When it comes to backtesting, the first procedure is to seek and get a reliable and trusted data source that can provide data that are almost perfect and most importantly zero-tolerance to manipulation and error. Do a background check to find those kind of platforms, ask other industry experts for recommendation, go online and do more research to work with few ones with great reviews.

Best practise is to choose right and precise data in time frame which suits your trading model. After that adjust that data with your strategy and time frame. Try it over multiple time frames and variety of statistics so you can find which strategy worked better in precise form. Last and most important point which people usually ignore while back testing of historical data is Slippages, slippages are something which is very important so never forget to add small slippage in your data while getting back testing results because in reality slippages will be there and in back testing it won't be.

In my experience, selecting the right data source is pivotal for effective backtesting. Aligning with your trading style and ensuring accuracy, completeness, and up-to-date information is crucial. For stocks, reputable exchanges with dividends, splits, and adjustments are essential, while forex trading requires data reflecting bid-ask spread and market liquidity. Considering cost, availability, and format, along with easy access and updates, establishes a robust foundation. This meticulous approach has proven instrumental in refining trading strategies through historical analysis.

Inaccurate data will give birth to inaccurate outcomes because of the accuracy gaps in the collation and optimization of data. To avoid this, double check the data, identify where error is and correct before using. Consider your strategy and approach and ensure the data representation aligns with them. As much as there are strategies a technical analyst can not adopt if he chooses to use line chart over candlesticks pattern.

The data must align with strategy for a technical analyst to backtest effectively. If my strategy warrants me to see the high or low of the day candle wick before I take a trade then I won't be considering line chart to align with that strategy because it doesn't provide such information that I need.

It is safe not to rely on a particular data source to be able to crosscheck and confirm the data in use is free from manipulation and errors. The use of different method is not abusive nor forbidden in this context, from walk-forward backtesting to Monte Carlo simulation, it is all good so long they fit in to supply value in direction of what the TA desires.