# Manufacturing Machine Control, from Kim Zussman

July 25, 2007 |

Thinking about use of manufacturing control systems to time market cycle-changes, take the example of a Malibu Oxycontin 120 tablet machine.

Say the FDA specifies that the tablets must have 120 mg of oxycodone with a standard deviation of 1mg. From this you can determine the frequency of pills which are too high, and adjust the machine (and keep checking it) so that it "never" makes 130 mg tablets (never > 10 st dev, etc).

But is the market like a pill machine that gets clogged with chemicals or wears out, and can be adjusted back to good behavior?

Here, for example is check on frequency of SPY c-c drops <-1%. The control system is regression of wait times (in trading days) between 1% drops vs. arbitrary timescale:

Regression Analysis: wait versus td

The regression equation is wait = 6.59 + 0.115 time.

Predictor     Coef     SE Coef     T      P
Constant     6.588    2.917  2.26  0.026
time            0.115    0.055  2.09  0.039

S = 14.1789   R-Sq = 4.6%   R-Sq(adj) = 3.5%

Looks like the wait gaps between such days is increasing, so is it time to adjust the machine?

More on engineering control:

"His face is melted to the wire!"