Please visit Puckerings for more hockey stuff by me   

It’s past the half-way mark of the NHL schedule now, so it’s time to once again provide an update on my 2002/03 predictions, and compare them to how The Hockey News is doing so far this year. Through January 15, 54.3% of the NHL schedule had been completed. And as you’ll see, the results have gotten much more in line with predictions since the last update.

To evaluate the predictions, I will use the coefficients of correlation between the predicted conference rankings and the actual conference rankings (based on projected points to account for differences in games played), as at January 15. Really good predictions will have high positive correlation coefficients, say around 0.70 or higher. So without further ado, here we are:

The Western Conference has been incredibly hard on prognosticators, with highly-rated Colorado and San Jose both being mediocre teams for the first half. The East has been far easier to predict; just look at the results above. The correlations coefficients for the east are incredibly high, for all three predictors.

I’m still feeling good, beating both last year’s actual results and the self-proclaimed “Bible of Hockey”, The Hockey News’ predictions as well (though just barely). Perhaps my hunch about my competition’s overpredicting is right; or maybe, the situation will be entirely reversed come the end of the season. After all, the predictions were meant to cover 82 games, not 44.5.