June 20th, 2016, 10:51 am
Hi Alan it's a good question - yes, the specific placement of the local peaks depends on the specifics of the implied vol we use -- a parametric fit, calibrated to the markets -- but what it does show is that there is a considerable amount of (Brexit related) tail probability. I would be reluctant to commit to any point estimates but you can certainly see probability mass clustered in various regions/intervals. I'm by no means surprised to see bimodality (EURCHF early last year had much the same property and look how that turned out) but three peaks does look odd. Whether it's a an artifact of particular smile interpolation or an artifact of an atypical vol surface marked by nervous traders is a good question and one I would like to investigate in a further revision of our work, as more data rolls in.Of course you can't expect every single fx smile model to calibrate nicely to such stressed markets...