Measuring the effect of a single event on the stock market

Question

Due to the already extremely random model of the stock market, this is a bit of a tough ask, but is it possible to attempt to measure the affect of a single annual recurring event on a specific companies price?

For example, if the google I/O annual developer's meeting affects the price at all? I plan on doing this for multiple different companies, and I plan on factoring in the conference date, to look at the prices around the weeks of the conference, as well as whether or not products were announced at these events.

I am thinking that the correct path to take here is an intervention analysis by taking the data before the event and generating an arima model, then using that model to forecast values after the conference, then difference the actual and forecasted values.

Answer 1

I suggest you could look here: https://vk.com/doc8733459_342746046?hash=7f1ab7634db4cb6828&dl=2086b693480c5a7d41

It is a paper presentation by Prof. Leonidov on the Econophysics or Statistical Enonomics, whatever you prefer. I particularly refer to slides 14 and 15 on exogenous effects to the stock price. They analyzed many years of news feed affecting the stocks. They call the problem, along with the physics conventions, a relaxation. It is an averaged (over many events) process of price volatility returning to its "normal" level following any significant news. They found actually that the news effect is an order or 2 orders less meaningful than the price dynamics itself (indigenous processes).

Just to give you an idea how to approach this on a high level. I am sorry that half of that paper is in Russian, but you could search through preprints authored by Leonidov. It is quite a mind blowing reading...