Thanks for keeping up with the Advanced Sports Analytics Tout Rankings. Today we will be looking at both tout performance for Week 3 and aggregate performance for the season. Just as a refresher, we cover 6 sites using the Draftkings scoring system. They are numberfire.com, swishanalytics.com, fantasylabs.com, fantasyinsiders.com, fantasyalarm.com, and 4for4.com. Here are the results for Week 3:

 

4for4 fantasyalarm fantasyinsiders fantasylabs numberfire swishanalytics
MSE 64.579 76.743 62.256 71.399 70.578 63.837
MAE 6.121 6.623 5.263 6.419 5.740 5.677
ME -0.184 -0.647 -1.619 -0.439 -2.076 -0.824
RS 0.288 0.258 0.798 0.391 0.148 0.221

 

This was a very tough week to project. Some of the top players, including most notably Julio Jones, seriously underperformed. There seemed to be an overall much higher variance in player performance relative to expectation. This is reflected in the Tout Rankings this week. The mean squared error scores were fairly high, with no site getting under 60. Another interesting thing to note is the inflated scoring this week. While in the past two weeks we have noticed sites overestimated total fantasy points, this week all six sites had negative mean error. Finally, note that fantasyinsiders had a ridiculously high RS this week. I would guess this is due to the fact that they project substantially fewer players than other sites.This means that they are only projecting high scoring players, which is conducive to a high RS while still have a relatively poor ME. Overall, it seems like 4for4 and fantasylabs had the best projections while keeping up a high depth of analysis.

The cumulative results are as follows:

 

4for4 fantasyalarm fantasyinsiders fantasylabs numberfire swishanalytics
MSE 52.864 58.930 40.855 62.631 49.321 49.210
MAE 5.315 5.739 4.333 5.930 4.731 4.874
ME 0.128 0.009 -0.390 0.702 -1.162 -0.290
RS 0.531 0.555 0.637 0.605 0.490 0.482

 

This table shows something very interesting. It seems that fantasyalarm is incredibly well calibrated on ME through three weeks, with nearly 0 error. They are not particularly impressive on any other metric, but it seems that they put a lot of effort into understanding the total fantasy output of the week. Also, numberfire’s rough week this week has crushed their ME score across three weeks.

Visit the site next week to see how the Touts performed in Week 4.