NHL 2018/19 Team Rankings

Since you 'had to make a decision' then the decision becomes non-mathematical in the strictest sense.

But a bigger K creating a bigger spread changes the odds and therefore the effect of the result of the next game, does it not? If I am missing the math, I would love to see, with rigor, how it makes no difference. And, I say that as a math geek, with a slight inclination in the back of my mind how that math would come out.

If this is subjective, then the 'guess' as the correct value of OT/SO wins and losses is also, as well as the value of K. And, as described below, there are mathematical ways to determine the value of home ice as a league average.

It is a cold mathematical process, but it can be skewed by the following. Let's say there are 32 teams, instead of 31, so we don't offend anyone, and let's say that the last team in (International Falls, MN) is globally acknowledged to be the worst team in the league. Now, let them win their first 3 games, 1 in OT, because their goalie is particularly hot. Everyone knows that this is a mirage. It won't last. But, their 4th opponent, who beats them, gets major points for this win. And, after they lose 9 more games out of 10, the next opponent, who beats them, gets minimal points. That seems mathematically correct, but not completely sensical to me.

Even though I know that all teams have injuries, and so one team's power level is not the same all year long, so teams do have scheduling advantages of playing some one "at the right time", I feel this system unduly rewards teams with scheduling luck to play weak teams with fast starts.

That could well be your preference, and I wouldn't argue with your preference, but it is really a preference.

This article (pardon the few typos. I wrote it as a post, and the mods moved it. Had I known it would be an article, I would have proof-read it better) gives the basics. Should you be interested in more mathematical rigor, or how I arranged the spread-sheet to accomplish this, please PM me.

And, it should be noticed that those rankings recalculate from scratch after every game.

 

The size of K doesn't matter as long as it's constant. A team can earn 3 points on K of 10 and 30 points on K of 100. But this spread applies to every team and every game, so the ranking doesn't change. It is more about the expected results, K just translates it into a number of points.

The scores I chose to apply for OTSO win and loss are subjective. I had discussions last year with people about what to do with OTSO, I knew this will always be the biggest source to controversy, is it that a win is a win is a win. Or is is OTSO different. I made the decision it is half a win. (A tie in Elo which doesn't exist in the NHL is 0.5, a win is 1, so half a win is 0.75 and half a loss is 0.25)

I don't really want to get into home/away games and how they sway expected scores. Too subjective for me, I'll get enough questions about OTSO.

You can come up with scenarios, but it balances out. If a team is hot and then turns cold the and stay cold then by the end they will end up where they belong. You might not agree that beating a bad 3-0 team is worth as much as beating a 3-0 good team, but in a basic outlook it is the same.

In the end there is so much time I can spend on this.

 

Thanks for the discussion. I'm going to run some trials with K later today just to convince myself of what you say. Am I remembering right, that the pre-game odds depend on only the Elo spread? Just so I have it right.

That's the real answer, isn't it, that there just isn't enough time? Modeling home/away into Elo becomes very difficult, because you really need some number by which to multiply the home teams Elo ranking before you create the pre-game odds. And, how to come up with that? It's very very difficult to do it with some real reasoning behind it.

As for OT/SO, morehockeystats treats an OT/SO win as a full win. And, an OT/SO loss as half a win. Which simply has the effect that the mean/median Elo Ranking of all teams creeps up through the year.

For my calculation, I need only the following things at any point in the year:
1- How many games each team has played both home and away with all the other teams.
2- How many total wins each team has
3- How many home there are in the whole league (I think I'm right about the last one).

But I run into the same problem with the OT/SO issue. At first, I went 2/3, 1/3. But then someone said that the results should use the same method as the standings. And, since I only need the total number of wins for each team, I take their points, and divide by the league-wide average number of points given per game (which is approx 2.25 and has been for years). So, that gets me out of the bind.

Anyway, thanks again. Take a look at my Bradley-Terry article and get back to me if you want. This has been good motivation for me to go take another look at the math involved in that.

 

Great work ! This is a really interesting way to look at the rankings.

 

Updated rankings for Oct 24 2018.

Latest changes:
[2018-10-23, Arizona Coyotes (4): 2406 -> 2467, Columbus Blue Jackets (1): 2485 -> 2424]
[2018-10-23, Anaheim Ducks (1): 2460 -> 2410, Chicago Blackhawks (3): 2457 -> 2507]
[2018-10-23, Los Angeles Kings (2): 2340 -> 2303, Dallas Stars (4): 2436 -> 2473]
[2018-10-23, Pittsburgh Penguins (6): 2541 -> 2561, Edmonton Oilers (5): 2508 -> 2488, OTSO]
[2018-10-23, Calgary Flames (2): 2589 -> 2532, Montreal Canadiens (3): 2542 -> 2599]
[2018-10-23, San Jose Sharks (5): 2529 -> 2605, Nashville Predators (4): 2727 -> 2651]
[2018-10-23, Florida Panthers (2): 2375 -> 2324, New York Rangers (5): 2365 -> 2416]
[2018-10-23, Boston Bruins (4): 2498 -> 2559, Ottawa Senators (1): 2579 -> 2518]

 

I’d be very interested to see last year’s and this year’s in a graph format.

The changes over time interest me the most

 

Currently I don't store these scores anywhere so I'm not able to do trend analysis. Basically, I only have the latest scores.

I'm thinking about setting up a website at some point that shows overtime trends of how teams go up and down, can track previous seasons as well.

Then I would run the program against other leagues as well: NBA, NFL, MLB, even MLS.

I will let you guys know if I ever do that. But I'm too busy for that now.

 

4 of the top 7 in the Central.

This division's gonna be a bloodbath.

 

TB looking very good again this year. 6-1-1 start.

 

Updated rankings for Oct 25 2018

Latest changes
[2018-10-23, Arizona Coyotes (4): 2406 -> 2467, Columbus Blue Jackets (1): 2485 -> 2424]
[2018-10-23, Anaheim Ducks (1): 2460 -> 2410, Chicago Blackhawks (3): 2457 -> 2507]
[2018-10-23, Los Angeles Kings (2): 2340 -> 2303, Dallas Stars (4): 2436 -> 2473]
[2018-10-23, Pittsburgh Penguins (6): 2541 -> 2561, Edmonton Oilers (5): 2508 -> 2488, OTSO]
[2018-10-23, Calgary Flames (2): 2589 -> 2532, Montreal Canadiens (3): 2542 -> 2599]
[2018-10-23, San Jose Sharks (5): 2529 -> 2605, Nashville Predators (4): 2727 -> 2651]
[2018-10-23, Florida Panthers (2): 2375 -> 2324, New York Rangers (5): 2365 -> 2416]
[2018-10-23, Boston Bruins (4): 2498 -> 2559, Ottawa Senators (1): 2579 -> 2518]
[2018-10-24, Tampa Bay Lightning (1): 2607 -> 2662, Colorado Avalanche (0): 2641 -> 2586]
[2018-10-24, Florida Panthers (3): 2324 -> 2365, New York Islanders (2): 2441 -> 2400, OTSO]
[2018-10-24, Vancouver Canucks (3): 2473 -> 2509, Vegas Golden Knights (2): 2551 -> 2515, OTSO]
[2018-10-24, Toronto Maple Leafs (4): 2511 -> 2576, Winnipeg Jets (2): 2616 -> 2551]

 

Jumped to first place after last night's win against Colorado.

 

Updated rankings for Oct 26 2018

Latest changes

[2018-10-25, Vancouver Canucks (1): 2509 -> 2453, Arizona Coyotes (4): 2467 -> 2523]
[2018-10-25, Philadelphia Flyers (0): 2448 -> 2413, Boston Bruins (3): 2559 -> 2594]
[2018-10-25, Montreal Canadiens (3): 2599 -> 2537, Buffalo Sabres (4): 2517 -> 2579]
[2018-10-25, Pittsburgh Penguins (9): 2561 -> 2607, Calgary Flames (1): 2532 -> 2486]
[2018-10-25, New York Rangers (1): 2416 -> 2379, Chicago Blackhawks (4): 2507 -> 2544]
[2018-10-25, Anaheim Ducks (2): 2410 -> 2369, Dallas Stars (5): 2473 -> 2514]
[2018-10-25, Washington Capitals (1): 2507 -> 2454, Edmonton Oilers (4): 2488 -> 2541]
[2018-10-25, Los Angeles Kings (1): 2303 -> 2285, Minnesota Wild (4): 2563 -> 2581]
[2018-10-25, Nashville Predators (4): 2651 -> 2658, New Jersey Devils (3): 2518 -> 2511, OTSO]
[2018-10-25, Columbus Blue Jackets (7): 2424 -> 2479, St. Louis Blues (4): 2462 -> 2407]

 

Updated rankings for Oct 27 2018

Latest changes
[2018-10-26, San Jose Sharks (3): 2605 -> 2564, Carolina Hurricanes (4): 2489 -> 2530, OTSO]
[2018-10-26, Ottawa Senators (3): 2518 -> 2478, Colorado Avalanche (6): 2586 -> 2626]
[2018-10-26, Winnipeg Jets (2): 2551 -> 2572, Detroit Red Wings (1): 2319 -> 2298]
[2018-10-26, Tampa Bay Lightning (3): 2662 -> 2692, Vegas Golden Knights (2): 2515 -> 2485]

 

Update rankings for Oct 30 2018

Latest changes
[2018-10-27, Tampa Bay Lightning (1): 2692 -> 2619, Arizona Coyotes (7): 2523 -> 2596]
[2018-10-27, Montreal Canadiens (3): 2537 -> 2595, Boston Bruins (0): 2594 -> 2536]
[2018-10-27, Buffalo Sabres (4): 2579 -> 2540, Columbus Blue Jackets (5): 2479 -> 2518, OTSO]
[2018-10-27, Washington Capitals (4): 2454 -> 2484, Calgary Flames (3): 2486 -> 2456, OTSO]
[2018-10-27, Colorado Avalanche (2): 2626 -> 2570, Minnesota Wild (3): 2581 -> 2637]
[2018-10-27, Florida Panthers (2): 2365 -> 2335, New Jersey Devils (3): 2511 -> 2541]
[2018-10-27, Edmonton Oilers (5): 2541 -> 2607, Nashville Predators (3): 2658 -> 2592]
[2018-10-27, New York Islanders (6): 2400 -> 2452, Philadelphia Flyers (1): 2413 -> 2361]
[2018-10-27, Chicago Blackhawks (3): 2544 -> 2475, St. Louis Blues (7): 2407 -> 2476]
[2018-10-27, Winnipeg Jets (2): 2572 -> 2523, Toronto Maple Leafs (3): 2576 -> 2625]
[2018-10-27, Pittsburgh Penguins (5): 2607 -> 2636, Vancouver Canucks (0): 2453 -> 2424]
[2018-10-28, San Jose Sharks (4): 2564 -> 2564, Anaheim Ducks (3): 2369 -> 2369, OTSO]
[2018-10-28, New York Islanders (2): 2452 -> 2513, Carolina Hurricanes (1): 2530 -> 2469]
[2018-10-28, Edmonton Oilers (2): 2607 -> 2614, Chicago Blackhawks (1): 2475 -> 2468, OTSO]
[2018-10-28, Dallas Stars (2): 2514 -> 2436, Detroit Red Wings (4): 2298 -> 2376]
[2018-10-28, New York Rangers (3): 2379 -> 2316, Los Angeles Kings (4): 2285 -> 2348]
[2018-10-28, Ottawa Senators (3): 2478 -> 2454, Vegas Golden Knights (4): 2485 -> 2509, OTSO]
[2018-10-29, Calgary Flames (3): 2456 -> 2529, Toronto Maple Leafs (1): 2625 -> 2552]
[2018-10-29, Minnesota Wild (2): 2637 -> 2560, Vancouver Canucks (5): 2424 -> 2501]

 

Updated rankings for Oct 31 2018

Latest changes
[2018-10-30, Philadelphia Flyers (3): 2361 -> 2412, Anaheim Ducks (2): 2369 -> 2318]
[2018-10-30, Ottawa Senators (1): 2454 -> 2423, Arizona Coyotes (5): 2596 -> 2627]
[2018-10-30, Calgary Flames (2): 2529 -> 2556, Buffalo Sabres (1): 2540 -> 2513, OTSO]
[2018-10-30, Boston Bruins (3): 2536 -> 2576, Carolina Hurricanes (2): 2469 -> 2429]
[2018-10-30, Detroit Red Wings (5): 2376 -> 2445, Columbus Blue Jackets (3): 2518 -> 2449]
[2018-10-30, Minnesota Wild (4): 2560 -> 2618, Edmonton Oilers (3): 2614 -> 2556]
[2018-10-30, Dallas Stars (4): 2436 -> 2507, Montreal Canadiens (1): 2595 -> 2524]
[2018-10-30, Vegas Golden Knights (1): 2509 -> 2471, Nashville Predators (4): 2592 -> 2630]
[2018-10-30, New York Islanders (6): 2513 -> 2580, Pittsburgh Penguins (3): 2636 -> 2569]
[2018-10-30, New York Rangers (4): 2316 -> 2372, San Jose Sharks (3): 2564 -> 2508, OTSO]
[2018-10-30, New Jersey Devils (3): 2541 -> 2502, Tampa Bay Lightning (8): 2619 -> 2658]

 

Updated rankings for Nov 01 2018

Latest changes
[2018-10-31, Chicago Blackhawks (2): 2468 -> 2423, Vancouver Canucks (4): 2501 -> 2546]


I didn't notice there was only one game yesterday. Hardly worth updating the rankings for

 

Central looks great. Pacific and metropolitan look like garbage.

 

Updated rankings for Nov 02 2018

Latest Changes
[2018-11-01, New York Rangers (3): 2372 -> 2389, Anaheim Ducks (2): 2318 -> 2301, OTSO]
[2018-11-01, Colorado Avalanche (5): 2570 -> 2518, Calgary Flames (6): 2556 -> 2608]
[2018-11-01, New Jersey Devils (3): 2502 -> 2444, Detroit Red Wings (4): 2445 -> 2503]
[2018-11-01, Chicago Blackhawks (0): 2423 -> 2391, Edmonton Oilers (4): 2556 -> 2588]
[2018-11-01, Winnipeg Jets (4): 2523 -> 2548, Florida Panthers (2): 2335 -> 2310]
[2018-11-01, Philadelphia Flyers (5): 2412 -> 2453, Los Angeles Kings (2): 2348 -> 2307]
[2018-11-01, Washington Capitals (4): 2484 -> 2440, Montreal Canadiens (6): 2524 -> 2568]
[2018-11-01, Pittsburgh Penguins (2): 2569 -> 2546, New York Islanders (3): 2580 -> 2603, OTSO]
[2018-11-01, Buffalo Sabres (2): 2513 -> 2450, Ottawa Senators (4): 2423 -> 2486]
[2018-11-01, Columbus Blue Jackets (4): 2449 -> 2507, San Jose Sharks (1): 2508 -> 2450]
[2018-11-01, Vegas Golden Knights (3): 2471 -> 2422, St. Louis Blues (5): 2476 -> 2525]
[2018-11-01, Nashville Predators (4): 2630 -> 2684, Tampa Bay Lightning (1): 2658 -> 2604]
[2018-11-01, Dallas Stars (2): 2507 -> 2563, Toronto Maple Leafs (1): 2552 -> 2496]

 

What the **** is this garbage?

 

Updated rankings for Nov 05 2018

Latest Changes
[2018-11-02, Carolina Hurricanes (3): 2429 -> 2430, Arizona Coyotes (4): 2627 -> 2626, OTSO]
[2018-11-02, Colorado Avalanche (6): 2518 -> 2497, Vancouver Canucks (7): 2546 -> 2567, OTSO]
[2018-11-02, Florida Panthers (4): 2310 -> 2390, Winnipeg Jets (2): 2548 -> 2468]
[2018-11-03, Ottawa Senators (2): 2486 -> 2431, Buffalo Sabres (9): 2450 -> 2505]
[2018-11-03, Chicago Blackhawks (3): 2391 -> 2369, Calgary Flames (5): 2608 -> 2630]
[2018-11-03, Edmonton Oilers (4): 2588 -> 2626, Detroit Red Wings (3): 2503 -> 2465]
[2018-11-03, Columbus Blue Jackets (1): 2507 -> 2431, Los Angeles Kings (4): 2307 -> 2383]
[2018-11-03, Tampa Bay Lightning (4): 2604 -> 2649, Montreal Canadiens (1): 2568 -> 2523]
[2018-11-03, Boston Bruins (0): 2576 -> 2541, Nashville Predators (1): 2684 -> 2719]
[2018-11-03, New Jersey Devils (0): 2444 -> 2415, New York Islanders (3): 2603 -> 2632]
[2018-11-03, Toronto Maple Leafs (5): 2496 -> 2553, Pittsburgh Penguins (0): 2546 -> 2489]
[2018-11-03, Philadelphia Flyers (3): 2453 -> 2428, San Jose Sharks (4): 2450 -> 2475, OTSO]
[2018-11-03, Minnesota Wild (5): 2618 -> 2655, St. Louis Blues (1): 2525 -> 2488]
[2018-11-03, Carolina Hurricanes (0): 2430 -> 2379, Vegas Golden Knights (3): 2422 -> 2473]
[2018-11-03, Dallas Stars (4): 2563 -> 2571, Washington Capitals (3): 2440 -> 2432, OTSO]
[2018-11-04, Columbus Blue Jackets (2): 2431 -> 2388, Anaheim Ducks (3): 2301 -> 2344, OTSO]
[2018-11-04, Buffalo Sabres (1): 2505 -> 2439, New York Rangers (3): 2389 -> 2455]
[2018-11-04, Tampa Bay Lightning (4): 2649 -> 2646, Ottawa Senators (3): 2431 -> 2434, OTSO]

 

Somebody sounds sour.

 

Lol that's awesome. Tampa won and went down 3 points because of OT against the Sens. And I don't even disagree.

 

The central owns the top two spots... surprising no one.

 

I love this thread, one of my favs each year. So far there are some surprises, the teams in red I expect to drop and the teams in green I'd think will rise by mid year with a larger sample size.

 

Surprised the Oilers aren't higher with their record and wins against Boston, Washington, Nashville, Winnipeg.