In MMT-02, it is discussed that we need to calculate EV based on past patterns i.e whether to stand at best bid 1 or best bid 2 and calculate the probability of this getting executed based on past data. Some questions:
How to capture tick data for order book and trade data in order to enable us to calculate the probabilities (this is the most important question)?
How many days' data to consider to calculate probabilities?
Suppose today is a high volume day in say TCS and stk is moving up. So shud we try to get data of other high volume days (when stk was up) in TCS to calculate probabilities?
Hello,
Below are the responses:
- How do you capture tick data? If we are using broker's API, we can use API specific solution to capture the tick data. For example, in TBP-01 class, we captured live ticks using the API method. If we are using the REST API (provided by a broker), in almost all cases, we can capture live ticks using WebSockets. If we are a direct trading member of an exchange, we usually capture live ticks using the exchange specific API.
- What kind of live ticks can we capture? That depends on what data is being broadcasted by the broker or exchange. Again, in TBP-01 class, we'd seen that we can get the best ask, best bid, high price, low price, and so on. However, it's not necessary that all brokers would broadcast all kinds of ticks.
- Usually 2 or 3 weeks of true tick data should be enough for such studies.
- Not really. You should not filter high volume or low volume days. Rather consider and study whatever data you have and get insights from the analysis that you do on that data.