These predictions (10000 simulations per control scenario) are based on the 2000 MCMC iterations from commit 336a146c. Corresponding plots are in dropbox, here, https://www.dropbox.com/home/challenge-ASF/sims_model9-MCMC4-10000sim
ps.1 - MCMC4 is currently at 4000, so the csv filess here are a little bit out of date (but probably doesn't change much).
ps.2 - model10/MCMC is currently at about 700 iterations, but there are 5 of them which we can combine for better MCMC integration. So long as I didn't screw up the new parameterisation of active search we should hopefully switch to using that Wednesday morning.
ps.3 - Good luck! Don't hesitate if you need more info from me.