Rink Bias: Adjusted Turnovers

In the RTSS (Real time scoring statistics) the nhl tracks turnovers(forced and unforced).  To clarify, an unforced turnover is known as a “giveaway”, where as a forced turnover occurs when a team’s opponent has a takeaway.  For example, say Chicago is noted in the box score with “8 defensive takeaways” vs San Jose, this means that San Jose had 8 “forced turnovers” which are inefficent offensive possessions.  Per my possession formula, I refer to these forced turnovers as “takeaways against”. Teams can have two types of turnovers:  “Giveaways” and “Takeaways against” In sorting through my data, I noticed large discrepancies with in each box score where giveaway and takeaway against totals seemed very high for some games as opposed to others.

The first step is to compare the home rinks opponent’s documented giveaways and takeaways against with how their opponents performed on the road against every other team in the league.  For example, say Anaheim is the home rink I am evaluating, I will look at the average giveaways and takeaways against all of Anaheim’s opponent’s averaged on the road at every rink except Anaheim’s.  This difference will reflect consistency issues within Anaheim’s scorekeepers.  I adjusted the opponent’s giveaways and takeaways against first using the calculated multipliers.  The “opponent’s giveaways and takeaways against” adjustments at the bottom reflect this.  By doing this I can adjust all of the away turnovers for the away teams.  Then I input those results into the breakdown below, to adjust the turnovers documented for the home team.

Below on the left is a breakdown of the home team’s documented giveaways at home vs on the road.  In the first category San Jose averages 16.5 giveaways at home, and 6.4 giveaways on the road.(16.5-6.4=10.1 giveaway differential)  Given the fact that there are ~10 giveaways documented per game for each team in the league, this large differential reflects poor consistency on behalf of the scorekeepers.

As you can see San Jose have the highest giveaway differential, followed by the NY Islanders.  These high differentials indicate home rink bias, where the scorekeeper’s main focus is documenting events at a higher frequency for the home team as opposed to both teams.  A difference of zero, would indicate good consistency amongst scorekeepers, minimizing the effects of rink bias.  To adjust for this, I first set an acceptable variance of +2 to -2.  Anything -2.1 and below would be adjusted upwards to -2, and anything 2.1 and above would be adjusted down to 2.

Per my formula, San Jose’s giveaways at the home rink will be weighted at 50.8% to adjust for their home rink bias.(16.5*.508=8.4).  The 8.4 represents San Jose’s adjusted home giveaways average, which is now within the acceptable range of +2 to -2.   8.4-6.4=2, where 6.4 equals the average away giveaways for San Jose.  A multiplier of 1 indicates that the comparison of home vs away giveaway totals show no sign of rink bias, therefore any giveaway scored for that team in it’s home rink will be weighted 1 to 1.

Above on the right, I then compared each home team’s takeaways against totals to their takeaways against totals on the road. Using the same acceptable range of +2 to -2, the teams with negative differentials reflect a home rink bias of the scorekeepers understating their opponent’s defensive takeaways. Therefore my adjustment for these home teams will increase their “takeaways against”(opponent’s defensive takeaways) towards a net difference of -2.  For every takeaway against Los Angeles scorekeepers document, I will multiply it by 2.322 to adjust for the home rink bias.

Below on the left, I take a look at the home team’s opponent’s giveaway differentials. This breaks down the home team’s opponents giveaways at the home team’s rink vs their opponents giveaways when said home team is on the road. This indicates that the scorekeepers in Colorado, Florida, St. Louis, and Pittsburgh tend to ignore their opponents in regards to documenting giveaways. The negative differentials you see below on the left indicate that when those teams are on the road, their opponents are documented with more giveaways than when said team is at home. This is a clear indication of home rink bias. My multiplier is designed to increase a home team’s opponent’s giveaway totals.

Above on the right, I look at each home teams’ opponent’s takeaways against compared to their opponents takeaways against when said home team is on the road. To clarify, opponent’s takeaways against = Home teams’ defensive takeaways.  Calgary stand out as having the highest differential, meaning their scorekeepers document defensive takeaways(opponents takeaways against) for Calgary at a higher frequency than scorekeepers do when Calgary is on the road.  This indicates clear scorekeeping bias, and as you can see, I adjusted Calgary’s opponent’s takeaways against totals with a multiplier of .629, adjusting the totals down to the +2 acceptable home vs away differential range.

Per my data, these are the home rinks which are the most biased in regards to documenting turnovers: 1. NY Islanders, 2. Calgary, 3. San Jose, 4. Los Angeles, 5. Montreal, 6. Winnipeg, 7. Toronto, 8. Buffalo, 9. Florida, 10. Pittsburgh.   In total, their documented turnovers for the home team and their opponent deviate the furthest from the average differential of documented turnovers at home versus on the road.