These NHL power rankings are based on two factors that are estimates of a team’s true talent level, both adjusted for strength of schedule: Pythagenpat win percentage and percentage of even-strength shots in their favor when the game is within one goal (“Fenwick Close“).

Pythagenpat estimates a team’s true talent level based on goals for and against, while a team’s Fenwick Close gives us a proxy for their ability to drive puck possession. The more shots in their favor after eliminating score effects, the more sustainable winning should be. For a complete explanation of the methodology, see the end of this post.

Objective rankings like these will have some volatility early on because of small sample sizes, but will stabilize as the season plays on.

Here are this week’s power ratings. Agree or disagree? Let Neil know on twitter @ngreenberg.

1. Pittsburgh Penguins (3-1-0), 0.724 AdjWin%, 0.605 AdjFenClose%

The Penguins made sure no team was still unbeaten after dispatching the New York Islanders, 3-1, on Saturday night. Ignoring special teams and lead-protecting situations, the Pens are taking 60.5 percent of the shots after adjusting for strength of opponent.

2. Tampa Bay Lightning (3-1-1), 0.704 AdjWin%, 0.57 AdjFenClose%

Tampa Bay scorer extraordinaire Steven Stamkos is back from injury, scoring five goals in his first five games. And if the Bolts aren’t playing well enough, they just recalled forward Jonathan Drouin, the third pick in the 2013 NHL draft and favorite on many rookie of the year lists.

3. St. Louis Blues (2-2-1), 0.638 AdjWin%, 0.577 AdjFenClose%

4. Anaheim Ducks (5-1-0), 0.725 AdjWin%, 0.54 AdjFenClose%

5. Minnesota Wild (2-2-0), 0.748 AdjWin%, 0.519 AdjFenClose%

6. New York Islanders (4-1-0), 0.606 AdjWin%, 0.558 AdjFenClose%

7. Chicago Blackhawks (3-0-1), 0.656 AdjWin%, 0.539 AdjFenClose%

8. Washington Capitals (3-0-2), 0.677 AdjWin%, 0.527 AdjFenClose%

Alex Ovechkin is scoring (five goals in five games), the defensive depth is now a team strength and the penalty kill has only allowed two goals on 20 power plays against.

9. Dallas Stars (2-1-2), 0.555 AdjWin%, 0.562 AdjFenClose%

10. New York Rangers (3-3-0), 0.461 AdjWin%, 0.558 AdjFenClose%

11. Nashville Predators (3-0-2), 0.663 AdjWin%, 0.488 AdjFenClose%

12. Vancouver Canucks (3-1-0), 0.514 AdjWin%, 0.531 AdjFenClose%

13. Detroit Red Wings (3-1-1), 0.595 AdjWin%, 0.502 AdjFenClose%

14. New Jersey Devils (3-2-0), 0.559 AdjWin%, 0.514 AdjFenClose%

15. Montreal Canadiens (5-1-0), 0.48 AdjWin%, 0.516 AdjFenClose%

16. Los Angeles Kings (4-1-1), 0.622 AdjWin%, 0.462 AdjFenClose%

17. Boston Bruins (3-4-0), 0.412 AdjWin%, 0.518 AdjFenClose%

18. Columbus Blue Jackets (3-2-0), 0.534 AdjWin%, 0.466 AdjFenClose%

19. Philadelphia Flyers (1-2-2), 0.429 AdjWin%, 0.495 AdjFenClose%

20. Toronto Maple Leafs (2-3-1), 0.416 AdjWin%, 0.498 AdjFenClose%

21. San Jose Sharks (4-1-1), 0.647 AdjWin%, 0.415 AdjFenClose%

The Sharks are winning, but they are not tilting the ice in their favor. Just 41.2 percent of even-strength shot attempts have gone in their favor, fourth-lowest percentage in the league. If they don’t turn that around, their winning ways won’t last.

22. Ottawa Senators (4-1-0), 0.612 AdjWin%, 0.421 AdjFenClose%

23. Arizona Coyotes (2-2-0), 0.296 AdjWin%, 0.507 AdjFenClose%

24. Florida Panthers (1-2-2), 0.284 AdjWin%, 0.495 AdjFenClose%

25. Calgary Flames (4-3-0), 0.56 AdjWin%, 0.396 AdjFenClose%

26. Carolina Hurricanes (0-2-2), 0.302 AdjWin%, 0.48 AdjFenClose%

27. Edmonton Oilers (0-4-1), 0.226 AdjWin%, 0.487 AdjFenClose%

28. Colorado Avalanche (1-4-1), 0.295 AdjWin%, 0.451 AdjFenClose%

29. Winnipeg Jets (1-4-0), 0.328 AdjWin%, 0.386 AdjFenClose%

30. Buffalo Sabres (1-5-0), 0.169 AdjWin%, 0.387 AdjFenClose%

Last season, the Sabres could not sustain any puck possession and were outshot, 2,154-1,643, at even strength. This season, things have started off even worse.

NHL Power Ranking Methodology

The methodology is simple and adapted from a guest post at Football Perspective.

It starts with an estimate of a team’s true talent to win games: their Pythagenpat record, a modification of Bill James’s Pythagorean win formula where the exponent used is derived from the individual team’s goal scoring environment — the more goals per game, the higher the exponent.

Let’s take the Washington Capitals as an example. They have scored 18 goals and allowed 11 in five games, so the best estimate for their true talent is a two-step process:

Step 1: Find the exponent to use in Pythagenpat by the following formula:

((GF + GA)/GP)^.285 = ((18+11)/5)^.285 = 1.7

Step 2: Use that exponent in the Pythagorean win formula:

Win Percentage = [(GF)^1.7]/[(GA)^1.7 + (GA)^1.7] = 0.693

Then, we find the average Pythagenpat of their opponents. The Capitals have played Montreal (0.480 adjusted win percentage), Boston (0.412), San Jose (0.647), New Jersey (0.559) and Florida (0.284), for an average of 0.476.

From there we convert these to win/loss ratios based on a full season. Washington’s win ratio is:

WinRatio_WSH = Wins (.693 * 82) over Losses [(1 – .693) * 82] = 56.8/25.2 = 2.3

The win/loss ratio for their opponents is:

WinRatio_WSHOpp = Wins (.476 * 82) over Losses [(1 – .476) * 82] = 39.0/42.9 = .91.

After that, we find the natural logarithm of the product of the two win ratios (designated here by x).

x = ln(2.3 * .91) = .739

Still with me? All that is left is to run a logistic function to find the win percentage which is adjusted for strength of schedule.

Logistic Function: 1/(1 + e^-x) = 1/(1 + e^-.739) = .677

This means that Washington is really a 0.677 win-percentage team that has the “true talent” to win 0.693 of their games against opponents who have won 0.476. Repeat this process for each team, each week, and then again for shot percentage when the game is close (Fenwick Close) and you get the power rankings above.