In previous parts I introduced the project and came up with a formula for franchise parity and then explored season parity. After a nice holiday break I'm going to look at game parity.
Game Parity
Game parity represents the likelihood that a worse team will still manage to win a game against a better team. In leagues with high game parity a few lucky bounces or a particularly good day on the field can lead to a big upset, whereas in low game parity leagues large upsets are very rare and there's little a bad team can do to overcome the quality difference.
Game parity, like the other forms of parity, has to be carefully balanced. If it's set right, then fans of even a losing team still feel like they have a good chance of winning any particular game, so there's more incentive to spend a night out at the ballpark or tune into a match. Too little and the results of individual matches will become nearly pre-determined as good teams will dominate bad teams. Too much and the league will seem random, frustrating fans of good teams who too often lose due to bad breaks.
Whereas franchise parity is governed by the roster construction rules of the league and season parity is governed by the organizational and playoff rules, game parity is largely governed by the rules of the game on the field. A game decided by a small number of runs resulting from a round ball hitting a round bat in just the right way is going to be much more random than a game decided by a large number of long plays or drives decided primarily by skill.
And it's no wonder that game parity is the hardest of the three to change. While leagues seem to be content fiddling with divisional structure and salary caps on an ongoing basis, changes to the play on the field are much rarer and more subtle. This is understandable, given that players and coaches train their entire lives to excel in a sport being played a certain way. Drew Brees is an excellent quarterback, and he's an excellent quarterback whether or not the Saints have a salary cap and whether or not the league has 4 or 16 playoff spots. Those external factors are mostly irrelevant to his ability on the field. But if teams were suddenly given only 3 downs to make 10 yards or forward passes were made illegal, that would dramatically change the nature of his game.
Changes to on-field rules seem to be generally limited to two categories: how ties are handled and the nature of and penalties for infractions. Leagues and fans are evidently comfortable switching around between shootouts, two-minute drills, fewer points, coinflips, and various other mechanisms to resolve ties as is evidenced by the changes in tie handling in MLS and the NHL (and the recent agitation for changes in the NFL). And changes in what actions should be considered fouls are also annual occurrences (see the two-hand check in the NBA and the recent crackdown on helmet-to-helmet contact in the NFL).
But it's evident that changes outside those categories are rare and in general you only see them when a new league is formed, rather than as a modification to an existing league. Arena Football, the CFL, and — if you go back far enough — Rugby League, the IIHF, and the NBA all took advantage of a fresh start to fiddle with the rules of the game. The last significant change to the rules of play that I can think of outside of the two categories already discussed is the addition of the three point line in the NBA in 1979 (which, as per the previous discussion about player training, immediately created the position of long range gunner from scratch). And before that the Designated Hitter in MLB in 1973 — a rule change that's still controversial.
So in short, problems with game parity are not something leagues solve by changing the rules of the game. Instead they tend to adjust around them. Leagues with high game parity require longer seasons and longer playoff series to help offset the randomness of individual results. Leagues with lower game parity can tolerate shorter playoff series (and more playoff teams) without getting excessively random results.
Methodology
The question we're asking is: what are the chances that a good team will beat a bad team in any given matchup? To answer this, I've divided the teams in each season into five roughly evenly sized groups and called the top fifth 'good teams' and the bottom fifth 'bad teams'. Then I've calculated the winning percentage (or the points percentage in soccer) of the good teams in games against the bad teams. The higher the winning percentage, the less game parity exists in that league.
The most obvious measure to rank team quality is winning percentage. However, this introduces some circularity (since we'd be using winning percentage to determine the best teams and then looking at the best teams' winning percentage) and it might obscure some of the game parity data we want to investigate, because teams that play really well and lose anyway are exposing high game parity but might not be included in our 'best teams' list because they lost anyway. So instead I'm going with goal differential or score differential, which coincide pretty well with winning percentage but don't have the same problems.
For example, in the 2010 MLB season the best teams by run differential were the Yankees, Tampa Bay, Philadelphia, San Francisco, Minnesota, and Atlanta (which incidentally is the same list you'd get with winning percentage). The worst teams were Pittsburgh, Seattle (yay), Baltimore, Kansas City, Arizona, and Houston (whereas winning percentage would have swapped Washington in for Houston). In games between one of the good teams and one of the bad teams, the good team won 65.8% of the time, or about two thirds.
In the 2009-2010 NBA season the good teams (by point differential) were the Magic, Cavaliers, Jazz, Spurs, Suns, and Lakers. The bad teams were the Timberwolves, Nets, Clippers, Pistons, Wizards, and Kings. The good teams beat the bad teams 87.4% of the time. So a bad team over good team upset (either home or away) only happened about one in ten games. Clearly MLB has higher game parity than the NBA.
To convert that percentage into a parity value, I've just taken the bad team winning percentage (or 1 minus the good team percentage) and doubled it to get it into the 0 to 1 range (since the bad teams will never be favored and therefore 0.5 is a practical max for their winning percentage). So our Game Parity value for the 2010 MLB season was 0.684 and for the 2009-10 NBA season it was 0.252.
Results
Again averaging the last three full seasons for each league, here are the Game Parity results for each of the leagues we're examining:
| League | Game Parity |
|---|---|
| MLB | 0.702 |
| MLS | 0.603 |
| EPL | 0.318 |
| NBA | 0.254 |
| NFL | 0.156 |
One thing to note here is the large difference in parity between MLS and EPL. This contradicts my earlier assertion that game parity is primarily determined by the rules of the game, since both leagues are using the same rulebook but getting different results. This is a result of extreme differences in team quality within the league affecting game parity. Referring back to the earlier parts in the series you'll see that the EPL is an extreme example of low parity (both season and franchise) and the MLS is an extreme example of high parity (likewise). The top fifth of teams in the EPL are contenders for the European Champions League, while the bottom fifth are relegated to the Championship. This massive difference in quality (resulting primarily from massive differences in payroll) means that the inherently high-parity nature of soccer (due to games being decided by a small number of goals) is skewed lower, though even those massive differences in quality can't bring soccer down to basketball or American football levels of game parity.
Keeping in mind that Game Parity is not a value that's easy (or likely) to change, we instead have to look at how each league copes with its game parity. In general, a league with high game parity should feature a longer season and longer playoff series to help offset the effects of randomness and help ensure that better teams advance. It should also have a smaller number of playoff teams (to offset the possibility of a dramatically lesser team lucking into a playoff series win). And in each case the opposite holds true for low parity sports. They can afford to let more playoff teams in (since marginal teams are likely to lose in the playoffs anyway) to shorten playoff series and to have a shorter regular season.
MLB comports itself well here. With the highest game parity of any of these leagues, it has by far the longest season, has at least a 5-game series in each playoff series, and allows the smallest percentage of teams into the playoffs. The NFL has the lowest parity, which justifies the decision to use single games rather than series to decide playoff rounds (also necessitated by the violent nature of the sport). Even in a one-game matchup, the better team is heavily favored to win. It also justifies the very short season of the NFL. It only takes 16 games in the NFL to make it entirely clear who the bad teams and the good teams are. Another 10 games per season wouldn't reveal anything.
Two leagues stand out as having structures that don't coincide well with their levels of game parity. The NBA with its low game parity can afford to let in such a large percentage of teams into the playoffs (more than half), since the lesser teams are unlikely to win. However, the NBA shouldn't need so many games in each series, and the prevalence of 7-game series has more to do with television revenue than the need to determine the better team. Also, the NBA's long season is completely unnecessary in such a low parity sport but is another concession to revenue.
MLS is much the opposite. While the season length seems sufficient to shake out the bad teams from the good, the high levels of game parity that we see with soccer almost necessitate longer playoff series to help identify the better teams. A home-and-away two legged series should be a bare minimum, with a three- or four-game series strongly considered, but instead we only get two matches for the first round, after which we go to single games to decide the championship. Similarly, the high parity should argue for reducing the number of teams in the playoffs to decrease the number of marginal teams that can go on a streak and win the championship. The results of the MLS playoffs this season and — to a lesser extent — last season are perfect illustrations of these flaws.
Next week I'll wrap up with an overview of the results and a more in depth look at each league and particularly MLS, with some recommendations on how to best manage parity in the league.