Following this little corner of the hockey universe has likely offered a small glimpse into the effects of schedules. In particular, back-to-back sets. In this post, I wanted to illustrate the impact of back to back sets, particularly the second game in the set, and that game’s influence in the standings.
Three scenarios exist for all back to back sets; I’ll refer to these as ‘B2B sets’ moving forward.
The first scenario has both teams playing a B2B set with the second game featuring two tired teams. This is actually a rare circumstance. During the period between lockouts, 894 out of 17,220 games (slightly over 5%) were played with both teams having dressed the previous night.
This is a direct contrast to the 2258 games played as a rested/tired combo (slightly over 13%). The other two scenarios feature one team is playing the back end of a B2B set designated as the tired team. The other team (opposition) did not play the previous night, designated as the rested team.
I’ve written reams about this in the past so I won’t go into too many details here, in an effort to push the conversation further instead of rehashing old ideas. A primer on this subject is found here, along with a detailed google doc with a team-by-team breakdown for the shortened 2013 shortened NHL season.
After a full introduction, it was only fitting to see the impact on the proposed realignment set to initiate in the 2013-14 season. The results are right here:
What kind of impact do these games have? How do they affect scheduling and standings? More than just a single season’s data is required to give perspective to final results.
If there’s an actual positive to emerge from two lockouts within seven seasons are the bookends to house data. Within those years are a separate and distinct set of circumstances that, as we shall see, contributed heavily to the overall effect of B2B sets.
Our story, however, begins over 30 years ago.
Let’s take a little trip.
Pre-lockout 2004
NHL expansion at the end of the 1970’s swallowed up the now defunct WHA beginning in 1979-80 and bloated the league to 21 teams with the addition of the Quebec Nordiques, Edmonton Oilers, Hartford Whalers and Winnipeg Jets.
**Note, my data starts from 1981-82 since it doesn’t make much of a difference in the grand scheme. The only difference of note took place beginning in the 1938-84 season, the implementation of overtime and we will touch on that.
I used a uniform calculation for win percentage throughout this entire project regardless of any fundamental changes in the way the NHL calculates points and the implementation of overtime, and further changes to OT in the late 1990’s. The calculation is below.
(Wins+(Tie/2))/GP -- OT games replace Tie in the case of post lockout. The (Tie/2) calculation is for the actual splitting of a point in the event of a tie. In the post lockout, technically, teams are splitting three points so the calculation could vary by changing that part to (OT/1.33) but for uniformity, I kept the calculation the same.
| Season | GP | B2B | AvTmSets | GP/Ssn | Teams | B2B Sched Percentage |
| 1981-82 | 1680 | 424 | 5.30 | 80 | 21 | 25.24 |
| 1982-83 | 1680 | 383 | 4.79 | 80 | 21 | 22.80 |
| 1983-84 | 1680 | 371 | 4.64 | 80 | 21 | 22.08 |
| 1984-85 | 1680 | 387 | 4.84 | 80 | 21 | 23.04 |
| 1985-86 | 1680 | 383 | 4.79 | 80 | 21 | 22.80 |
| 1986-87 | 1680 | 392 | 4.90 | 80 | 21 | 23.33 |
| 1987-88 | 1680 | 358 | 4.48 | 80 | 21 | 21.31 |
| 1988-89 | 1680 | 339 | 4.24 | 80 | 21 | 20.18 |
| 1989-90 | 1680 | 337 | 4.21 | 80 | 21 | 20.06 |
| 1990-91 | 1680 | 344 | 4.30 | 80 | 21 | 20.48 |
| 1991-92 | 1760 | 317 | 3.96 | 80 | 22 | 18.01 |
| 1992-93 | 2016 | 368 | 4.38 | 84 | 24 | 18.25 |
| 1993-94 | 2184 | 438 | 5.21 | 84 | 26 | 20.05 |
| 1995 | 1248 | 236 | 4.92 | 48 | 26 | 18.91 |
| 1995-96 | 2132 | 409 | 4.99 | 82 | 26 | 19.18 |
| 1996-97 | 2132 | 389 | 4.74 | 82 | 26 | 18.25 |
| 1997-98 | 2132 | 405 | 4.94 | 82 | 26 | 19.00 |
| 1998-99 | 2214 | 405 | 4.94 | 82 | 27 | 18.29 |
| 1999-00 | 2296 | 433 | 5.28 | 82 | 28 | 18.86 |
| 2000-01 | 2460 | 521 | 6.35 | 82 | 30 | 21.18 |
| 2001-02 | 2460 | 425 | 5.18 | 82 | 30 | 17.28 |
| 2002-03 | 2460 | 442 | 5.39 | 82 | 30 | 17.97 |
| 2003-04 | 2460 | 469 | 5.72 | 82 | 30 | 19.07 |
Of note, in 1981-82, over 25% of the NHL schedule involved back-to-back sets. That’s an NHL season high never attained again to present day. That amount reduced to an average of about 20% of the schedule by the time of the first lockout in 2004. That encompasses the league growing from 21 teams to 30 overall as it stands today through staggered expansion, a shortened season sparing a season-wide shutdown and further complication of divisional realignments.
An unlikely effect of expansion was new markets altering the NHL B2B dynamic and reducing overall B2B sets as a league on the whole. For example, teams could travel through LA and then down the road to Anaheim, or from Florida to Tampa Bay. Travel and scheduling changed as new teams were added, removed and moved.
Post lockout, between the bookends with 30 teams, the schedule never exceeded 350 sets, or more than 14% of the schedule, with the NHL average around 13.11 percent while dropping the per team average to just under four sets per team (3.93). Prior to the lockout, the standard deviation was 57.21 for the amount of B2B sets. Between bookends, the standard deviation almost doubled (98.78). Post-lockout the amounts of sets dropped almost an entire standard deviation from the pre lockout.
| Season | GP | B2B | AvTmSets | B2B Sched Percentage |
| 200506 | 2460 | 334 | 4.07 | 13.58 |
| 200607 | 2460 | 305 | 3.72 | 12.4 |
| 200708 | 2460 | 341 | 4.16 | 13.86 |
| 200809 | 2460 | 325 | 3.96 | 13.21 |
| 200910 | 2460 | 310 | 3.78 | 12.6 |
| 201011 | 2460 | 332 | 4.05 | 13.5 |
| 201112 | 2460 | 311 | 3.79 | 12.64 |
RESTED vs TIRED
The main basis of this study was to determine advantages of being a rested team over playing a tired team (and the disadvantage of being a tired team).
When the data is stretched out over a long period, even with different scoring systems and league makeup, the numbers are very similar.
Tired teams in the era pre-lockout era, between 1981 and 2004 posted an overall .419 win percentage with a record of 1966-2863-724. Even when breaking down that period into three where fundamental changes to the NHL point systems and overtime had an impact, there’s a singularity. Basically, playing as a tired team historically smoothed out to about .420 win percentage.
This is shown below.
| Season | W | L | T/OT | W% |
| 81-83 | 137 | 217 | 61 | 0.404 |
| 83-98 | 1205 | 1747 | 444 | 0.420 |
| 99-04 | 624 | 899 | 219 | 0.421 |
The designation 81-83 is only two seasons and not necessarily relevant in today’s discussions, but I wanted to demonstrate the uniformity.
In 1983-84, the NHL implemented the first five-minute overtime period, designed to give teams more time to win a game, awarding a full two points for a win, and zero (0) for a loss. In 1999 until 2004, as an aim to curb the lack of scoring plaguing the league, the NHL awarded a guaranteed point to teams tied at the end of 60 minutes, with an extra point on the line.
The design was meant to encourage scoring, yet has since been ironically dubbed ‘the loser point’. The shootout after the first lockout changed that. It also had an almost dramatic effect on the winning percentages for tired teams.
| Season | W | L | T/OT | W% |
| Pre-Lockout | 1966 | 2863 | 724 | 0.419 |
| Post Lockouts | 988 | 1009 | 261 | 0.495 |
Based on the data, it’s safe to assume that, historically, tired teams win about half of their games. It started from about .420 and grew to just under a .500 record aided by the implementation of the shootout. That was to be expected, as teams have the opportunity to salvage another point and bloat their win percentage with a guaranteed winner for each game.
Teams could now play for the overtime, and then hope for the shootout for an extra point at the end of the two-game stretch.
Teams posted .490 overtime win percentage prior to the 2004 lockout (187-203-390), which is close enough to the .485 post lockout (261 shootout games) to show the consistency from different eras. Regulation win percentage is .424 post-lockout.
Here are the post lockout numbers.
| Game Type | Games | Wins | Win% |
| Total GP | 17220 | 8607 | 0.558 |
| Non B2B GP | 14168 | 7222 | 0.569 |
| B2B both teams | 794 | 397 | 0.552 |
| Rested B2B | 2258 | 1270 | 0.596 |
| Tired B2B | 2258 | 988 | 0.495 |
Based on these, a season’s set of games as a tired team, expectations could be to win half of those games.
A rested team is likely to win about 60% of their games. When looked at in a vacuum, these can produce significant enough results to affect the standings, especially in this era of three-point games, and general NHL parity.
Expressed as a ratio, teams records could be evaluated on a strength of schedule by taking the tired/rested games.
For example, take the Detroit Red Wings. In a lockout shortened season, they had 11 games as a tired team. Based on history, they would be expected to get half of the total points, so 11 points in 11 games. They currently sport a 7-2-1 record, for 15 points, already exceeding expectations. At the time of this writing, that record is the saving grace for the Red Wings. Expectations of a .500 record would have them at 5-5, for 10 points and out of a playoff spot.
Winnipeg offers a great example of this scenario I tend to fall back on a lot.
Winnipeg with the rested advantage and Ottawa as a tired team are great 2011-12 examples.
Winnipeg iced a rested lineup versus a team playing their second game on consecutive nights an NHL high 17 games, sporting a nifty 13-3-0-1 record, for 27 points.
On December 1, the Jets sat in 13th in the East with 24 points. By New Years, and they moved up to 7th overall sporting a slippery record of 19-14-5.
Six games (5-0-1) in December featured a tired team on the back end of games on consecutive nights, with a record of 5-0-1. Five of those games featured a team flying into Winnipeg an average of 547 miles, including the first four occurrences, all wins.
In fact, the Jets did not lose their first of these 17 games until the end of January, posting a record of 10-0-2 along the way. The Jets were only 8 points out of a playoff spot by season’s end, surely aided by 27 points in these games.
Oh, and in games where Winnipeg was the tired team, they sported an eye-opening 1-10-1 record for three points
HOME vs ROAD
A significant amount of B2B sets involve some element of travel. In the West, teams that travel through California will take one the Kings and Ducks, or the Oilers and Flames. In the East, traveling teams may pass through both New York teams, or New Jersey as an alternative. Florida and Tampa Bay take turns playing teams that played the night before as they travel through Florida. Buffalo and Toronto are only one hour away off the 401 highway.
Most B2B games also occur on the road. In the pre-lockout era, 4,336 games were played on the road, 1,217 at home featuring a combination of rested/tired teams. Post lockout, the numbers are similar (525 home games, 1733 road games).
The breakdown from pre-lockout to post is evident here. The change has been dramatic.
| Home/Road split 81-04 | |||||
| Tired | Hm | 554 | 506 | 157 | 0.520 |
| Rd | 1412 | 2357 | 567 | 0.391 | |
| Rested | Hm | 506 | 554 | 157 | 0.480 |
| Rd | 2357 | 1412 | 567 | 0.609 | |
After the lockout, road games as a tired team saw a modest boost in win percentage while home rested teams eclipsed the .500 mark.
| Hm/Rd '05-12 | ||
| Tired | Rested | |
| Hm | 0.555 | 0.554 |
| Rd | 0.477 | 0.636 |
In both categories, a rested team on the road was clearly an advantage with a .636 win percentage. The logic behind that would be that most likely in these situations, the actual home team is likely to have played on the road the previous night, and returned back home, with their opponents already in town and waiting for them.
Conclusions
Half. It’s not just what a former spouse receives upon a divorce anymore, it’s also the expectation of winning games as a tired team on the second night of a back-to-back set against a team that has not played the previous night.
Every season when assessing the strength of schedule, breaking down the games where teams play tired can now come with the historical expectation that they should win about half their games.
Rested teams have been historically winning games at a 6 out of 10 clip, offering a distinct advantage. Further distinctions can be isolated, such as home and road splits. For instance, a team that is tired is likely to win more games at home than on the road and rested teams on the road are more likely win than the home team.
It’s like a paradox in this instance, where the home tired team may win more games, but a rested road team has the best win percentage of the lot. The amount of home games as a tired team plays a bigger part in the winning percentage, rather than just the number alone.
When it comes to determining strength of schedule, keep the amount of games as a tired team in mind, especially down the stretch where every point changes the standings on a nightly basis due to the perceived parity of today’s NHL.
Data is courtesy NHL.com and shrpsports.com
Follow the McKeen's team on Twitter:
@KatsHockey
@mckeenshockey