There is no simple solution, but I can give you some pointers. You should take the following into account when choosing the time frame for backtesting your online day trading system:
How many trades does your day trading system produce per day? It doesn't matter how long you backtest a day trading system—what matters is that you have enough transactions to base your assumptions on statistically sound data* A year of testing offers you enough data to establish trustworthy assumptions* if your online day trading system generates three trades per day or 600 trades per year.
However, if your trading system only generates three trades per month or 36 trades per year, you should backtest for a couple of years to obtain reliable data.
You must consider the underlying contract's characteristics. The chart below depicts the e-mini S&P 500's average daily volume:
Back-testing a trading system for the e-mini S&P before 1999 makes no sense because the contract simply did not exist! Backtesting an e-mini trading system before 2002, makes no sense because the market was completely different; less liquidity and different market participants.