Machine learning in trading: theory, models, practice and algo-trading - page 3593
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
The next floor is interesting - not the regulations of the Central Banks of various countries, but the theoretical models on which these documents are based. Obviously, it is full of poorly performing predictive models, but one of the ideas of MO (ensemble) is precisely to assemble a strong predictor on the basis of weak ones. Imho, only by putting together everything available can we hope for something stable (or rather, with as much stability as we can afford).
I've already written that I don't believe in "teamwork" outside of monetary and contractual terms. I'm more inclined to just start a thread on the topic and try to find a couple of pearls in the pile of dung that is bound to form there).
You want to jump a notch higher without examining the materials of the notch below - that's not how understanding will come. I think that the Central Bank uses relatively conservative forecasting methods, and their methodology is partially described in open sources. For example, there is a lot of information here on the models used by the CBR, I think all banks use something similar. Here from there:
"
For short-term forecasting of the ruble exchange rate we use the model of short-term dynamics of the exchange rate (Generalised autoregressive conditional heteroscedasticity
model (GARCH), which is based on daily data and is able to take into account the most timely information about the influence of financial market indicators on the formation of the ruble exchange rate.
on the ruble exchange rate formation
"
As I understand, different heads of the Central Bank adhere to different schools of economics, and their team will work in their paradigm, which should also be taken into account, as different works of economists may highlight different criteria for inflation detection and forecasting.
"
The Bank of Russia uses econometric models and expert judgement for short-term forecasting. They use a very wide range of data and other information on the current and expected dynamics of domestic economic conditions, financial markets, and the world economy and are an important part of forming a comprehensive picture of economic development. The results of the short-term forecast have both independent value for forming a picture of the economy in the near term and are used as input parameters ("starting conditions") in medium-term forecasting models.
"
So, the way I see the challenge is to first understand what is an important indicator for the Central Bank and how that indicator affects their model. It is clear that publicly users receive information later for more than one day, so ideally these indicators should be forecasted based on significant factors - for example, reports of large companies.
This all requires a lot of work, which must be done regularly - to be constantly in the news economic and statistical events, as well as to take into account the factors of politics. It is not a job for a single person.
You want to jump up a notch without studying the materials of the notch below - that's not the way to understand. I think the Central Bank uses relatively conservative forecasting methods, and their methodology is partially described in open sources. For example, there is a lot of information here on the models used by the CBR, I think all banks use something similar. Here from there:
This all requires a lot of work, which must be done regularly - to be constantly in the news economic and statistical events, as well as to take into account the factors of politics. It's not a job for a loner.
Well, I wrote recently that the task of a complete description of market behaviour for all values of signs is really unsolvable. But the search for small working "pieces" is quite possible for many people.
I'm not against teams, I'm just not ready to participate in them in a free format. I lack "enthusiasm")
That's how we put an end to our favourite curvafitting :)
You just need to diversify a bit sometimes, change your position)
Without general theories (about balances, parities, etc.) we will have only a huge unordered pile of facts. There are basic models and secondary models that refine the former.
Do you want to predict expert opinion? In general - of course you need to understand the different schools - that's what I wrote about.
Best to start by getting to know a mid-level Central Bank economist :)