Overfitting Celebrity Pitches

The Washington Post created a fun infographic of celebrity baseball pitches.

I use this graphic in my Data Analytics class. Students are tempted to draw inferences about individuals from this data set. John Wall and Michael Jordan are great athletes, but in this case they are underperforming Avril Lavigne and George W. Bush. Do we conclude that Sonia Sotomayor missed her calling as an MLB player?

The first lesson here is that we should not assume we can predict where Harrison Ford’s next pitch will go based on observing just one pitch. A single pitch should be considered a random draw from a distribution centered around Ford’s average ability. Any single pitch could be an outlier.

Snoop Dog features twice on this graph. In 2012 he got the ball in the strike zone. Had we only seen that, we would want to conclude that he is a great pitcher. However, in 2016 he was way off to the right. In either case, overconfidence that he is predictably near a single pitch would have been a mistake.

Lastly, I use this graph to illustrate the concept of overfitting (investopedia definition). I suggest a model that is obviously inappropriate. What if we conclude from these data that anyone with the last name of Bieber will not be able to throw the ball in the strike zone? That model surely will not generalize. The problem is that if we test that prediction on the same data we used to train the model, the misclassification rate will be zero. If possible, start with a large data set and set aside some portion of the data for validation, before training a model. Having validation data for assessment is a good way to check that you haven’t modeled the noise in your training set.

The Revealed Preferences of the National Hockey League

American sports leagues are different from their international counterparts for a variety of reasons, but perhaps the simplest and most important is that they exist as singular entities, otherwise natural cartels whose network effects are explicitly codified as clubs whose barriers to entry ensure a steady stream of profits so long as their sport remains sufficiently popular. Negotiating against player unions of varying levels of organization, they routinely negotiate collective bargaining agreements that neatly establish the division of proceeds between capital and labor.

A common mistake made is questioning the choices made by teams as if they were independent firms competing against each other in a ruthless marketplace for economic survival like Sony, McDonalds, or Manchester United, when in fact their survival is largely pre-ordained by the cartel, their choices salient only to the prestige and short-term windfall profits of annual trophies.

Tom Wilson plays for the Washington Capitals, which happens to be my favorite team in the National Hockey League. He is extremely good at hockey. He scores goals, makes good choices in transition, plays commendable defense, and is extremely adept at physically hurting other players. It is for this last bit that he has received the most attention. His team gains a notable advantage when he is on the ice simply because the other team must allocate a disproportionate amount of their attention to where Wilson is and their own relative vulnerability. The other teams in the league, and many of their players, are increasingly of the publicly held opinion that this advantage is not gained in a manner within the rules of the game. Tom Wilson is a cheating bully who threatens the safety of every other player beyond an acceptable level who simply must be stopped immediately.

To be clear, they do not believe this.

The other teams and their players believe he is dangerous (he is). But they clearly do not think he is too dangerous. Tom Wilson is occasionally suspended or fined, his salary donated to charity. The players’ union (the NHLPA) has worked tirelessly to minimize the punishments he incurs for physically injuring the other members of the same union. The other teams within the league cartel has never once imposed a punishment on his employing team. Based on the relatively modest punishments doled out and the minimal interest the players union has in ensuring their members’ physical safety, it would be foolish to conclude that the NHL views Tom Wilson as a net negative or even symptomatic of a net negative institution within hockey.

The NHL sells hockey. Their cartel members aren’t competing with each other, they are competing as a league against other sources of entertainment, principally other sports. They are competing for attention. Three John Wick movies have left me convinced that violence is an excellent means of eliciting attention. The NHL isn’t punishing Tom Wilson or the Washington Capitals because every time he punches a player prone on the ice in the back of their neck, the possibility that a player may be paralyzed or killed receives twenty-five fold the attention that Connor MacDavid receives for being the most skilled player I’ve ever seen.

To be clear, the NHL doesn’t sell hockey or violence, they sell a bundle of goods that includes athletic skill, regional identity, cultural identity, and violence. Compared to the other major US sports, it’s not unreasonable to consider the violence within hockey to be the bundle component that overlaps the least with other competing products and, as such, contributes the most, at the margin, to their share of the market. Violence may literally be the most profitably thing the NHL sells.

Every time Tom Wilson or another players seriously injures a player, possibly ending a career or reducing the quality of the rest of their life, people will speculate on what sort of event will cause the NHL to change the nature of their sport, but I don’t know why there is any uncertainty.

They’ll change when revenues decline because fans prefer less violence in their sports entertainment consumption or when young athletes with brief peak earning windows express willingness to receive smaller wages in exchange for safer working conditions. Such things have been happening steadily for the last 25 years with all of the major sports, but hockey has put itself in a uniquely bad position to continue transitioning away from selling violence, one what may demand that teams earn smaller profits, and players smaller wages, in the short run in order to enjoy greater success in the long run. I guess it could happen naturally through artful negotiation, earned trust, and thoughtful planning.

You ever know a joke that you know only a small fraction of people will understand, but you tell it anyway?

The Top Shot Blockchain Phenomenon

Samford Business student Wes Crane writes: Remember when people used to buy and sell physical sports trading cards? Recently the NBA has partnered with Dapper Labs, a company that specializes in blockchain (the same technology used for Bitcoin and Ethereum), to create and sell a digital art form of non-fungible tokens, or “NFT”s, that contain video clips of highlight moments which can be traded online at https://nbatopshot.com.

Continue reading

I could do better

My favorite soccer team has been badly coached for 2 years and I am regularly convinced I could do better.

These are not the thoughts of a rational man and its causing me no small amount of consternation, bordering on intellectual crisis. Which is, of course, a lie, but adding a touch of intellectual melodrama never hurts when you’re trying your damnedest to write something new every week.

It is a puzzle, to be sure. There have been two coaches in the last two years, the second having only been there a week. The first was experienced, accomplished, and internationally famous. I’m quite confident he was wrong in the majority of decisions he made, but I at least had a model for why he was so often wrong.

When an ostensible expert appears to be failing at their job far worse than a hypothetically cheaper replacement, I always look for the rational reason why someone might be choosing to fail. In this case, we were observing an individual who could achieve mediocrity without effort. His past accomplishments gave him credibility with the players and his stock of knowledge as of 2011 was sufficient to carry him to large pay checks. To achieve mediocrity required near minimal effort. Could he update his tactics, both within the structure of the game and his management of personnel? Of course. But doing so would require enormous amounts of effort. His salary had peaked, his future managerial prospects dimmed by aged and recent results, and as such the returns to effort were dwarfed by the returns to leisure. And, should you be so inclined, allow me to enter in ego into the calculus. What sounds for cognitively costly: acquiescing to reality that your human capital has been rendered obsolete and rebuilding your modus operandi from scratch with the full knowledge that you may spend your wealth-laden golden years failing in public? Or denying it fully, shifting all blame for failure onto the personnel, and bemoaning that it is not your human capital that is obsolete, but rather that the labor pool available to you is fundamentally flawed? To me its a no-brainer, and it’s why I am fully of the belief that there actually are bloggers in their mom‘s basement who could have better managed a team.

The new manager is a temp. He’s never managed a team before. Then again, neither have I. He has, however, played professional soccer at the highest level. He has been placed on the management training track by a world-class organization. He has none of the maladapted human capital or rational-addiction-adjacent reasons to fail at his job. He has all of the local and tacit knowledge from being on the training pitch and in the locker room that I don’t.

I’m still confident I could have done a better job than he did today. Why is that?

I can construct a model to rationalize my beliefs, but that model gets awfully “just-so” very quickly. It relies on assumptions I can’t justify and broad generalizations that, if evenly applied, would hurt the case for myself as superior even more so than the current job holder. Of course, I can invent a narrative where I am the superior sports team manager, but that narrative would have to rewrite my entire personal history going back so far as to render me a completely differ human, and one who no doubt would have just as many (and possibly the same) blind spots.

I guess what I’m saying is that I know I shouldn’t be the manager. Every rational bone in my body knows that is a silly idea and I would fail miserably. But I think there is a case to be made that sometimes we can look at the person making decisions for our favorite team, look at their track record, and confidently say “They would be making better decisions if they talked it over with me.” When the armchair quarterback says ‘the coach is an idiot” they’re not saying they want to be the coach. They’re saying they want to be in the room. They want a voice because they think they could contribute.

Someone tell Tottenham Hotspur that I’m available. I’m not free, but I can be had.

Moneyball for March Madness

Sinclaire Green, a Samford business school student, writes:

Since Billy Beane transformed the 2002 Oakland Athletics baseball team by utilizing data analytics, propelling the team to a 20 consecutive game win streak, sports fans, coaches, and players have all become more attuned to the role data can play in baseball. As a result, professional basketball teams also began to use data analytics to improve their game plans, skills, and recruiting. In the past few years, Colton Houston and Matt Dover have begun to use data analytics to help college basketball teams similarly to how Billy Beane helped transform the Oakland A’s in the early 2000s.

Houston and Dover’s company, HD Intelligence, provides a service to college basketball that was previously only feasible for professional teams. HD Intelligence eliminates the need for internal data analysts. Analysists at HD Intelligence compile data and present it to college basketball coaches to improve decision making. HD Intelligence prides themselves on making meaningful insights that coaches can understand. Instead of coaches having to rely on watching video and looking at statistics from box scores, HD Intelligence provides reports for teams. Coaches can know their team better, know their opponent better, evaluate recruits effectively, and optimize their schedules.

In the 2019-2020 basketball season, HD Intelligence had two primary college basketball clients, The University of Dayton Flyers and the University of Alabama Crimson Tide. Similar to the Oakland A’s, Dayton does not have quite as robust of a budget as many of the nation’s other top programs. Also similar to the Oakland A’s, the Flyers had a 20-game win streak last year and many basketball aficionados think that Dayton would have won the NCAA tournament had the COVID-19 pandemic not halted the tournament. Similarly, Alabama basketball has had an excellent 2020-2021 season. The program has risen within the SEC to win the 2021 conference tournament. More importantly, the Tide made themselves a legitimate contender for the national title by making it to the Sweet 16 in the NCAA tournament. Even though Alabama did not make it to the Elite Eight, they had a great season with many wins. HD Intelligence helped both Dayton and Alabama optimize their talent and resources by providing data analysis of each game and by assisting with pre-season non-conference scheduling for the two programs. Looking to next basketball season, Houston and Dover have over 10 schools who they will assist with data analytics.

Rules versus Discretion in Soccer a.k.a. Football

It has been pointed out that the only thing funnier than “someone who never played the game” trying to improve soccer is someone who calls football “soccer”. Admittedly, the nomenclature I use is endogenous to my audience, but that is neither here nor there.

The rules of football are perfectly distilled examples of the merits of rules versus discretion in optimal policy. Historically, the “laws of the game” codified by FIFA was a relatively sparse tomb that made copious use of the phrase “in the opinion of the referee”. This reliance on referee discretion has contributed as much to the evolving game as globalization, nutrition, and greater athleticism. YouTube is a wonderful place to watch footage of older matches and stare aghast as the world’s very best players try to shatter each other’s tibias and femurs every few minutes. (Brief digression: to really appreciate this, watch this collection of George Best dribbling and note how to succeed at dribbling at any time meant the opposition would inevitably try to break his legs). As our cultural norms have shifted, away from preferences for ultraviolence on fields of play, so too have the enforcement norms amongst football referees. This is at the margin, mind you, with many inframarginal fans and participants left indignant by the cowardice imposed on the game.

I previously questioned the overreliance of the game on referee discretion, much the way I sometimes questioned discretion in monetary policy. My views on monetary policy have shifted somewhat away from pure rules commitment, in part because of what I have viewed in football, and the introduction of a massive institutional shift away from discretion in two dimensions has been nothing sort of disastrous for the experience of both watching and playing a high-level professional game. The introduction of Video Assisted Replay (VAR) and its application to the enforcement of 1) Offsides and 2) Handball infractions has changed not just how the game is optimally played, but the entire emotional arc of observing and playing the game.

  1. Offsides
Premier League are considering removing the drawn lines when using VAR for offside calls

The offsides rule is quickly defined as such: you may only pass the ball to a player who has between themselves and the goal either i) the ball or ii) two opposition players. The two player bit always confuses people until they remember that the goalkeeper counts as a player.

The rule is, historically speaking, revealing in its continued structure. First, it is on its face silly that they’ve never changed the rule to “one non-goalkeeper between the player and the goal” and make an already cognitively challenging rule to monitor that much easier to enforce. It seems like the kind of rule change easily smuggled in with little opposition, not unlike eliminating offsides for throw-ins, which happened roughly a century ago. Second, there has always been a looseness to what body parts can place a player offsides – in a sport where hands cannot be used, it seems odd that they might shift the imagined lines.

For the purposes of this discussion, what matters is that what might seem a rule with little gray area was actually rife with two forms of important discretion. First, the already alluded to application to body parts. Second, given that a linesmen 40 yards from the middle of the pitch must track the ball and players often 50 yards from it, the triangle of vision they must manage simply does not allow for fine-grained analysis. They’re not tracking limbs, they’re tracking center mass, and barely at that. All of this adds up to enforcement guided by the discretion of the referee and the norms that inform them. Those key norms over time boiled down to 1) Even is onside, 2) When in doubt, the benefit is given to the attacking player, 3) the only parts that should matter are those that can play the ball i.e. flailing arms don’t count. Defensive lines were welcome to play an offside trap i.e. a high line with a governing centerback managing the line and yelling “step!” when a key offensive player might be put in an offside position. But such a strategy came with the risk of relying on a cognitively overwhelmed linesman not leaving your hapless goalkeeper one-on-one against a marauding forward.

VAR was erroneously introduced on the false premise that the weakness of offsides enforcement was the fuzziness of observation when, in fact, the entire institution was predicated on that fuzziness. Without that fuzziness, the advantage shifts strongly to the defensive player(s) because they are facing the passer, giving them the half-second to step forward, placing the offensive player offsides. Such a strategy was too much of a gamble before – placing a player 3 inches offside was sufficiently unlikely to be acknowledged, and goals too scarce, to warrant frequent reliance. Further, as it turns out, the “even is on” norm is critical to offensive counter-attacking. Which leads to the single greatest error in the introduction of VAR: the comically thin 1-pixel lines with which positioning is assessed. Presented with the fallacy that video technology could assess position with <1inch precision, “even is on” ceased to exist because effectively no two players would ever be deemed even. Without additional explication, I will simply note that assessing when a pass was executed is sufficiently fuzzy that <1 inch precision is not on offer.

How to fix it

It’s actually fairly simple in this case – you reconstruct VAR to mimic the ideal referee of the past. You make the lines wider. If those lines overlap at all, the players are deemed even. The system should either a) assign the body parts that are relevant to the rule and make the lines 6-10 centimeters wide or, b) construct the lines at center mass and make them 15-25 centimeters wide. In this manner, we get the best of both worlds – the key elements revealed by 100 years of discretionary enforcement and the uniformity of computational augmentation.

2. Handball

FIFA's takeover of the handball law has led to greater controversy and uncertainty

There is this silly posture that players now frequently assume, with their arms hidden behind their backs, as they try to move about athletically without the balance of their arms (it’s really hard, try it some time). This posture is a product of the rule that any violation inside the your defensive penalty box results in a penalty kick which results in a goal roughly ~70% of the time. That’s a very high value event when fixtures average fewer than 3 total goals per game, and is actually even higher value when you consider that scoring opportunities are endogenous to the current score i.e. teams are more conservative with a lead.

In case you did not know, you can’t use your hands, or arms, in football. You’d be within your rights to imagine that players have always gone about pegging the ball at the other teams arms when in the penalty box, trying to draw penalty kicks. You’d have been largely wrong though, for the simple reason that referees have historically been reluctant to reward such tactics. Under the loosely codified rubric of intent of action, proximity to the strike of the ball that eventually contacted their arm, awareness of the ball, or other such language, referees repeatedly made it clear that the penalty they least desired to award was one for a nebulous handball.

VAR stepped in, along with some mind-bogglingly stupid reinterpretation of the handball rule and said “Nope, if it touches the defending player’s arm in the box, it’s a penalty, and absolute chaos ensued. professional players quickly realized that any outstretched arm was to be chipped at and any leaping defender was to be collided with, in the hopes of producing a random arm-ball contact and, in turn, seven-tenths of a goal. Everyone hated it.

So what went wrong? Once again, it’s a case where the equilibrium of the game had evolved to entirely depend on discretion. The penalty box, and its single-sanction system for violations, was designed to deter teams for being overzealous in defense, and give attacking teams fair opportunity to score. That single-sanction, however, was so strong, that it only held in equilibrium through its discretionary, and therefore unpredictable, application by the referee. Sure, it’s arguably the single most powerful referee tool in major sports, but given that soccer is the most popular sport in the world, it certainly warrants respect as a stable second-best solution.

When VAR robbed the institution of its discretionary fuzziness, however, the equilibrium was shattered and the combination of a single-sanction system with a rule that cannot be perfectly complied with, well, the game was sent into minor chaos. The de factor rules had shifted so rapidly and so completely that neither players nor fans understood what was happening. Everyone was thinking more about if and when balls were in contact with arms than if and when it might go in the net. That’s not the ideal equilibrium for a sport, particularly if you’d prefer highlights that are more than penalty kicks (which make for exceedingly boring viewing).

How to fix it

There’s no getting rid of the handball rule. There’s no way to eliminate random contact with arms/hands. There’s no way to adjudicate intent well. If you can’t change the rule and you can’t change the monitoring quality, there’s only one thing left to do: change the punishment. Football should walk away from the single-sanction system.

Indirect free kicks from the spot of the handball

Why Indirect free kicks in the penalty box?

  1. They offer lower expected value than penalties,
  2. The expected value that does emerge reflects the ability of both teams, rather than just the shooter and goalkeeper.
  3. They sneak in a little bit of referee discretion when they identify the spot of the foul i.e. location determines value.
  4. They are fun as hell.

Which obviously brings me to monetary policy

Classic arguments about rules versus discretion are typically about constitutional constraints of elected and appointed officials. Maybe the most salient to our modern lives is how the Federal Reserve should go about it’s policy of increasing or decreasing the money supply i.e. what is the optimal amount of inflation? Regardless of what that number might be, or whether it should be adjusted with the assessed stage of the business cycle, the underlying argument is really about whether or not the targeted number should be chosen by people or set by a rule to which we are bound by a codified pre-commitment.

I myself was once a hard line “rules” person, or at least as hard line as one can be without being a monetary economist by field or training (for an informed opinion, ask this guy). Over time, though, I’ve come to appreciate that rules only work if we know what we are doing when we set them and if we can credibly commit to them. These are big “if’s”. The reality is that there is no such thing as a “pure rule” setting – some amount of discretion is always baked in. If you can’t identify exactly where the status quo discretion is, or, more importantly, why we arrived at the current equilibrium level of discretion, you should proceed with extreme caution. You may find that the discretion you didn’t know was there was the only thing keeping the system afloat this whole time.

Camping in Style in an “Instant Cabin” Tent

This is the time of year when we often think of gifts to give to others, or for others to give to us, if they are so moved. So I will share an item which took a bit of research to lock in on, and which has worked out very well in practice.

When I was in my teens, I was content to throw a sleeping bag on a tarp right on the ground when camping. In my 20s, I used a half inch thick dense foam pad, a classic Ridge Rest. I wanted a little more cushion under me in my 30s, and so graduated to a 1.5 inch thick self-inflating sleeping pad like this Stansport. For backpacking in my 40s and 50s, I craved yet more air space underneath me, especially for curling up on my side, and got good usage out of a narrow, 2.5 inch thick inflatable sleeping pad.

Now my wife and I are pretty much done with roughing it. We still enjoy the great outdoors, but find we enjoy it even more when we have essentially all the comforts of home, which includes a full size queen air mattress. It takes a pretty big tent to accommodate that plus all our other gear, without feeling squashed.

I have had some large tents in the past, which were very tedious to set up. So I was pleased to find a huge, airy tent, which almost erects itself. This is the Ozark Trails 9-Person Instant Cabin.

Image from Amazon website

The main room is 9 x 14 ft, which is plenty big for glamor-camping (glamping) for two people. In huddled masses mode, probably 8 bodies would fit comfortably. The tent has a screen room across the front, for a bug-free place to sit. The fly over the screen room provides a roof over the door to the main room, keeping out rain even when the tent door is opened.  Here are two views from within on our latest camping trip, first looking out the door through the screen room, and then looking straight up through the roof before we put the fly over the tent at the end of the day.

As an engineer, I am tickled by the clever joints that allow you to make the structure arise with just a few strategic tugs. Going from stage 2 to stage 3 in the photo below takes all of fifteen seconds. Taking the tent down for storage simply involves doing all these motions in reverse. The tent itself stays always attached to the poles.

Image from Amazon website

The only major drawback is the price, about $300, which is a lot for a tent. Considering our wants at this stage, for us that is a fair exchange. It gives us much of the space and utility of a small pop-up camper trailer, for a fraction of the cost.