Analysis of Curling Team Strategy and Tactics using Curling Informatics

Hiromu Otani

, Fumito Masui

, Kohsuke Hirata

, Hitoshi Yanagi

2,3

and Michal Ptaszynski

Department of Computer Science, Kitami Institute of Technology, 165, Kouen-cho, Kitami, Japan

Common Course, Kitami Institute of Technology, 165, Kouen-cho, Kitami, Japan

High Performance Committee, Japan Curling Association, 1-1-1, Jin-nan, Shibuya-ku, Tokyo, Japan

Keywords:

Curling Informatics, Digital Scorebook, Winter Olympics, Tactics and Strategy, Game Information, Differ-

ence in Shot Accuracies, Difference in Game Scores, Correlation.

Abstract:

The 2015-2016 season became a historic season for Japanese curling. Japan national curling team has won

a silver medal for the ﬁrst time at the 46th Women’s World Championship 2016 in Canada. However, it is

still necessary to work on strengthening the team performance in order to aim for the top. Such strengthening

needs to include material factors such as the physical factor and the human factor, but also the strategic/tactical

factor which is crucial in curling. Bradley (2009) points out the strategic/tactical factor as the most important

at top level. As an example of a research aimed at supporting such strengthening of the strategic/tactical

factor, Masui et al. (2016) proposed the concept of Curling Informatics. They built an environment for

strategic/tactical support which makes use of ICT by allowing digitally collect and analyze game informations

in real time. Speciﬁcally, as the ﬁrst step of implementation they developed a digital scorebook iCE as a

method for digital collection of game information for further analysis. In this research we analyzed game

information collected with iCE to establish the effective analysis for tactical support and verify the knowledge

which can be obtained empirically and what kind of new knowledge can be obtained from it. We report on the

new knowledge we obtained regarding the relationship between shot accuracy and difference in game scores,

and difference in correlation for each level in 93 games collected during the 2013-2014 season.

1 INTRODUCTION

Japan national curling team won second place at

the Women’s World Championship 2016 in Canada,

Saskatchewan. It was the ﬁrst time a Japanese team

won a medal in curling. In the background of this suc-

cess, there are various strengthening programs devel-

oped for Japanese curling (Yanagi, 2011; Takahashi,

2011).

However, it is still necessary to continue working

on strengthening such support in order to make Japan

capable of obtaining a gold medal at the Pyeongchang

Winter Olympic to take place in two years. For exam-

ple, Japan played against Switzerland three times in-

cluding round robin

. This suggests that Japan should

have already captured the tactics of Switzerland team

which should help in winning a gold medal or at least

improve the team’s ranking during the Pyeongchang

Winter Olympic in 2018 .

Type of a tournament in which every team competes

against every other team in turns.

An Information Science approach can be men-

tioned as an example of a method for improving the

performance of a team. In the past few years a num-

ber of cases to support sports with ICT (Information

and Communication Technology) have been reported

(Fujimura, 2004; Kagawa, 2006), including curling

(Masui, 2016).

Masui et al. (Masui et al. 2016) have proposed

a new ﬁeld called Curling Informatics. As Figure 1

shows, this research ﬁeld deals with the strategic and

tactical factors of top curling teams and improves the

strategic and tactical skills of curling players. It also

focuses on realizing several support environments,

such as recording and referring to game information,

tactical navigation, and assistance for reﬂection and

tactical training. Speciﬁcally, general plan developed

within this ﬁeld aims at implementing the following

methods to (1) collect, (2) analyze, (3) visualize and

(4) share game information. In the ﬁrst step, they de-

veloped the digital scorebook iCE which runs on a

tablet computer to collect and analyze game informa-

tion and they conﬁrmed validity of the system. By

182

Otani, H., Masui, F., Hirata, K., Yanagi, H. and Ptaszynski, M.

Analysis of Curling Team Strategy and Tactics using Curling Informatics.

DOI: 10.5220/0006044601820187

In Proceedings of the 4th International Congress on Sport Sciences Research and Technology Support (icSPORTS 2016), pages 182-187

ISBN: 978-989-758-205-9

Figure 1: The concept of Curling Informatics.

the application of iCE, it is possible to check the shot

accuracy of each team or player based on collected

information sequentially and visually. In the near fu-

ture, they plan to analyze game information stored

with iCE in detail.

As a part of Curling Informatics, in this study we

manually analyze game information with the aim to

specify the team’s characteristics and establish the ap-

propriate analytical method. Especially, we discuss

the inﬂuence of shot accuracy against the game result

by focusing on the relationship between shot accuracy

and score.

The outline of the paper is as follows. Firstly, we

describe the team strategy and shot accuracy in Sec-

tion 2. Next, we describe an overview of targeted

game information, and the analytical method. The

results are discussed and explained in Section 4. Fi-

nally, in Section 5 we conclude the paper.

2 TEAM STRATEGY/TACTICS

AND SHOT ACCURACY

Curling is a winter sport in which two teams compete

to obtain points by throwing 16 stones at the center of

a circular area called a house in an ice-based square

area called a sheet. One team consists of four play-

ers. Each player throws two stones (called shot) in a

rotation, and the score is calculated by each rotation,

when all 16 stones have been thrown. At this time,

The team that has a stone at the nearest position to the

center of the house gets the number of points equal to

the number of stones on the inside of the opponent’s

stones in the house. The scoring team has the ﬁrst

move in the next rotation.

This rotation is called an end. One game consists

of eight or ten ends, but the game is extended if the

score is tied in the ﬁnal end (Howard, 2009; Coleman,

2014).

In addition, Curling is often called “chess on ice”

because it is a kind of sport in which tactics is very

important. It requires a player to form complex strate-

gies in search for effective moves with consideration

of the ice condition and stone position in the house.

Shots in curling are roughly divided into two

types: draw shots and takeout shots. In draw shots the

stone stops in the house. Takeout shots force opponent

stones out of the house. The thrown shot is recorded

by a team coach or a substitute player and is given

from 0 to 4 points. In the special case that affects the

position, such as takeout shot which hits more oppo-

site stones then advised by the skip, 5 points can be

given. These points are called “shot-score” in curl-

ing. Additionally, shot accuracy is calculated. Shot

accuracy is the percentage that represents the overall

accuracy of the shots and is based on shot-score of

each end or the whole game, according to equation

(1) and is one of important measurements to estimate

player’s skill and condition.

shot accuracy =

total of shot −scores

number of shots

× 25(%) (1)

In general, the higher accuracy shots there are, the

more accurately the team was able to throw shots.

Therefore, it is advantageous in the game. The team

Analysis of Curling Team Strategy and Tactics using Curling Informatics

183

Figure 2: Shot accuracy for each level.

which has higher shot accuracy is more likely to suc-

ceed when drawing the shot in critical situations. And

the team which has lower shot accuracy will raise the

probability for scoring by opposite team.

iCE compiles shot score databases from basic data

of shot accuracy, which makes it possible to organize

and analyze scores from a variety of viewpoints. To

grasp the characteristics and trends for each teams we

could analyze this information on shot scores in more

detail. And this in result could lead to ﬁnding the dif-

ference of strategy between a rival country and Japan

in an objective way.

Masui et al. (Masui et al. 2016) suggested that

there is a strong correlation between the difference in

shot accuracies and the difference in the game scores.

In addition, they observed that Japanese national class

is stronger than Japanese junior national class. And

Japanese national class is rarely inﬂuenced by missed

shots. It means that the game result could be predicted

before the game ends if we knew the difference in the

playing teams’ shot accuracy. Furthermore, the cor-

relation in world class becomes weaker than Japanese

national class because of smaller differences in shot

accuracies.

3 METHOD AND TARGET DATA

To verify the analysis of Masui et al. (Masui et al.

2016) mentioned in Section 2, we analyzed the re-

lationship between difference in shot accuracy and

score for game information of world class teams.

Speciﬁcally, we extracted the difference in scores

and shot accuracy for each collected game. Next, we

calculated the difference in shot accuracy and Pear-

son’s correlation coefﬁcient based on extracted data.

In addition, we examined a correlation between dif-

ference in scores and difference in shot accuracy.

Table 1: Target game information for analysis.

Year Championships Number of games

30th Japan Championship 26

PACC Japan Palyoff 16

Paciﬁc Asia Junior Championship 20

2012 World Junior Championship 11

3th College Championship 10

Universiade 2013 of Japan Playoff 7

Japan Junior Championship 24

31th Japan Championship 11

2013 Olymipic Winter Games of Japan Playoff 21

4th College Championship 11

5th College Championship 13

2014 Universiade 2015 of Japan Playoff 30

Advics Cup 13

Sochi Olympic Winter Game 93

2015 World Woman’s Championship 72

Total 378

The target data is the game information in the

database for 285 games collected by Masui et al. (Ma-

sui et al. 2016) and 93 games of Olympic Winter

Games 2014 (total of 378 games, covering around

sixty thousand shots). Table 1 shows target game in-

formation for analysis in detail.

4 RESULT AND DISCUSSION

In this section, we describe the relationship between

difference in shot accuracy and difference in scores.

Figure 2 shows team shot accuracies for 285 games

collected by (Masui et al. 2016). In Figure 2 , the

X axis shows level, the Y axis shows shot accuracies

each game and the Error bar stands for standard devia-

tion. As shown in Figure 2, higher level teams achieve

higher shot accuracy and lower standard deviation. In

addition, we calculated Pearson’s correlation coefﬁ-

cient between difference in shot accuracy and differ-

ence in scores for each level. Japanese junior top level

was 0.72, Japanese national top level was 0.80 and

World national top level was 0.74, World Best4 was

0.68. It means that the correlation for Wold class is

lower than Japanese class.

icSPORTS 2016 - 4th International Congress on Sport Sciences Research and Technology Support

184

Figure 3: Correlation between difference of shot accuracies

vs. difference of game scores for the Sochi Winter Olympic

Games.

Figure 3 shows a diagram representing a correla-

tion between difference in shot accuracy, and differ-

ence in scores and regression line for 93 newly col-

lected games. In Figure 3 , the X axis shows differ-

ence in teams’ shot accuracies each game and the Y

axis represents the difference in game score.

Figure 3 indicates considerable positive correla-

tion between difference in shot accuracy and differ-

ence in scores. Pearson correlation coefﬁcient be-

tween the two differences for total data was 0.68. This

result was lower than the above mentioned Japanese

junior top level. However, teams’ shot accuracies was

82.01±5.80S.D. It was higher than Japanese national

top level in Figure 2. In other words, game informa-

tion of world class teams indicated a very high shot

accuracy. And yet the correlation between the two

differences was low. These results were similar to

those of Masui et al 2016. This means that tactics

or planning each end has an impact on game result or

difference of game scores

It could be possible that the correlation is nega-

tively inﬂuenced by outliers. The more games con-

taining outliers, the lower the correlation. It can be

expected that we could expose the process of how tac-

tics or strategies affect game result or difference in

game scores by analyzing the games of outliers.

Next, we analyzed two games in which the team

of superior shot accuracy lost due to failure in tactics.

Figure 4 and Figure 5 represent graphs showing

transition of teams’ shot accuracies for every end of

each game. The X axis means number of ends and

the Y axis shows team shot accuracy.

As shown in Figure 4, game deployment does not

indicate the difference in game scores until the mid-

dle of the game. While near the end of the game, team

A gained multiple scores and won the game. As the

rate of shots performed by each team, ratio of take-

out shots for team A was 64% (51 shots in total of 80

shots) and ratio of draw shots was 36% (29 shots in

total of 80 shots). Team A performed more takeouts

than draws. On the other hand, ratio of takeout shots

For example, intentionally performing a missed shot.

for team B was 51% (41 shots in total of 80 shots) and

ratio of draw shots was 49% (39 shots in total of 80

shots). Team B performed similar number of takeout

and draw shots. Takeout shots reduce the score prob-

ability of opponent team because they forced a stone

out of the house. Draw shots raise the score prob-

ability of one’s own team because they accumulated

stones the house. In short, as the cause of victory it

can be considered that team A took the tactics of risk

aversion by performing selected takeout shots and ac-

curately taking advantage of missed shots performed

by team B. In fact, in third end and ninth end, team

A obtained their scores because the situation changed

due to missed shot of team B.

As Figure 5 shows team A’s draw shot accuracies

is 100% from fourth end to eighth end. However, they

made some scores only at the ﬁfth end. Also in this

game, team B which won the game had thetakeout

shots as 64% (49 shots in total of 76 shots) and the

draw shots as 36% (27 shots in total of 76 shots).

It means that team B performed more takeouts than

draws. In the game information of world class, the

teams’ shot accuracies exceed 80% and a standard de-

viation is small. It suggests that one missed shot can

have an impact on match situation more than in games

of Japanese national top level. Also in this game, it

can be considered that tactics of team B was based on

purposeful using missed shots of team A.

Therefore, as the cause of victory we can deter-

mine that team A took the tactics of risk aversion as

their priority. Thus the selected tactics and a con-

tributing shot (or miss) had an impact on game result.

5 CONCLUSION

In this paper, we performed the analysis of game in-

formation of a number of curling game matches by

using the digital scorebook iCE developed by Masui

et al. (Masui et al 2016).

The result suggested that the difference of shot

accuracies is related to the difference of the game

scores. Also, we conﬁrmed that this correlation is

lower at the world class. Furthermore, we analyzed

the game information of outliers from tactical point

of view. It was proven that the selected tactics and

a contributing shot (or miss) had an impact on game

result.

In the near future, we plan to record game infor-

mation of World national top level and analyze it in

detail. In addition, we will specify the process of how

the team strategy/tactics inﬂuences the game results

or the difference of game scores.

Analysis of Curling Team Strategy and Tactics using Curling Informatics

185

Figure 4: Team’s shot accuracies for every ends and transition of game score(1).

Figure 5: Team’s shot accuracies for every ends and transition of game score(2).

ACKNOWLEDGEMENTS

This work was supported by JSPS KAKENHI Grant

Number 15H02797.

REFERENCES

Bradley, L, J. (2009). The sports science of curling. A Prac-

tical Review, Journal of Sports Science and Medicine

vol.8, pp.495-500.

Coleman, G. (2014). Introduction to curling strategy (en-

glish edition). Amazon Services International.

Fujimura, A. (2004). Quantitative evaluation of sport team-

work using generalized voronoi diagrams. IEICE

Transactions D J87-D2:818-828.

Howard, R. (2009). Curl to win: Expert advice to improve

your game. HarperCollins Publishers Ltd.

Kagawa, M. (2006). Effect of multimedia information on

web pages in physical training class of university.

Journal of Japan Society for Educational Technology

29 37-40.

Masui, F. (2016). Informatics to support tactics and strate-

gies in curling. Int J of Automation Technology10(2).

icSPORTS 2016 - 4th International Congress on Sport Sciences Research and Technology Support

186

Takahashi, H. (2011). Support the japan womens curling

national team by a trainer. Journal of Training Science

for Exercise and Sport 23(1):7-12.

Yanagi, H. (2011). Training science on college curling

team. Journal of Training Science for Exercise and

Sport 23(1):13-19.

Analysis of Curling Team Strategy and Tactics using Curling Informatics

187