For any kind of detection or diagnostic system, there is always a balance between sensitivity and specificity. In a perfect world, you would like to see 100% sensitivity and specificity. That is, whenever you make a call, you are always correct, and you do not have any missed calls. In reality, that does not happen.
If one intends to increase sensitivity, he may end up with decreased specificity, that is, he will pick up lots of false signals. Whereas if he wanted to increase specificity, he may end up with decreased sensitivity, that is, missed signals. The only thing one can try is, based on existing data, draw the line somewhere for him to get the best combined results.
Real trading examples can be seen in the chart I posted above. Point b was a buy signal given in last December, after the red line spiked to 21.4. Point e was May 15, close to the bottom of the 20-something-point drop that day, when the red indicator rose to 19.5. Retrospectively, the red indicator just cleared the threshold of 20 in last December to trigger a buy signal, whereas it missed a buy opportunity last week by a hair.
But 20 is not a magic number, no more special than 19 or 21. My point is, one has to be prepared for the possibility that his system may give a false signal or miss a signal, even though when that happens, it may still not be the end of world. For example, if the buy signal in last December had been missed, and one had been shorting the market into January, market eventually did drop to point d, which was lower than the initial sell point a, and much lower than the second sell point c. The roller coaster ride would have been stomach churning, but the guy could still have come out with a nice profit. This may be attributed to the possibility that a weak signal (therefore missed) indicates a weak market movement in that direction. However, I cannot prove that with confidence, and it also could have been pure luck, therefore there is no guarantee that one can be as lucky again next time he misses a market turning point.
Hence the reason I keep reminding myself how important risk management is. I don't think there is a single correct answer to how one manages his risk, be it position sizing, loss-cutting, hedging or some other strategies, whatever works for him.