordinates on the first axis in each of the one or more 2D data items; and

2D data including a value indicative of coordinates on the second axis in each of the one or more 2D data items.

5. The method according to claim 1 , wherein the converting includes:

normalizing a quantity for each of the plurality of ask prices and a quantity for each of the plurality of bid prices; and

associating the normalized quantity with each of a plurality of coordinates defined according to the times on the first axis and the prices on the second axis.

6. The method according to claim 1 , further comprising, as the current time point changes, repeatedly performing the acquiring, the calculating, and the converting with respect to the order book data from the changed current time point to a time point in the past that is the predetermined period earlier.

7. The method according to claim 6 , wherein

prediction information on the price of the stock at the time point in the future is a predetermined period of time after the current time point.

8. The method according to claim 7 , wherein the prediction information includes:

a first class indicating that the price of the stock at the time point in the future would be higher than a price at the current time point;

a second class indicating that the price of the stock at the time point in the future would be the same as the price at the current time point; and

a third class indicating that the price of the stock at the time point in the future would be lower than the price at the current time point.

9. A non-transitory computer-readable recording medium storing instructions that, when executed by one or more processors, cause performance of the method according to claim 1 .

10. A system for converting order book data into 2D data for a machine learning model, comprising:

a memory storing one or more instructions; and

one or more processors configured to execute one or more instructions in the memory to:

acquire order book data including data items for a plurality of bid prices and data items for a plurality of ask prices for a stock traded in a first stock exchange according to time from a time point in the past that is a predetermined period of time earlier, to a current time point;

calculate a mid price based on a highest bid price of the plurality of bid prices and a lowest ask price of the plurality of ask prices;

convert the order book data into data in tensor form having a first axis of times at certain time intervals and a second axis of prices in units of tick, wherein the data in tensor form includes one or more 2D data items having data for a quantity for each of the plurality of ask prices or data for a quantity for each of the plurality of bid prices as values for each of a plurality of coordinates defined according to the times on the first axis and the prices on the second axis, wherein

the second axis of prices in units of tick includes the calculated mid price;

input the data in tensor form to the machine learning model; and

predict, using the machine learning model, information on a price of the stock at a time point in a future based on the data in tensor form, wherein the machine learning model includes a plurality of successive layers and wherein each of the successive layers configured to reduce the size of a feature map corresponding to an output of a previous layer.

11. The system according to claim 10 , wherein the data in tensor form includes 2D data corresponding to each of a plurality of channels,

the plurality of channels includes a first channel and a second channel, and

the data in tensor form further includes:

2D data corresponding to the first channel, which is for a bid price of a stock traded in the first stock exchange according to the time; and

2D data corresponding to the second channel, which is for an ask price of a stock traded in the first stock exchange according to the time.

12. The system according to claim 11 , wherein the data in tensor form further includes a third channel and a fourth channel, and

the data in tensor form further includes:

2D data corresponding to the third channel, which is for the bid price of the stock traded in a second stock exchange according to the time; and

2D data corresponding to the fourth channel, which is for the ask price of the stock traded in the second stock exchange according to the time.

13. The system according to claim 11 , wherein the data in tensor form further includes a fifth channel and a sixth channel, and

the data in tensor form further includes:

2D data including a value indicative of coordinates on the first axis in each of the one or more 2D data items; and

2D data including a value indicative of coordinates on the second axis in each of the one or more 2D data items.

14. The system according to claim 10 , wherein the converting the order book data into the data in tensor form includes:

normalizing a quantity for each of the plurality of ask prices and a quantity for each of the plurality of bid prices; and

associating the normalized quantity with each of a plurality of coordinates defined according to the times on the first axis and the prices on the second axis.

15. The system according to claim 10 , wherein the one or more processors are further configured to, as the current time point changes, repeatedly perform the acquiring, the calculating, and the converting with respect to the order book data from the changed current time point to a time point in the past that is the predetermined period earlier.

16. The system according to claim 15 , wherein

the time point in the future is a predetermined period of time after the current time point.

17. The system according to claim 16 , wherein the prediction information includes:

a first class indicating that the price of the stock at the time point in the future would be higher than a price at the current time point;

a second class indicating that the price of the stock at the time point in the future would be the same as the price at the current time point; and

a third class indicating that the price of the stock at the time point in the future would be lower than the price at the current time point.