So far, I am really impressed with this indicator, which I mentioned in my previous post. It uses estimated market cycle periods to derive a signal to noise ratio. I made a histogram out of it, which crosses the 0 line at +6dB, meaning the signal is twice as strong as the noise. Anything less, and it’s much more risky to trade.
I was looking at the action on BBY around 11:00 as a possible double bottom… but since SNR was dropping I stayed out, and I’m glad I did.
It lags the market by about 4 bars, but it moves pretty smoothly, so you can kind-of think a couple moves ahead. Like, if it’s dropping quickly, you might take into account that it might be red soon… and if it’s red soon that means it’s too noisy right now. Make sense?
I’ll be watching it for a few more days, to see if the performance holds up. When the noise level is high, you see a lot of listless chop, like at the tail end of the chart above. I’ve also found that, if you simply _must_ trade the stock, you can move up to a bigger timeframe and see if that’s less noisy. That will usually mean larger stops, which is exactly what a noisy market calls for.
September 24th, 2007 at 7:08 pm
Hey,
I groc the Control system’s understanding of signal to noise, what is the signal to noise ratio indicator used here?
-Karl
September 24th, 2007 at 8:28 pm
The signal amplitude is estimated by taking a hilbert transform of the sampled price action. It uses the median price of each bar as the sample. The noise level is derived from the high-low range of each bar. And, of course, signal-to-noise ratio is the ration of those. The idea comes from the Rocket Science book.
September 25th, 2007 at 4:57 pm
[...] am also using the signal to noise indicator that I’ve been writing about. I think it works pretty darn well. You might say, [...]