Best Metrics to Predict "average stock price"

Hello,

I am getting into algo trading as a mechanical engineer and would like to see if anyone had any information on the best metric to use for tracking a stocks "average" price and then extrapolating that to predict it's average price in the near future. I will elaborate on this further.

My goal: to use algo trading to make high frequency trades scalping small profits off a stable stock.

My theory: By avoiding the OTC and penny stocks I don't have to contend with extremely high risk, in which in the end I would lose to more experienced investors. I will be taking advantage of low price commission trades to facilitate a net profit from many trades.

In the attached image I show what I mean by average price(BLACK LINE). This black line is what the average price of the stock is if you don't take into account the many fluctuations that happen every minute, the bounds of the fluctuations are the (RED and GREEN) lines. I don't see why I couldn't code an algorithm to calculate what the black line will be and predict what it will be a minute or so in advance. The beauty of this is it doesn't have to be precise because I am not trying to buy at the lowest part of the fluctuation or even sell at the highest point. I am just trying to buy and sell between the bounds to make a small profit. By being able to predict just a little ahead you could theoretically set limit orders to buy and sell rapidly making small profits even though the net stock movement might be zero.

Main question: Are there any metrics out there to track the average or "black" line and then predict what it will be(TEAL line)? I am hoping I don't have to go as far as getting into the order price and size of an order to calculate some coefficient of predicted movement. From calculus I would think that there is an equation that predicts what the price will be(in the near future) based on the derivative of the price chart. Perhaps using a sliding polynomial best fit line.

Graph of "average price"