November 22 1900 UTC TWS - 10 to 15 knots TWD - 090 true Sea State - 1 to 2 meter swell Sky - partly cloudy, littered with cumulus clouds some of which make mild threats of bad intent This being American Thanksgiving, I am exceptionally thankful that I have enough excess fresh water that I was able to bathe today. So is every other biological creature downwind of Dragon. The things you take for granted. And for those in the office making book on me appearing with a beard - you should have taken the under. I do want to thank everyone who sent emails of good cheer today. They were much appreciated and I envy all of you your feasts and families and definitely the pumpkin pie. Yesterday afternoon's gybe to the southwest was earlier than the routing suggested, taking a bit of a gamble that in trying to cut the corner I could save some time. When I made the gybe, I was not going to be able to get to the island without at least one more set of gybes back out to the west and then south again. Every gybe is slow, and every gybe is a chance to screw things up. The gamble was that the breeze continue to swing towards the east, then the SE and thereby allow me to point the bow right at the north west corner of Guadelouoe, Tete Anglais. And it's not quite a gamble, but a bet on one set of forecasts (the USA' GFS model) and against another (Europe's ECMWF model). Each is built using massive super computing power and very similar methods. But with different assumptions , they can yield different results. The Euro model has frankly been more often correct in the past few years. I used to have to deduce that through my own observation and discussions with others, but for this race, I started using a new piece of software called Model Accuracy. This software takes the data from my wind instruments (the actual outcome) and statistically compares it to the GRIB files that represent the forecast from each model. And then Model Accuracy tells me which one is more accurate, and what the delta is between the actual and the models. I can use it to know which forecast I should put more confidence in, and use it to adjust the GRIB files to help improve the accuracy of my routing. The first half of the race, the two models had lots of differences, which led me to make the big gamble of going west, which I wrote about way back on Novemeber 11. The Euro model was more often correct, and predicted the lane that would develop through the high and allowed me to escape into the trades. But since then, while the models have been fairly closely aligned, the GFS has been more accurate. And the GFS tells me that it was time to get south west and quick. It's worked out well so far. Pressure has been solid and I have seen to the shift that brought me on course. And its given me an edge that allowed me to re-take 12th place. Time and a few more hundred miles will spell the real truth.