With a rookie head coach at the helm and a rookie quarterback probably under center, Jets fans this season. But they can take a deep breath and relax. Preseason N.F.L. predictions are completely worthless. Using the consensus overunder win totals set by the Las Vegas “futures market” over the last four seasons,Cheap Jerricho Cotchery Jersey, we can judge just how accurate expert predictions are. First, let’s grade accuracy in a straightforward way using something called Mean Absolute Error (MAE). This measure is just the average difference between the predicted and actual win totals for each team. But real statisticians don’t like the simplicity of MAE. They like to square everything, which penalizes faroff guesses much more than close estimates. A more accepted measure of statistical error is something called Root Mean Square Error (RMSE). In this case, the RMSE for the consensus predictions is 3.1; for the braindead 8win predictions, it’s 3.2, just slightly higher. So over the next few weeks, when you read and hear the innumerable “expert” preseason predictions for season win totals, just chuckle to yourself. Let the season play out and enjoy every twist and turn. Anything can happen in the N.F.L., which is why we love it. Over the past four seasons, the expert consensus predictions had a MAE of 2.6 games. That doesn’t sound so good, but it’s not really meaningful until we compare it with some benchmark. What if I guessed 8 wins for all 32 teams each year? My MAE would be … 2.6 wins. But can hundreds of manhours of game charting, data collection and statistical analysis improve on those accuracies? Not really. One Web site with an annual publication (which I’ll pick on only because it is widely respected) touts a complex and advanced system for predicting season wins. It earns a 2.4 MAE and a 3.0 RMSE, only slightly better than a coma patient’s 8win predictions. (Editor’s note: of ,Cheap Ricky Williams Jersey, who will be a guest blogger for a week., a combination of football, math and common sense, is well worth a visit.) In 2005, the expert consensus was that the Jets would win 9 or 10 games. They won 4. In 2006, the prediction was 5 or 6 games. They won 10. In 2007, the experts said 8 wins and the Jets won half that. In 2008, the Jets accommodated the predictions slightly better by winning 9 games compared to the prediction of 7. I’m a big fan of advanced stats. But “advanced” doesn’t always mean complicated. Sometimes advanced is straightforward and elegant. A simple rule of thumb using a regression, which is basically using a ruler to fit a line to points on a graph, is just as accurate as any expert or system out there. It’s based solely on last year’s wins. Take last year’s wins divided by 4,Roy Williams cowboys Jersey, and add 6. A 12win team from last year would be expected to win about 9 games this year. A 4win team last year would be expected to win about 7. The MAE is 2.4 and the RMSE is 3.0, no worse than the vastly more complicated systems. I’m not claiming my rule of thumb is terribly accurate, just that other predictions aren’t any better. The author introduces： VA Cheap **** Butkus Jersey on sale 057 365d JERSEYS is a privately owned sports goods company with a history of nearly 4 years.We have the professional factory behind us manufacturing all sports jerseys.china nfl jerseys Read More