Friday, October 12, 2018

Recovery, Part 3

So after seeing a couple of good weeks, I couldn't help but be curious.  I pulled down Todd's data archive for the season to date from Prediction Tracker.  After filtering the first week out, darn if I wasn't the most accurate (in terms of Mean Square Error) of anyone, including the line (which was second).  Doing this well early in the season is really hard.  I hate that I've messed up my numbers for the year.

If you're curious, I had a formula in a spreadsheet that pulled over the returning quarterbacks data.  Unfortunately, I forgot to convert the formulas to values before I sorted the teams.  The teams "sorted," but the returning QB numbers never moved.

Such is life.

Sunday, October 7, 2018

Recovery, Part 2

I did another spot check on last week (week #6).  Again, most accurate in terms of mean square error, again by a pretty good margin, beating even the updated line* (~268 to ~276).  Now I'm beginning to wonder how the season numbers would look if I hadn't boogered up week #1.  Oh, well.

*Disclaimer:  I very highly recommend the stock market over sports betting.