Introduction

• Home advantage, or home bias, is common phenomenon not just in sport.
  • Bias: Decisions in financial portfolio composition, international trade.
  • Advantage: Outcomes in competitive contests distorted towards local participants.
• Home advantage more commonly investigated in team sports.
  • Pollard (2005), Pollard and Pollard (2005), Nevill et al. (1996), Koyama and Reade (2009), etc.
  • Exceptions: Koning (2011), Holder and Nevill (1997).
• Individual sports should offer less complicated picture:
  • Single individual to analyse rather than group dynamics to control for.
• We analyse 500,000+ tennis matches over many decades and consider home bias in outcomes.
  • Sample allows great variation in travel costs and other aspects of competition.

Theory and Literature

• Home bias distorts (economic) outcomes away from optimum.
• Home advantage an outcome rather than a choice:
  • Could it be outcome of tit-for-tat choices made by teach team/player?
• Home bias may exist in choice of events to enter.
  • Throughout years, fields have become much more international.
  • Representatively international?
• Does home advantage exist in individual sports as it does in team sports?
  • Previous literature suggests it is less prevalent.

Possible Causes

1. Cost of travel.
   • High costs discourage travel, fewer possible non-home wins.
   • More fatigue, expect lesser performance.
2. Unfamiliarity.
   • Players unfamiliar with surroundings, procedures, institutions, etc.
   • Ought to apply (to lesser extent?) to new tournaments in native country?
3. Support/expectation of spectators.
   • Local spectators support local players up to point.
   • At some threshold, local spectators expect particular level of outcomes.
4. Support/competition from compatriots.
   • Is it familiarity with competitors that matters?
5. Official decisions (discrimination via officials).
   • Discrimination by officials on part of spectators?

Possible Causes: Measures

1. Cost of travel.
   • Monetary cost of travel where available.
   • Impact of World Wars, 9/11 (2001), Ash cloud (2010).
2. Unfamiliarity.
   • Number of times event has previously run.
   • Number of times player has played in event previously.
3. Support/expectation of spectators.
   • Different types of event attract different interest.
   • Proxy expectations by player’s Elo ranking.
4. Support/competition from compatriots.
   • Number of compatriots in (remaining in) tournament.
5. Official decisions (discrimination via officials).
   • Cyclops at Grand Slams.
   • Hawkeye challenge system in last decade.
   • Clay courts and markings on them.

Method

• We investigate outcomes using a linear regression method.
• Dependent variable binary: \[ y_{it} = \left\{\begin{array}{lll} 1 && \text{ if alphabetically listed first player wins match,}\\ 0 && \text{ if alphabetically listed second player wins match.} \end{array}\right.\label{eq:DepVar} \]
• Regression model is: \[ \begin{align} y_{it} = \beta_0 + \sum_{j=1}^2 \left(\beta_{1j} home_{ijt} + \beta_{2j} home_{ijt}\times rank_{ijt} + \beta_{3j} home_{ijt}\times tech_{ijt} + \beta_{4j} home_{ijt}\times travel_t + \beta_{5j} X_{ijt}\right) + \eps_{it}, \end{align} \]
• Issue: Which player at home?
• home1 and home2 binary variables, or home trinary variable?
• Essentially an issue of an F-test of the restrictions; F-test statistic 1.7.
• Thus report slightly easier to comprehend ‘trinary’ results.

Data

• Collected from http://www.atpworldtour.com/:
  • All tournaments, roughly speaking, since 1877.
• Via player profiles pick up more matches: Davis Cup.
  • Also gives us player-specific tournament outcomes (ranking points, financial rewards).
  • Including player-specific info: DOB, handedness, etc.
• In total:
  • 25,589 players from 166 countries.
  • 2181 tournaments in 143 countries.
• Potential to add doubles events into picture:
  • Is home advantage stronger/weaker in team events?
  • For now we restrict attention to Davis Cup and World Team Championships.
• Potential to add financial information:
  • Prize and total financial commitment for tournaments back to 1980s.
  • Individual financial outcomes back to 1985.

Data: Descriptive

• How often are players at home over the years?

• In early years, nationality often not known for players.
  • Cost of travel would appear instinctively important.
  • WWII in particular, but also 1979 oil crisis, and recent oil price spikes.
• Do players win more when at home?

• Since the millenium, no?

Results

Results Synthesised (a bit)

• Various explanations….
• Travel?
  • Travel costs insignificant (not in table reported).
  • Little impact of `big’ one-off events (ash cloud, wars, 9/11).
• Unfamiliarity?
  • Number of times tournament played (tourn_no) matters.
  • Player experience at tournament matters.
  • More generally, age not experience appears to matter.
• Support/expectation?
  • Ranking quadratic, lesser advantage at Futures/Challenger level.
• Social impact?
  • Number of fellow nationals at tournament and remaining don’t appear to matter.
• Official decisions?
  • Home players on clay do worse than other surfaces.
• Team effect?
  • Stronger home advantage in team events (World Team Cup, Davis Cup).

Conclusions?

• Very detailed home advantage investigation:
  • Vast new dataset offering great potential.
• Highly tentative (but interesting) results:
  • No travel effect in regression but something in plot?
  • Official decisions may have impact.
  • Team events relative to individual ones (add doubles later).
• More work still to be done…