Sounds crazy. By definition alone, "randomness" should not be able to be "exploited". "Regularities" might be "taken advantage of/exploited".
I would take careful measurements over a long period of time and apply RSI and MACD to it. When I get an RSI cross and MACD confirms it (or is it MACD cross with RSI confirm?), then I'd put the sail up.
I think the example is a false analogy. It is easy to exploit market randomness. This is what brokers and HFT do. Set up the game so that you extract a very small gain out of random motions. Then you always profit. Casinos have done that for ages although their edge is high, for American roulette it is 5.26%. European roulette is 2.70%. Actually, most money in the world are made because of randomness. In the presence of orderly processes, no money can be made because of efficiency. If you are a retail trader, the only way to profit out of randomness is by "not being fooled by randomness". Some fools think that randomness precludes patterns. Actually, randomness hides patterns so that they are hard to detect.
"So perhaps traders should look for markets where trends tend to be longer?" Winner, winner, chicken dinner. The next logical progression in the calculus is "of the markets that tend to trend longer, which ones demonstrate the least variance from the trendline during the lifetime of the trend".
"It is easy to exploit market randomness. This is what brokers and HFT do. Set up the game so that you extract a very small gain out of random motions. Then you always profit. " Not true. I know from direct first-hand experience that HFT's are abandoning the "shake" algorithm and other strategies that flip market noise.
From direct experience I am telling you noise is all over the place and HFT profits from it. Actually, the number of real traders is very small. This is understood by anyone who actually trades like me. Noise has gone up. The HFT people say so because they want to avoid upper limits placed on number of orders per unit of time.
Yes it is. I think many people overlook this cliche and disregard it because they do not know how to define trend. I could easily give another trader my definition. But they would fail with it because we do not have the same model, timeframe and psychology. If you want to exploit randonmess you must become a master of your time frame first. Then it ceases to be random. Its just an entry/exit point.
The poster assumes there are tradable (exploitable) trends in random data. There aren't. Try it in Excel. Throw up 1000 random time series and devise a trading system that systematically makes money from the random data. Post your results.