Loss aversion, Father-Son racquetball, and the Anaheim Ducks

One of my favorite forms of exercise this summer has been playing racquetball with my father, as we match up pretty well against each other. He’s been playing for almost 24 years longer than I have, but I am about 26 years younger than he is, and thus quicker, more athletic, and less prone to minor tweaks and injuries.

Over the past month or so, we’ve played 22 games, and my record is 17-5. We usually trade wins at a 60/40 clip, though I won the first 11 games that we played.

What caught my attention wasn’t just that I was winning often, but that I was consistently winning by the lowest possible score. In racquetball, the first player to reach 15 points wins unless the opposition has 14 points. Then, the player has to win by at least two points. If the score is 15-14, the game will continue until one player is two points ahead of the other; this leads to scores such as 18-16, 21-19, and 17-15.

Out of my 17 wins, I would say at least 12 of them were only by two points, with at least nine of the 11 wins during the winning streak being only by two points. When I won by more than two points, it was never a blow out, and I never won by more than five points.

My losses, however, have been nowhere near as close. When I lose, I get blown out. My most recent losses feature embarrassing scores of 15-3 and 15-7.

As a result, the total “point differential” I would have, if calculated, would likely be much closer to zero than one would expect, especially given my record.

As soon as I noted this phenomenon, I started to take note of how I played in close games, as a way to see if I could change my style of play and keep the games from getting so close in the first place.

What I noticed was that my strategy changed, almost unconsciously, based on the score of the game. If I had a lead, and I started getting close to 15 points, I would start trying for more high risk-high reward “kill shots”; I could usually afford to give my father a couple of points if I missed, and if I hit the shots, they were almost sure bets to score. The volleys would be much shorter here, and the pace of the game was much quicker, as points would be awarded quickly until someone won the game.

On the flip side, if I was losing, and my Dad started getting close to 15 points, I would play incredibly conservative. In order to keep him from scoring, I would avoid those kill shots like the plague, and instead, I would try and keep my shots in the corners in order to prevent him from getting any kill shots off on me. The volleys would usually go on for a considerable amount of time, which was in my favor; I have more stamina than my Dad, and if I could keep the game going long enough, he would tire out, and I would be able to come back and win.

So, when I need to score, I increase the pace of the game by going for lots of high risk, high reward shots.

When I need to keep my father from scoring, I slow the pace of the game down, forget all about getting points of my own, and instead make it much harder for him to score.

I play my best racquetball when I was trailing, however, as a result of loss aversion. I hate to lose more than I like to win, and my strategy for keeping my Dad from scoring points worked so well that I came back from large deficits (seven plus points) on several occasions. Unfortunately, this wasn’t something I can control, as the pressure of facing a loss is what makes me play my best.

If you’re still with me at this point, and you’re a fan of hockey (or you actually read the title), you can probably tell where this is going.

The Anaheim Ducks have become something of an anomaly to the statistical community, as they are an average team in almost every category (5 on 5 play, power play, penalty kill), and yet their record in one-goal games propels them to the top of the NHL standings. Their unique record has been noted, on several occasions, and it’s the biggest reason for their run of success of the past two seasons.

Though racquetball and hockey are two completely different sports, the mindset that one has to have late in the game in order to pull off close victories is essentially the same.

If you need to score points in order to win, it’s best to increase the chances for you to score, and go for higher percentage shots that let you get the points you need.

If you need to hold the other team, your offensive should definitely slow down. Low percentage shots on your behalf can help slow the game down, and prevent the number of chances that they have to score, making the amount of time in between each goal/point much, much longer.

With that in mind, I looked at how the Anaheim Ducks had performed in two specific situations over the past three seasons, and whether or not there is some validity to the theory that they are one of few teams in today’s NHL that “know how to win”. What I mostly came up with, however, is that the Ducks, instead of knowing how to win, know how to not lose.

Let’s get started.

When the Ducks are trailing (need to score)

The first situation that I looked at was when Anaheim was trailing late in a game. Previous research has shown us that trailing teams typically have a below average shooting percentage, though they also typically see an increase in SAT% and shot attempts for.

Most of the general rules apply to the Ducks, as they get more shot attempts on goal, and get a more high percentage of the total shot attempts whenever they trail late.

What is unique about Anaheim is that their shooting percentage actually jumps whenever they are losing in the third period, to an absurd 10.5%. Most of this is a result of the increase in scoring chances that they get; they average almost 4.8 more scoring chances for per 60 minutes while trailing in the third period, which leads the league from 2012-2015.

Over that same time span, the Ducks also have the highest among of “comeback wins” (the team trailing at the start of the third period ends up being the victors) in the league. Part of that has to do with their equally absurd shooting percentage while tied in the third period (10.6%), but I think there’s a case to be made that the players in Anaheim have a strong case of loss aversion. When they need to score, they increase the amount of opportunities that they get, and they convert on an impressive amount of them.

Here’s where the analysis gets dicey. Do we genuinely believe that Anaheim is able to elevate their shooting percentage when the threat of a loss is growing larger and larger? There is almost zero statistical data that backs this up in hockey, especially when isolated to just one team.

On the other hand, loss aversion has been shown to have effects in golf, tennis, and even NHL playoff series. Players perform better when they’re losing, and if we want to bring sociology and latent variables into the mix, then the locker room culture present in Anaheim could genuinely be birthing a team that refuses to go down without a fight.

I doubt the latent variable/loss aversion theory carries more truth than the luck theory, but either way, it’s still something that should be thought about, if nothing more.

(Just for fun, I ran a correlation between “comeback wins” and the increase a team sees in their SCF/60 while trailing in the third. The R value I got back was 0.48, so there’s definitely a relationship between increasing scoring chances and being able to come back from a deficit).

When the Ducks are in the lead (need to keep their opponents from scoring)

The second situation that I looked at was when Anaheim was in the lead late in hockey games. At this point, they don’t need to worry about scoring a goal, and instead should be focusing on reducing the number of opportunities their opponents get.

I was rather surprised to find the degree to which the team tried to follow this ideology, however.

For starters, they do manage to decrease the pace of the game, though it isn’t quite as drastic as one would expect. From 2012-2015, there were, on average, only three less scoring chances per 60 minutes for both teams when Anaheim had the lead in the third period.

Offensively, however, they buy into the “We’re in the lead, we don’t need to score again” train of thought a bit too much. While leading in the third period, the Ducks saw their shooting percentage plummet to 5.7% (as compared to 8.5% overall), and their PD0 of -1.9 is 26th in the league from 2012-2015.

As a result, they get massively outscored when they lead in the third period. Only Winnipeg has a lower goal differential in that specific situation.

The difference between the team’s performance while leading and trailing is that the scoring chance/shot attempt numbers don’t exactly support the incredibly low shooting percentage while leading. Anaheim’s SCF/60 while leading in the third period decreases by about the league average; they’re one of three teams that experience a drop in shooting percentage, and even then they see a -1.6% higher drop than the next closest team.

I don’t know if that’s just some terrible luck, or if the team genuinely stops trying to score while in the lead late. Either way, I would be willing to bet that their inability to add to their lead contributes to their middling goal differential, and incredible record in one-goal games.

Believe it or not, this actually contributes to the idea that the Anaheim Ducks suffer from loss aversion to a greater degree than the rest of the NHL. Not only would they play better while trailing, but their play while in the lead would be focused on minimizing defensive risks. We know that this is a recipe for disaster in the NHL, as playing conservatively results in dump-outs, chip-ins, and terrible possession numbers. If Anaheim genuinely had a stronger case of loss aversion, they would be pretty bad when they had the lead. The second worst goal differential in the NHL while leading late in the game? That’s pretty bad 

When the Ducks are tied (need to score AND keep their opponents from scoring)

This is by far the situation where the Ducks earn most of their one-goal wins. From 2012-2015, while tied in the second and third periods, Anaheim’s goal differential of +28 was the best in the league.

The only issue here is that they weren’t getting that goal differential through puck possession. Their PD0 in this situation was well over +3.0, which is a total that is almost unheard of and entirely unsustainable.

Given their performances in the other two score states, however, it’s easy to see how they racked up one goal wins. Late in hockey games, they dominated the opposition if the score was tied. If the ended up with a lead late in the third period, their offense dried up, leading to opposing teams coming close to tying the game, but not often succeeding due to the slower pace of the game. If they were trailing headed into the third period, they turned on their offense, and were able to have more comeback victories than any other team in the NHL.


I do think that the Anaheim Ducks are a team that, someway or another, will end up winning a decent amount of one goal games. Whether it be mental fortitude or a coaching strategy, their style of play is conducive to winning without blowing their opponents out, which not only results in one goal wins, but also, a middling goal differential.

Circling back to the racquetball analogy, there’s definitely reason to believe that I won’t continue to beat my father 77% of the time. Based on points differential alone, you would guess I would likely beat him around 55-60% of the time, but that would be underestimating my ability to win close games (God, I sound so conceited here, and part of me genuinely hates having to type that), and thus underestimating my true talent level compared to my Dad’s.  Loss aversion can be a powerful mental influence, and ignoring evidence of its effect is probably going to hurt any type of prediction. The best bet for my true winning percentage, then, would be someplace in the middle (around 65%).

When it comes to the Anaheim Ducks, there is definitely reason to believe that they aren’t a team that should be getting 66.5% of the available points in the games that they play. Going off of goal differential and score-adjusted SAT% alone, their point percentage should probably be closer to 51%, but that would be underestimating certain aspects of their game. Do they genuinely play better when the threat of loss is imminent, or have they just gotten incredibly lucky? There’s evidence to support both arguments, and in the end, the best prediction for their true point percentage is likely somewhere in the middle of their previous records, and their statistical projections (so, around 58%).

The team will have high hopes in 2015-2016 after being eliminated in Game 7 of the Western Conference Finals during the 2015 Stanley Cup Playoffs. The roster didn’t get considerably worse, and it could even be said that the team’s offseason transactions made them better.

It’s likely they won’t lead the league in possession metrics, or suddenly see a massive increase in their goal differential. If the team wants to remain atop the NHL standings again, it will have to rely on a strong record in one-goal games for the fourth consecutive season.

Whether it be a result of puck luck, loss aversion, or some other phenomena, there’s a good chance that the Ducks get that favorable record in one-goal games. Though it’s impossible to quantify latent variables by their very nature, the Anaheim Ducks just might be a case of one such variable (locker room culture resulting in a strong loss aversion that leads to their impressive record in one goal games) influencing the results of a hockey game.

Perhaps the best answer to the question lies between the purely statistical outlook, and the purely subjective outlook. The integration of qualitative and quantitative analysis is essential to NHL teams adapting to analytics, and being successful with them. For the Anaheim Ducks, this means accepting that the team isn’t as statistically solid as others around the league, but continuing to do whatever it is that results in the team’s loss aversion. The team is winning hockey games because of it, though there’s no real guarantee that it will last.

Leave a Reply

Your email address will not be published. Required fields are marked *