In a search to implement a model from Robert Close (1) Chantal Roth found a model that reaches about 73% detection while having a CHSH of 2.8. From the literature it was clear that these model existed (2), but so far I had not been able to find code for this.

The hidden variable is a common angle theta, and the measurement formula (to calculate the spin for a particle) is:

public int measure(double filter_angle, Particle particle) {

double theta = particle.getTheta();

int spin = (int) Math.signum(Math.sin(theta+ filter_angle));

double pdetect = 2.3*Math.abs(Math.sin(theta + filter_angle));

if (Math.random()<=pdetect) return spin;

else return Integer.MIN_VALUE;

}

double theta = particle.getTheta();

int spin = (int) Math.signum(Math.sin(theta+ filter_angle));

double pdetect = 2.3*Math.abs(Math.sin(theta + filter_angle));

if (Math.random()<=pdetect) return spin;

else return Integer.MIN_VALUE;

}

When Integer.MIN_VALUE is returned the particle pair is not counted.

- Robert Close, http://www.classicalmatter.org/Bell/SpinCorrelation.pdf and http://www.classicalmatter.com/Bell/SpinCorrSim.pdf.
**Att.: the model discussed above is different from that of Robert Close.** - Part of LH models non-issue for scientific community, http://challengingbell.blogspot.com/2011/04/part-of-lh-models-non-issue-for.html