Professor Kyle MacLean invited me to present a guest lecture to his bizskool class on Sports and Entertainment Analytics at UWO's Ivey Bizskool today. The talk involved lots of war stories, but some things, too. What a great time!!!
Here are the notes for that presentation:
Notes for February 26, 2018:
Baseball, background
Bill James Historical Abstract and intuitive stat techniques
- Ability to win close games? Or is it luck…
James argued that it is luck and that if a team’s win% in close games is greater than its overall win%, that’s a sign the team was just lucky and will likely have a lower win% in the next season.
If W%c1 > W%t1, then we should expect W%t2 < W%t1, and
If W%c1 < W%t1, then the team was unluckly in t=1 and we should expect W%t2 > W%t1.
Not for all teams, and generally, roughly speaking, on average.
To test, must have a good defn of close games. One-run? Extra innings? Overtime games in hockey? 3pts in basketball or football?
Of course it doesn’t always work:
- Teams change from year to year
- Some teams have more depth for later in games.
- Test for significance? Normal approx. to binomial?
- Confirmed by James and numerous students.
- Coin toss. Need honesty. Find out who the champion coin-tosser is.
That person has a true ability to predict coin tosses. Ha ha. Relate to Finance.
What about players who come through in the clutch or those who choke? Are there players who demonstrate clutch ability?
Well, they may have a string of clutch performances:
- Tom Brady
- Joe Montana
- Pat Tabler’s batting average with the bases loaded.
- Dan Gladden
- [choke?] Scott Norwood’s missed field goal in last 9 sec of Superbowl.
Did these players have clutch ability? Choke ability? How do we know they weren’t just lucky or unlucky.
[rispl = runners in scoring position, late innings]
Studies showing if BA(rispl)1 > BA1, then the distribution of BA(rispl)2 – BA2 is normally distributed the next season. More generally the distribution of BA(rispl) – BA is not consistent year-to-year.
The point: don’t confuse luck and ability
- - - - - -
Why do all these tests? To measure ability and to test assertions about player and team ability. GMs want to know who to sign and how much to offer, and that requires measurements and forecasts.
What are the goals of professional sports franchise?
Profits and narcissism (aka utility?)
To max profits, produce that rate of output for which MR = MC.
What is the output of a sports franchise? What do they produce?
Wins. So want MR(W) = MC(W)
But a team generally makes most decisions ex ante, before the season, and so it tries to use inputs to produce expected wins up to the point at which the eMR(eW) = eMC(eW) [Here, “e” means “expected”; also, in reality should discount for present value]
What does eMR(eW) look like? Humped because extra wins as a team begins to contend for the playoffs, each additional win is worth quite a bit.
What does eMC(W) look like? Upward sloping.
Put the two together to show eW*. Note that if a team has tonnes of die-hard fans (Cubs, Leafs) then eMR(eW) is pretty low and it pays to not to have such a good team.
Cubs in 1965-6 had very low attendance. Averaged <8K/game.
https://www.baseball-reference.com/teams/CHC/attend.shtml
Back to graphs.
How to produce eW? By hiring playing ability (PA)
Hire ePA up to the point at which eMR(ePA) = eMC(ePA)
Problems:
- can’t always get smooth increments
- complementarities (less of a problem in baseball but still there)
- how to measure PA or predict ePA?
Too many measures are team and/or teammate dependent
e.g. Wins [often used for goalies or pitchers. Ugh. Teammate dependent and somewhat arbitrary]
hockey: +/- ? save%
Baseball: Triple crown – BA, HR, RBIs
Better baseball: OBP and SLG
Rogers contract [when I did play-by-play of the London Werewolves games, I insisted that the contract specify they would NOT show RBIs on screen but WOULD show OBP]
OPS = OBP+SLG
Pitchers? Not wins. ERA? WHIP? Why not OOPS?
What does WAR imply re: the supply curve of playing ability?
Economic Impact on a city from sports and arts.
- Multiplier [not 7, as some sports owners argue].
Eastbourne story for World Cup
- much spending leaves the community
- negative impact
- diversion from other community spending
- likely somewhere between +0.5 and -0.5
- but it puts the city on the map as big league.
If that’s valuable does it show up in real estate values?
Green Bay vs Oshkosh and Appleton? Nope. Explain.
Stratford and Niagara-on-the Lake? Yes. Explain.
- Olympics? [didn’t get to this]