What is the accuracy of historical data required for correct testing?

Just came across this thread, and it’s interesting how much detail goes into backtesting. I always thought historical data was just about past price movements, but it seems like execution details really matter too.
I wonder how professional traders deal with this. Do big firms use the same historical data sources as retail traders, or do they have something more advanced? If data quality is such a big factor, it feels like having access to the right data is almost as important as the strategy itself.

Yeah, this is a common issue. The accuracy of your historical data can make or break your backtests, especially if you’re dealing with scalping or high-frequency strategies. I remember back when I was testing a news-trading strategy—looked like a goldmine on 1-minute data, but in reality, spreads widened so much that most trades would’ve been losers.
From my experience, the best approach depends on your trading style. If you’re a long-term trader, even 1-hour candles might be enough. But if you’re doing anything intraday, you need at least 1-minute data, and for scalping, tick data is a must. I found that using https://forextester.com/blog/best-demo-trading-simulators strategies helped me get more accurate results because it accounts for real market conditions like spread changes and execution delays.
Are you mostly backtesting short-term or long-term strategies? That could help figure out what data quality you actually need.

I’ve been playing around with backtesting for a while now, and one thing keeps bothering me—how accurate does historical data really need to be? I used to think that as long as you had a general price trend, it should be fine. But after testing the same strategy on different datasets, I started noticing big differences.
One time, I ran a backtest on a dataset with 1-minute candles, and everything looked solid. Then I tried the same strategy on tick data, and suddenly, my win rate dropped because the fills were completely different. It made me wonder—how much detail do we actually need for backtesting to be reliable? Do you guys always go for tick data, or is lower resolution good enough for most strategies?