What should I use? Daily bars or 1440 minute bars?

I have stopped using daily bar data quite a while ago.  They have issues that crop up in both backtesting as well as real-time trading.

Daily bars do have an advantage of calculating very quickly in a strategy, but the disadvantages outweigh any such advantage.

When real-time trading, the close of a daily bar can change due to a final reconcilitation of the settlement price at the end of the day.  So at the close, the closing price may be represented by the final tick price.  However, it can then change once the final settlement price is determined by the exchange.  In trading, I am only concerned about the final tick price since this will not change.

When backtesting, I have often found a discrepency between daily bar OHLC prices and OHLC prices of minute bars.

For example, the open of the daily bar below is 1479.75 and the close is 1462.75.

However, when you examine the 1 minute bar values, the opening price is 1480.25:

If ran a strategy that had a market order to buy at the open of the daily bar on 1/25/2007, the backtest engine would fill it at 1479.75.  If you turned on LIBB with 1-minute resolution, then the backtest engine would fill it at 1480.25.  I have found the daily vs minute bars like this can have a large “trickle” effect on the results of a backtested strategy.

If you use 1440-minute bars instead, then the data discrepency disappears since TradeStation will build the 1440-minute (i.e., “daily”) bars from 1-minute bars.

Now the question could be asked, “Is the open of the 1-minute bar correct?”  Without the tick data, we will never know.  But I’m willing to bet that the 1-minute bars may be a bit more reliable.  Plus, using 1440-minute bars brings better consistency for backtesting and removes the closing settlement price issue during real-time trading.