Econometrics: bibliography - page 3
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
Incidentally, Sims did not develop the method, but applied it to macroeconomic data. VAR had been known for decades before him.
"Vector AutoRegression (VAR) is a model of the dynamics of several time series in which the current values of these series depend on the past values of the same time series. The model was proposed by Christopher Sims"(C)
"Christopher Sims (Sims, 1980) created such a construct - vector autoregressions (VARs)."(C)
Yes? I could be wrong. Who developed it?
Sort of. In fact, he wrote an article saying "...why do you use conventional autoregression, it doesn't work. Let's use vectorial! You know, like we do here on the forum.
Well, in fact, that's not what his paper, for which he was awarded a Nobel Prize, says.
"Vector AutoRegression (VAR) is a model of the dynamics of several time series in which the current values of these series depend on the past values of the same time series. The model was proposed by Christopher Sims"(C)
"Christopher Sims (Sims, 1980) created such a construct - vector autoregressions (VARs)."(C)
Yes? I could be wrong. Who developed it?
Well, that's not actually what his Nobel Prize-winning work says.
The problem is that if the forecasting method actually works on the forward, no one will publish it (until it stops working). So what we're seeing is mostly worked out stuff. For example, the same VAR model on historical data can show any accuracy, but again, the selection of parameters... What's there to explain, we already know that)
I don't get it at all - this model is designed for macroeconomic research. It is of interest, for example, to a central bank, not to a trader
The problem is that if a forecasting method really works on a forward, no one will publish it (until it stops working). So what we're seeing is mostly worked out stuff. For example, the same VAR model on historical data can show any accuracy, but again, the selection of parameters... What's there to explain, we already know that)
I have to admit that the links posted above are definite news to me. I now realise that making a model, like VAR, is only part of the job. The prediction that almost any model will give you needs to know how to use it. And that's no less of a problem than making the model itself. That's what the links above are about.
And what's more, these links show that there's a huge amount of literature on time series forecasting, which has especially blossomed after 2008. Six months ago I googled forecasts and got some pathetic set of links, but now it's not like that.
I couldn't agree more.
Classical models are not a worked out material, the same VAR, ARIMA, ARCH etc. They are not models, they are model building tools. So you could argue that a 19 wrench is waste material. Here you are confused with TA, where mass application of an indicator leads to its fading.
Try to build a TS in which the model is selected from the quotient analysis. You will see that in some areas there will be AR, in others ARMA, in others ARMA + GARCH and in some areas there is no maths at all. The larger the set of models you have, the smaller the area where there is nothing to apply.
completely misunderstood - this model is designed for macroeconomic research. It is of interest, for example, to the central bank, not to the trader