Using Hollinger Stats, LoadedO Predicts 51-31 season
Using the Hollinger rankings as a reference point and reviewing the Blazers' record against the league to date, I created four tiers of teams in the league and assigned a potential winning percentage over the teams in those tiers. Basically it's the Big 3, the other teams fighting Portland in the West, a smallish group incl. Toronto and Detroit, and then the bottom 13 teams.
From that I calculated the aggregate potential for wins and came up with 51-31. See how I did it, at LoadedO...
10 months ago
torridjoe
3 comments
0 recs |
Comments
Interesting calcs...
Did you try back-testing your model (i.e. use your model to “predict” the first 35 games of the season and see what our theoretical win total is versus our actual win total)? It may be interesting to see where/if variances appear. I suspect, since your model is based on Hollinger’s stats and his stats are based on previous performance (which is clearly very strongly correlated with our current record) your model should line up pretty well…
by pantelakis on Jan 9, 2009 12:59 PM PST reply actions 0 recs
Interesting calcs...
Did you try back-testing your model (i.e. use your model to “predict” the first 35 games of the season and see what our theoretical win total is versus our actual win total)? It may be interesting to see where/if variances appear. I suspect, since your model is based on Hollinger’s stats and his stats are based on previous performance (which is clearly very strongly correlated with our current record) your model should line up pretty well…
by pantelakis on Jan 9, 2009 12:59 PM PST reply actions 0 recs
didn't think of that
but it’s an interesting exercise. Let’s see…I get 18.8 victories expected based on the games they’ve played so far. Nineteen vs 21 is pretty close, yah. Getting six wins out of MIA-DET-TOR, when the model expects basically 2, is where most of the gravy is. :)
by torridjoe on Jan 9, 2009 1:26 PM PST reply actions 0 recs



























