The polar file used with your routing software should reflect the boat's actual average performance, including the fact that in light air offshore there is often a sloppy sea state, and in heavy air the boat is often sea-state limited and cannot be pushed above a seamanlike speed upwind in the typical sea-state associated with heavy air. Further, the boat's average performance includes the full mix of helming skills, not just the performance of the best helmsman. So the polars you use for routing will end up being a separate file from your target polars. The routing polars need to be achievable. If you run a route on a GRIB file, and it turns out the first four days of the GRIB file forecast are accurate, in four days the boat actually needs to be exactly where the router calculated it would be. If your polars consistently overestimate your boat's performance, and you don't actually get to where the router calculates you will be at various times in the future, then the calculated route will not be relevant for your boat, and the routing software will be "playing" shifts you will never experience. There are, however, many pitfalls in computing optimum routes from polars. Lets look at them individually.