Nick D

ATTENTION! This is not my native language, so there can be mistakes. Hello. I'm in my last year of University and now it's time to think about the future. At University I was interested in the topic of time series analysis. I've made some research of problem how news impact on time series. And I've found that apart from the news curves in GARCH models, this problem is poorly investigated. But with this news curve, it's hard to predict how the time series will be affected by an event that hasn't happened yet, or just happened(or we can't predict it at all(I hope i am right, because i've studied GARCH models on my own)). So it looks like a good niche for business: to develop a software product that at least approximately assesses, for example, the risk to the stock price caused by the occurrence of particular event. So even if my current implementation, how to predict the behavior of the time series after the occurrence of a particular event, is still raw (I did it myself and no one helped me so sorry), but the fact that inventing a working algorithm will change the way of trading on the stock exchange is obvious. So, how my idea looks now? Now it only works for GARCH models. To predict influence of news on company A, we should: 1)Collect historical data of stock prices of company A, and news related to this company for the same period of time 2)Make a GARCH model for this period of time 3)Classify news 4)To teach the classifier to recognize the news 5)In GARCH models errors = news impact, so we we put in line errors and news by date 6)Calculate total error for class of news So now we have some kind of table news class|news impact, and the trained classifier. When new event will appear (or there will the possibility of a new event), we will know how approximately it will affect the share price. This is a very general description of the idea. And the idea itself is still raw. I've done the tests a couple of times, but unfortunately, they include data related to trade secrets. All I can say right now is that I have reason to believe that this is working idea. Logically, similar events affect the time series in almost the same way: how the disaster affected the previous time, so will the next. From the theory we have the assumption that the errors in GARCH models is the impact of the news. But why am I writing all this here? First, I need supporters. I am a bad programmer, I do not know the theory perfectly, and it is not very easy to do business in my country (so as not to lose it later). I've tried to find supporters among my entourage, but no one wants to take risks. After all, it is much easier to work on a regular job than to try to do something new. But, you know what, there's no risk. There is only question how long will it take to implement this idea. Yes I know that now, perhaps I in many ways am mistaken. But I also know that there is a niche for business and that this algorithm will work maybe in a very modified form. So if you want to create something new, to participate in the creation of history, i need you, join! (I've created a discord server for further discussion and work https://discord.gg/efz46Zj) But what will you receive? After the project will be finished, closed joint-stock company will be created with equal distribution of shares among the participants (maybe this is not a good option for people sitting on the site with the word objectivism in the title, but this is not final yet), and yyou will get the job you like (you wouldn't be involved if you didn't like it, yes.?). And why i am doing all it? For me it is the realization of dream and the opportunity to make history. For me the meaning of life to go down in history. And how can a person with a technical education go down in history if he does not do something new or make a discovery?