The Parabolic SAR becomes pretty meaningless at the points of a chart that are preceeded by an extended trendless or choppy period. During this period, minute gyrations in price can randomly push the indicator to either side of the price, and when the next big trend begins there is no repeatable guarantee whether it will be on top or on the bottom. It is also succeptible to the initial conditions of the price series. Look at a 7:30 - 4:00 intraday chart of the ES with the PSAR indicator and compare it with the same chart with just the a time template changed to 9:30 - 4:00. The differences in the morning hours could be vast. It only becomes meaningful after a fairly strong trend is underway, after which it will always be lagging the trend. If you tried to use it during a trendless period, your results could be pretty much random. I don't use it, BTW.
I see what you mean, but thats the point-you dont know what the price will do, you just have to find an entry point, a dot on the chart somewhere that "looks good" for an order, and the sar does that. Ive been testing my stuff for a long time, with spectacular (theoretical) results. But given a pure discretionary method, its always going to be argued the results could be "random". I like them, 'coz they make the chart look purdy. You COULD always use them as a stop loss..........