The objective of this project is to produce a predictive model that will identify [90-120 seconds in advance] when a currency pair will be up or down for a period of at least 45 seconds. The data provided is historical but in actual practice, the analysis needs to be done real time with data that enters a database every second.
Sample data for a single currency pair is attached. I can provide MUCH more data for that pair and several other pairs for analysis.
Your job is to analyze the data, notice repeating patterns and match the real time data to the patterns for the purpose of triggering an alarm when a pattern is recognized.
I would like to look for patterns in the data repeat at or about the same time every day or day of week.