I have to admit that I find it very hard to follow all of the great research that people people are producing every day on the web. It's hard enough just checking the two or three largest sites, but when you add in team blogs, other stathead blogs and everything else, it becomes impossible. This doesn't even include trying to locate recent research in baseball, basketball, football and soccer.

So we've decided to do something about it. Every weekday, the Stathead blog will summarize the best research-related studies, news, conferences, and resources for baseball, basketball, football, hockey and soccer.

The blog will primarily be edited by Neil Paine and will typically feature 20-30 links to analytic content around the internet. We've been working out the format over the last two weeks, and we think you'll soon appreciate our concise summaries for all of the articles we write about.

Feedback as always is welcome.

The Elo rating system is a method for calculating the relative skill levels of players in two-player games. The creator of the system, Arpad Elo, was a professor of physics at Marquette University who wanted an improved chess rating system. Although the system has its roots in chess, today it is used in many other games.

We decided to take Elo's work and apply it to the ranking of hockey players. Our player pool consists of all NHL players who meet at least one of the following career criteria:

• 200 goals (skater)
• 300 assists (skater)
• 150 wins (goalie)
• 3 NHL All-Star games (skater or goalie)
• 3 NHL All-Star teams (skater or goalie)

All players have an initial rating of 1500 points. These ratings are then updated by randomly selecting pairs of players and having them "play" each other.

We start by computing the win probabilities for each player (let's call them A and B):

P(A wins) = 1 / (1 + 10^((RB - RA) / 400)) P(B wins) = 1 / (1 + 10^((RA - RB) / 400))

where RA = rating for A RB = rating for B

After the winner has been determined, the ratings of the two players are adjusted. If A wins the match then the new ratings are:

RA_new = RA + K * P(B wins) RB_new = RB - K * P(B wins)

The "K" above stands for the "K-factor" and has a value of 10.

While if B wins the match then the new ratings are:

RA_new = RA - K * P(A wins)
RB_new = RB + K * P(A wins)

For example, suppose Wayne Gretzky ("WG") has a rating of 2500 and Mario Lemieux ("ML") has a rating of 2450. The win probabilities for each player are:

P(WG wins) = 1 / (1 + 10^((2450 - 2500) / 400)) = 0.571
P(ML wins) = 1 / (1 + 10^((2500 - 2450) / 400)) = 0.429

If WG wins then the new ratings are:

WG_new = 2500 + 10 * 0.429 = 2504
ML_new = 2450 - 10 * 0.429 = 2446

While if ML wins then the new ratings are:

WG_new = 2500 - 10 * 0.571 = 2494
ML_new = 2450 + 10 * 0.571 = 2456

Here are a few more notes about the way we have chosen to implement the Elo rating system:

• This is a community-based project with the goal of rating the best players in NHL history. In each matchup, the user should choose the player who they believe was the better player. It is up to the user to determine how much weight to give to offense versus defense, peak value versus career value, regular season versus playoffs, etc.
• Before opening this up to the public we simulated thousands of matchups in order to give the players more realistic starting ratings. These starting ratings do not necessarily represent the opinions of the owners of this site.
• Pairs are not chosen completely at random. The first player is randomly selected to begin the process. Following that, a second player with a rating within 250 points of the first player is randomly selected to complete the pair. We did this in order to prevent bizarre choices (e.g., Tiger Williams over Gordie Howe) from distorting the ratings.

As always, please send us some feedback if you have any comments or questions.

A Brief Explanation:

Individual players (goalies or skaters) can contribute to team success in either of two ways: by creating marginal goals on offense, or by preventing them on defense. Since there is a relationship between marginal goals and points in the standings, we calculate the marginal goals each individual contributes on offense/defense, and use the league's ratio of marginal goals per point to determine how many points a player's contribution was worth. The sum of a team's individual player point shares will roughly equal the team's point total in the standings.

The Full Explanation:

Calculating Point Shares | Hockey-Reference.com

Hey! Aren't there similar systems to this out there?

Yes, this is hardly the first all-in-one hockey stat to be featured on the internet. In fact, while it was inspired primarily by baseball's Bill James, Point Shares also drew on research done by hockey analysts Tom Awad, Iain Fyffe, and Alan Ryder, among others. But since there was no easy place to find those metrics for every season in NHL history, it was only logical for us to develop our own player value stat that would be easily accessible for hockey fans.

Where to Find Point Shares on the Site:

• Leaderboards: Single-Season, Career, Active, Year-by-Year, Progressive
• Player Pages: Stat tables, leaderboard appearances
• Team Pages: Skater and goalie tables
• Franchise Pages: Season and career leaders
• H-R Play Index: Player Season Finder (example)

If you're at a Super Bowl party this weekend and there's a game of squares set up, this app will tell you which squares have the best chance of winning based on game data from 1994-2010. Also, if you're new to the game, there's a page devoted to basic rules that will help you navigate your first SB squares experience. The app costs $0.99, but will pay for itself many times over if you win the pool.

For more information about its features, as well as links to posts about Super Bowl squares theory, click here.

• Show your support for your favorite player or team.
• Drum up traffic for your own site & draw in fans with a common interest.
• Get some well-deserved recognition for your support of BBR.
• Make your voice heard by the tens of thousands of people who visit Hockey-Reference every day.

Here's all you have to do to get involved:

1. Create a membership account.
2. Find the page(s) you'd like to support, and click "sponsor" (available pages).
3. If the page you want is already sponsored, click "Alert Me!" to be informed when the current sponsorship expires.
4. Follow the instructions to create your message and make your payment.
5. Your message and links will be visible on the page after we approve them (usually in less than 24 hours).

And who knows, if you're clever enough, your message might end up on lists like these.

User Feedback - Hockey-Reference.com

You can also email us directly at this address. We'll try to respond to everything within a week (although it may take longer this time of year, due to the holidays).

• Show your support for your favorite player or team.
• Drum up traffic for your own site & draw in fans with a common interest.
• Get some well-deserved recognition for your support of BBR.
• Make your voice heard by the tens of thousands of people who visit Hockey-Reference every day.

Here's all you have to do to get involved:

1. Create a membership account.
2. Find the page(s) you'd like to support, and click "sponsor" (available pages).
3. If the page you want is already sponsored, click "Alert Me!" to be informed when the current sponsorship expires.
4. Follow the instructions to create your message and make your payment.
5. Your message and links will be visible on the page after we approve them (usually in less than 24 hours).

And who knows, if you're clever enough, your message might end up on lists like these.

Also, once you do that, you probably will find yourself wanting to order the new Puck Prospectus book, available in hard copy or .PDF format. For my money, there's no better way to prepare for the upcoming season.