# Finding patterns in data sets

Background Info: There is a game called Roblox. It has two virtual currencies, Robux and Tickets. People can exchange them for each other, but their exchange rates vary based on how much people want for each currency. Essentially it is a virtual economy. I have made a program that records the exchange rate data every 5 minutes, so that means that there are 288 points of data every day.

Objective: So here is the Robux rate data and here is the Tickets rate data. Every 288 data points on both data sets marks 1 day (because it is recorded every 5 minutes). Basically, I want to see if each day's data in either data set has a pattern. I think there will be a pattern because people will exchange less during night, meaning the rates are less-competitive and will possible drop.

Why: The basic idea behind profiting on the market is to get Tickets at a high rate and Robux at a low rate. So if I find that each day, there seems to be higher rates of Tickets at 1:00PM, then I will trade my Robux for Tickets (Which will multiply). Then if there seems to be lower rates of Robux at 1:00AM, then I will trade my Tickets back into Robux (Which will divide).

So is there some algorithm I can use to find patterns in several data sets? Then use that pattern data to find expected rates for the next day?

I don't think you should look for precision in the case of a game, which changes with every patch and evolves over time. I would recommend looking for a pattern visually by graphing the data. You can smooth out noise with simple moving average (https://en.wikipedia.org/wiki/Moving_average) before building graphs. 12 period moving average makes sense (5 minutes * 12 is 1 hour). It will basically average the rate for every hour. You can then visually look for a pattern. It would also make sense to build graphs for every day and see if the same pattern persists.

Also, there will be a limit on how much you can buy/sell at night. When the trading volume is low it is hard to trade large amounts.

It will also make it easier for you to do some fundamental analysis, e.g. to see if patch days have different patterns, holidays, weekends, Friday nights, etc.

• Thanks for the answer. Because people who don't have a paid-membership to Roblox earn 10 Tickets per day as a login reward, most people have Tickets. And when items in the game are released (Like holiday items), people have to trade their Tickets into Robux to buy the holiday item. This creates a more competitive market, raising the Tickets rates. So holidays and other events where themed items are released in the virtual market certainly affect the rates, and I think I will see that in my data. I did make a moving average of 12 in Excel, but I will have to wait a few days to collect more data – GShocked Nov 23 '15 at 10:41
• Cheers! Good luck! – Arthur Tarasov Nov 23 '15 at 10:44