Mailbag: Sports Randomness Rankings, Important Holder Question

Mailbag: Sports Randomness Rankings, Important Holder Question

Submitted by Brian on February 4th, 2014 at 12:41 PM


anecdotal comparison demonstrates that the best basketball player ever, Michael Jordan, was considerably better than the best baseball player ever, a chimpanzee named Carl who led the 1883-1884 Erie Mudjacks with a .546 batting average.

Question about randomness in sports that evolved into a very long answer with rants about how to fix hockey.

Hey MGoBlog team,

I'm of the opinion that individual games in any sport are decided by a combination of four things: player athleticism, player skill, player/coaching strategy, and chance. If you want you could also pull refereeing out of the chance bucket to make a fifth. Do you think this is a valid and relatively complete model and if not how would you change it? A related question is which of the five major sports (football, baseball, basketball, soccer, and hockey) relies most on chance to decide the outcome of a single game? How does that affect the fan experience?

My answer is basketball and hockey as the most reliant on chance, which I think negatively impacts my ability to get too caught up in the outcome of those games. That could be because I'm relatively ignorant of the strategy and skill elements in those games though, which is why I seek your most esteemed opinions.


Those four things cover just about everything, though athleticism and skill are sometimes difficult to parse out from each other.

As far as randomnesss goes, basketball in fact seems like the sport least impacted by chance, especially at the NBA level. The best NBA teams win something like 80% of their games, and teams that good generally plow through four rounds of playoffs without issue. A whopping 14 of the 20* teams with the best regular season win percentage in NBA history went on to win the championship. Two that didn't (the 96-97 Utah Jazz and 1995-96 Seattle Supersonics) lost in the finals to teams ahead of them on the list.

Meanwhile, no baseball team in a much longer history has touched an 80% win rate and only five of the 14 teams that Wikipedia references made it through much shorter playoff structures. Also, eight of them played before WWI. Baseball is the king of randomness. They've done studies and everything.

And when you think about it, it makes sense. Basketball is structured as 60-90 random trials worth 0-3 points for each team. That's a lot of trials. Baseball has nine. Football has 10-12, though that's a bit different because those trials are a lot less independent of previous ones. Hockey defies this sort of categorization. But just think about the MAXIMUM THUNDERBOLT LEVEL possible in any particular sport as a proxy for randomness:

  • A man makes a three pointer and is fouled for four points out of 100.
  • A goal is scored!
  • A grand slam!!

Despite the high thunderbolt rating of the top football plays, the sport as a whole is less random than hockey and baseball, primarily because each play has an impact on the one that follows it in a way that doesn't happen in sports with less memory. You can load the bases in baseball and come away with nothing; ripping off a 30-yard run in football is always worth something.

Soccer also defies this analysis, as goals rank high on the thunderbolt scale but soccer Cinderellas are exceedingly rare. When Cypriot club APOEL made the final 16 of the Champions League a couple years back there was a veritable tizzy; they were immediately bashed out by Real Madrid 8-2.

And that makes sense, too. The structure of the game is such that good teams always have the ball and bad teams give up on the idea of competing on equal footing, instead preferring to pack everyone in front of their goal in hopes of a tie or a fluke goal. Outmatched teams regularly get outshot 10-1, compared to 2-1 in hockey. Hell, when Barca comes to town even teams backed by Russian plutocrats have a tendency to park the bus and hope for fortune.

Your randomness rankings, then, from least to most random:

  1. Basketball
  2. Soccer
  3. Football
  4. Hockey
  5. Baseball

It's a wonder they bother to play baseball instead of just guessing what color the ump is thinking about.


College basketball is quite a bit more random than the NBA, primarily because the game is shorter, possessions longer, and three pointers more readily available. Michigan just lost a 55 possession game in which the opponent went 8 of 13 from three primarily because one guy was unconscious. If that was an NBA game it would be the middle of the third quarter and they would have a chance to right the ship.

But overall, college basketball still spits out the "right" team enough that it's satisfying to me. The playoff structure very rarely sees top teams not reach the Sweet 16 and even more rarely puts a true interloper into the Final Four. Meanwhile, the single elimination nature of it makes any team vulnerable. March Madness straddles the line between unsatisfyingly random (baseball) and boringly rote (the NBA) almost perfectly.

I am with you on hockey, though. I find myself increasingly discontent with 2-1 games in which goals are mostly a matter of which plinko ball makes it all the way through the mass of bodies. Watching MSU and Michigan play in the GLI was illuminating, as Mickey Redmond took some time to compare MSU's defensive strategy of packing all six players in or near the crease to the current NHL vogue, then complain about how the game was better when people were checking guys on the points. The game has shifted such that save percentages added in with the number of shots that don't get to the goalie  means that maybe 1 of 50 point shots does anything useful.

Meanwhile, the single elimination format that works so well in less-random basketball is a complete disaster in college hockey. Hell, the addition of the shootout and the loser point in the NHL has made even the 82 game regular season(!) way less predictive than you'd think. This then bleeds into the playoffs, where 8 seeds beating 1 seeds is commonplace. While it's somewhat controversial, these days statisticians struggle to find meaningful differences between most NHL goalies.

As a result I'm in favor of rather radical changes that would help teams who dominate in shots and attack time win more games, starting with expanding the net by approximately as much as goalies have improved in the last 20 years. Nine of the top ten save percentage marks in NHL history have been posted in the last five years. The only guy in the top 35 to post his mark before 2000 was Dominik Hasek, who owns five seasons in that range. There isn't a season older than 1993 in the entire top 250. This kind of goalie dominance makes hockey unsatisfyingly random.

I prefer a world were everything that goes off the post now is a goal. This gets goalie save percentages back down to a place were .900 is pretty good, differentiates the goalie pool, and makes standing around waiting for a hail of rubber a worse idea. Then gradually introduce Olympic ice in new buildings, do something to fix the wreck that is the offsides rule**, and maybe futz with goalie pads so they're more like they were back in the day—heavier, harder to cover your five-hole with, more cumbersome.

Death to plinko hockey.

*[I'm setting aside the 1946-47 Washington Capitols from Wikipedia's list since that season was considerably shorter than a regular NBA season, and it wasn't even the NBA yet.]

**[Current position: allow play to continue as long as the offsides player immediately moves to tag up at the blue line. Any other action by the player brings a whistle. There's no reason for rushes to get broken up because a guy is one foot offsides.]


With Dileo departing, an overlooked change that will take place next year is at holder.  Who do you think will assume this role?  My preference would be Norfleet or Peppers since I could see them executing some razzle dazzle that would lead to me dumping marinara sauce on my head in joy.



I have bad news, Mike: it's almost certainly going to be Kenny Allen, the backup punter, with an outside shot of one of the backup quarterbacks.

Dileo was uniquely qualified to be the holder because he secretes a sticky substance from his fingers that gives him super catching powers. Peppers and Norfleet are unlikely to be in Dileo's class in that department, so Michigan will replace him with someone who takes a ton of snaps as part of his daily routine. This is the way of the world.

Yeah, that change will limit Michigan's ability to fake field goals. Michigan will probably just go for it instead, which is fine.


Was wondering if you could project who you think the starting CB will be next season. I may be in a minority but I was really impressed with Jourdan Lewis.

Peter F

It's going to be tough for Lewis to fight his way past either incumbent starter, especially with Jabrill Peppers arriving on a cascade of hype, torrent of praise, and all other water-related-muchness of thing.

First, the incumbent starters. I know both Raymon Taylor and Blake Countess had issues with Tyler Lockett. This puts them in good company, as everyone who played a healthy Lockett this year got shredded. Overall Michigan finished 47th in yards per attempt and in a big pile at 23rd in interceptions despite having a mediocre-at-best pass rush. Countess and Taylor acquired ten of the seventeen interceptions, nearly all of them great plays instead of fortunate deflections. The eyeball test was pretty kind to both when in non-Lockett situations. (Also non-Indiana-tempo'd situations, which are a problem but one of a different sort than not being able to cover dudes.)

Meanwhile, the freshmen behind the starters were game but did give up a ton of yards in big chunks, whether it was because they phased out of reality or irritated the wrong gypsy. I don't think either starter is getting passed.

Then: Peppers. Historically, guys like Peppers have a breaking-in period of about half a season before emerging into a starting job late in their freshman year. Woodson, Hall, Countess, Jackson: all followed this path. With Michigan short on boundary corner types that seems like the most likely path for Peppers as well, starting the year as the third corner and gradually displacing Taylor as the season plugs along.

This doesn't leave much room for Lewis or Stribling this year. Michigan's corner depth is such that 247 is reporting that redshirt freshman Ross Douglas has been flipped to tailback despite being 5'10" and 180—the opposite of the rough and tumble brawler Michigan seems to want. While I expect both sophomores to get spot duty, a breakthrough is unlikely this time out.

This Week's Obsession: Handicapping the Position Battles

This Week's Obsession: Handicapping the Position Battles

Submitted by Seth on August 14th, 2013 at 10:38 AM


Weird line combinations and depth stuff have been coming out of practice. What should we believe, what's motivational, what's a boo boo, what's anything in this crazy world? We go under cover and bring you the NEWS that HITS the HARDEST (at least until Jake Ryan returns, because he hits harder than our news). The staff:

  • Brian Woodward
  • Seth Bernstein
  • Ace Bradlee
  • Heiko Felt, Sr.
  • Blue in South Katherine Graham
  • Watergathlete
  • Deep Throw

And the question:

Heiko and I have been arguing about this on gchat and I thought I'd bring it to the team. We want to know where you think there are real positional battles going on right now, and how you're handicapping them. For example:

Center: 55% Glasgow, 40% Miller, 5% Kugler
SAM: 85% Gordon, 15% Beyer
Norfleet: 100% Norfleet, 400% Norfleet, Norfleet% Norfleet.

Mathlete: Shouldn't it be Norfleet: 75% Slot, 20% RB, 5% DB, 100% Awesome?

Ace: The lineup seems refreshingly settled at most spots (hooray depth!); the only positions I see having real competition are tailback, outside receiver, center, and strongside linebacker, though it'll be interesting to see if there are any surprises in the defensive backfield with the return of Blake Countess and Dymonte Thomas's potential early impact at nickel. I'm operating under the assumption that Keith Heitzman earns the nod at SDE and Jibreel Black starts at three-tech with Chris Wormley playing a key role as a backup.

Starting from the top, we've discussed the running back battle ad nauseam; Fitz Toussaint should start against CMU and Derrick Green will push for more and more carries as the season wears on. I'll go 99% Toussaint (barring injury) and a 1% chance that Dennis Norfleet gets the season's first carry as Al Borges epically trolls Brian.

[Surprising (and probably meaningless) revelations and scandal, minus the scandal, after the jump]