This past month or so has been all about context. Let me explain. Gokul paid our sleepy campus a visit and a few sat up to listen. Here is a well spoken, tall guy talking about cricket, holding forth on how a run is not a run or a wicket is not a wicket. In other words, while we reel off the stats at will, he was making the point that runs scored in a successful 4th innings chase of over 400 runs in faraway West Indies were worth a lot more than those scored on a flat deck in Rawalpindi on days one and two. Runs and wickets were simply a means to the end: winning the cricket match.
I was happy because his work with Jaideep Verma and the rest of the crew at Impact Index was just the material I was hoping to find to bolster my sports analytics elective, SAIS. Students had been somewhat bemused as to why so many sections of the course dealt with the American big-four sports and not cricket or for that matter even soccer. I had maintained that when Bill James arrives in cricket, I would be happy to include cricket material. Glad that starting from the academic year 2015-'16, should the offering be subscribed to, there would be cricket material.
When the Research Papers track of the MIT Sloan Sports Analytics Conference started at the Boston Convention and Exhibition Center (BCEC) this Feb 27th, at least 4 of the 8 finalist papers dealt with...well...context. Patrick Lucey presented the methodology to use spatio-temporal data to better predict soccer outcomes. Shots on goal in soccer devoid of context can be confusing. Shot quality mattered. Each shot attempted needs to be weighed by its quality, that is the chances of the shot being a goal. Of course, shots taken from closer to goal/ from better angles would receive better weights because their chances of scoring a goal are higher.
Joe Rosales and Scott Spratt discussed the need to parse out the individual contributions to a called strike in baseball. In so doing their work better evaluates the catcher's ability to frame a pitch and thereby earn additional called strikes for his team. While the umpire, catcher and pitcher all play a role in a called strike, if we can parse out the impact of the catcher, then that directly affects valuation of catchers in the free-agent market. In their presentation they mentioned the "context of a called strike" and this rang familiar to how wicket keepers in cricket may earn additional wickets through their footwork to spin bowlers...getting perhaps close LBW shouts.
Stephen Pettigrew presented his work on assessing the offense productivity of NHL players using in-game win probabilities. This made me jump in my seat because it appeared so intuitive. He spoke of how the sixth goal scored in a 6-2 demolition may not be as valuable as a winner scored with time winding down in a 3-4 humdinger. Going from here productivity of the offense in hockey (ice-hockey) is better measured by incorporating the context in which the goal was scored.
Alexander Franks and Andy Miller introduced counter-points - a measurement of players' defensive productivity in the NBA. In "enriching the measurement of defensive play" their work directly dealt with context which is key to measurement of defensive productivity, defense being an inherently group endeavor.
The above papers used spatio-temporal, statistical, player tracking and visualization methods to get their hands around the context...to ensure that metrics reflected difficulty or sophistication levels well. It looked like here were II's intellectual brothers. It is interesting that different sports were at different maturity levels when it came to the extent to which current metrics reflected context - and thereby true importance of metrics in the manner they relate to the ultimate objective, wins. If I were to hazard a guess, in decreasing order of impact-context measurement sophistication (ICMS) would be Baseball >= Basketball > American Football > Ice Hockey, Soccer > Cricket. To the student queries in my SAIS elective, this should have been my answer. Please review this list:
Baseball >= Basketball > American Football > Ice Hockey, Soccer > Cricket
Top and bottom are bat and ball games their countries adore. Flowing games such as ice-hockey and soccer have been recent starters on impact-context road, with spatio-temporal data allowing researchers to model them. Most importantly, cricket has a long way to go.
Baseball >= Basketball > American Football > Ice Hockey, Soccer > Cricket
Top and bottom are bat and ball games their countries adore. Flowing games such as ice-hockey and soccer have been recent starters on impact-context road, with spatio-temporal data allowing researchers to model them. Most importantly, cricket has a long way to go.
We had a poster too at this conference set in the world of golf. Come to think of it, "accentuated loss aversion" is perhaps simply adding richness to the psychological context in which PGA golfers attempted putts. What we stopped short of doing was understanding how layering this psychological context might change rankings...in terms of golfers' putting productivity...and perhaps in other sports too!
Of course, the trip to Boston had other highlights too: Manu and Sangeetha drove down and dropped me on the way back at Foxwoods, got introduced to IPA, discovered Marathon Sports, stumbled upon Charlie, and realized that some of the old way-finding faculties still worked!
All the Best, Impact Index!!!
All the Best, Impact Index!!!
