When I first began to study technical analysis and how to apply it in the equities markets, I listened to an audio program that focused on day trading. The program was a recording of a live seminar on the topic. Quite early on, an audience member asked if technical indicators work as well on an hourly or 5 minute chart as they do on a daily or weekly chart. The speaker did not provide any evidence, but rather simply stated that for the purpose of the seminar, we are presuming that technical analysis works just as well on intraday charts as it does on longer-term charts.
Short Term Trading and Randomness
My experience in the markets tells me that this is certainly not true. Short term bars can be dominated by single large trades, and the fear, greed, and mass hysteria of fast markets that surround news events. Simply looking at a 5 minute bar chart, we will see one hell of a lot more noise…not to mention the fact that in order to trade this chart successfully, the trader must continuously execute split second decisions with consistency and clarity. Even if possible, this method must be quite stressful, and lead to a shortened career as a trader due to burnout or blowup!
Long Term Trading
The first thing that we can do to increase the signal to noise ratio on our charts is to go to a longer time frame. Of course, this in itself is not a cure-all, and even in fact, the distinction between short term and long term is subjective and seems to be different depending on which trading instrument we are referring to.Â
As an example, Cable Glider operates on a 30-minute bar chart. This is undeniably a short-term chart.  I have no explanation for why 30 minute bars work well in this case. For any other instrument that I have tried to build a trading system for, it always required using a 1-hour or greater bar time in order to generate a system with positive expectancy. I’ve sometimes heard this disparity referred to as the “personality†of a trading instrument, though I suspect it has a lot to do with the depth of liquidity in that market and the number of traders actively participating in the market.
Closed bar trading
Closed bar trading means only making decisions about entering a trade based on the previous bar, and if all conditions for entry are met, open a new position at the open of the next bar. This is a simple concept that improves the reliability of system backtests, and also mitigates the effects of bad data (presuming that “bad†data is more likely to printed as erroneous highs and lows and not erroneous opens and closes)
Tick Chart = Random path
When I first began to code automated trading systems, I thought that I would need to run my tests on tick charts, that is, charts that record EVERY trade and price movement in the market. This makes intuitive sense, but I now believe that that tick charts are useless at best and harmful at worst. If we can see that 1 minute bars are entirely too noisy to make any trading decisions on, what can we gain by looking at every tick!Â
It is true that a tick by tick data should allow us to reliably backtest a system that is making decisions based on the current bar, but this leaves us open to a lot more over-optimization and curve fitting. At worst, we might presume that the way that a trading instrument moved on a tick chart in the past will repeat itself. This is certainly pure folly, particularly in retail forex trading with a market maker, as the market maker can move the bid and ask wherever they chose, even without any trades occurring.
The bottom line is to chose a timeframe or set of timeframes that allow us to react quickly when timing entries and exits, but that also are long enough to filter out enough of the random noise so that our technical trading signals become reliable enough to generate consistent profitability. As usual, there is no quick solution or magic formula…. this is the ART of the technician! 
Â