Ever wonder how I generate the ESPN Fantasy basketball projections each season? If you read my work regularly, it should come as no surprise that I lean heavily on advanced analytics to drive the process.
Here is a look at my general approach and, specifically, how I use Second Spectrum to take my projections to another level. I'll break down my process for Pascal Siakam's projections for 2019-20 and projections for Joel Embiid, Paul George and Ja Morant this season.
Last year at the MIT Sloan Analytics Conference I spoke about using different analytics to project player production. One of the examples I gave was Pascal Siakam, who at the time was in the midst of his first season as the Toronto Raptors' primary option after Kawhi Leonard had moved on to the Clippers.
During the 2018-19 season, Leonard had led the Raptors in scoring at 26.6 PPG on 18.8 FGA, shooting 49.6% from the field, while Siakam had been second on the team with 16.9 PPG on 11.8 FGA at 54.9 FG%. Entering the season, there was an expectation that Siakam might be called upon to be the lead option for the Raptors, which thus made him a popular sleeper in fantasy basketball drafts. But on a more granular level, how could we predict what exactly Siakam would look like in the main role?
Well, one of the better tools to predict future performance is