Welcome back to our AlgoTrading blog, today we are going to continue talking about how it is possible to automate a program that is able to read news and in particular we will talk about how the words in our libraries will be translated.
Basically each word in the list #2 and #3 of the previous post will be written numerically into two-digits-strings.
The first digit will represent the category in which the word falls into, whilst the second digit will represent the position of that word inside that category (the value of the word).
An empty default sequence will be an empty string and slowely filled, while translating the words we listed, by the values will insert in the sequence, the main thing to remember is that since the default value is zero we won't use it in our translations.
These are the value of the first digits and their correspondent category:
0: Default value 5: # Don't know yet
1: Positively Related to Stock/Option 6: # Don't know yet
2: Negatively Related to Stock/Option 7: # Don't know yet
3: Positive Word Meaning 8: # Don't know yet
4: Negative Word Meaning 9: Reliable Sources' List
The second digit instead, since it will represent the value inside the category is different is linked to a different lists of words for each category that we won't go through today.
Since it is different for every category the value will represent to whcih extent the words is related to the first digit's category here are some examples: category 1 number 9 is the CEO, category 2 number 9 the Primary Rival, cat. 3 num. x, x is how good the word is, and so on, in category 9 the second digit represent how reliable and effective a source of news is.
So basically 9 is the max. value while 1 is the lowest.
This is important because it dictate us some guidelines to calculate the prevision made by the news.
The second digit 9 is considered as 100%, the digit 1 as 10% and all the values inthe middle according to this proportion.
This short explaination is fundamental in order to understand how the compression works, therefore it will help you understand our next post on how to elaborate a compression function that will work with our datas.
This is it for today, hope you understood, see you tomorrow!
StatsApplied