Hi, I predict the market prices with 5% error. I use ARIMA methodology, among others. I charge for it.

Background of stock market has various factors that affect the prices such as volumes that are traded, economic factors, political factors and trader's sentiment factors such as bullish or bearish. Like @Ajit said, there will always be error but when you account for all the above factors you can minimize it. A good example is Apple stock, when you look for a time period between 2000 to Nov 2018 you will probobly see an exponential curve based on that you can fit a curve to estimate it. Unfortunately due to political trade wars, the stock crashed over 20%. That show how each factor plays an important role in predictability

Source

https://www.investors.com/news/technology/click/apple-stock-iphone-purchase-intent/

Here is an open access article which modeled serial closing price of a randomly selected (utility) stock using maximum-accuracy Markov analysis. Analysis demonstrated the sensitivity of such analyses to weighting by magnitude of change, but unfortunately utility stocks don't move up or down very much in short-term study--they tend to remain stable. I will be following this up with a second analysis of the subsequent three-months of action in this stock. In the meantime I studied optimal weighted Markov using one outcome and also studying two time-lagged variables for individual daily ratings of physical symptoms--which when plotted resemble a sky-to-sky lightning bolt (i.e., zig-zag across time): these worked surprisingly well.

Here is the link to the stock price study which started this program of research:

https://odajournal.com/2019/01/19/maximum-precision-markov-transition-table-successive-daily-change-in-closing-price-of-a-utility-stock

And a link to an article using optimal weighted Markov to model two time-lagged outcomes with startling success:

https://odajournal.com/2019/02/18/optimal-markov-model-relating-two-time-lagged-outcomes

ODA also enables evaluation of the stationarity of a Markov process, as well as of the number of processes underlying the Markov process.

https://odajournal.com/2018/11/14/using-oda-to-confirm-a-first-order-markov-steady-state-process

https://odajournal.com/2018/11/14/using-oda-to-determine-if-a-markov-transition-process-is-second-order

Of course, (weighted) entries in the Markov transition table may themselves be modeled optimally, using propensity scores:

Article Computing Propensity Score Weights for CTA Models Involving ...

https://odajournal.com/2017/04/18/what-is-optimal-data-analysis

I hope (intend) to study these latter methods after I get a better "handle" on the greased-lightening trajectory of interesting temporal series.

https://www.quora.com/What-are-the-best-stock-price-forecasting-methods

There are loads of articles using various kinds of time series data prediction tools for stock price prediction. Machine learning nowadays has become an emerging tool for making forecast.

Check ANN and ANFIS. Both are good technique in using historical data to train and test forecasting model.