3rd March 2015
One of the more common subjects for queries we receive at Sports-Reference is our SRS (Simple Rating System) figures. For some background, the first of our sites to add SRS was Pro-Football-Reference, when Doug Drinen added it to the site in 2006 and provided this excellent primer. The important thing to know is that SRS is a rating that takes into account average point differential and strength of schedule. For instance, the 2006-07 Spurs won games by an average of 8.43 points per game and played a schedule with opponents that were 0.08 points worse than average, giving them an SRS of 8.35. This means they were 8.35 points better than an average team. An average team would have an SRS of 0.0. The calculation can be complicated, but the premise is simple and it produces easily interpreted results.
However, there are some variations in the way we calculate SRS across our various sites. We'll break down these differences below.
Pro-Football-Reference.com SRS: PFR's SRS is unique in that a home-field advantage is included as a part of the calculation because of the short schedule compared to the other sports (we don't want a team to look relatively weak at the halfway point because they've only played 3 of their first 8 at home, for instance). This HFA fluctuates yearly based on game results, but it is generally somewhere between 2 and 3 points (2006 being an outlier, as you'll see). Below is a look at the HFA numbers we have used since 2001. If you'd like to calculate these HFAs yourself, just sum up every team's home point differential and then divide by the total number of games played across the league that season. This data can easily be found in the Play Index for each season:
- 2001: 2.0081
- 2002: 2.2461
- 2003: 3.5547
- 2004: 2.5078
- 2005: 3.6484
- 2006: 0.8477
- 2007: 2.8672
- 2008: 2.5586
- 2009: 2.2070
- 2010: 1.8945
- 2011: 3.2656
- 2012: 2.4336
- 2013: 3.1055
- 2014: 2.4883
College Football SRS: Our CFB SRS does not contain a home-field advantage element, but it does have some other quirks. Most importantly, we have capped the margin of victory considered for the formula. Due to the number of mismatches seen in college football, the maximum point differential a team can be credited with in a game is 24. We also credit all wins as a minimum of plus-7 margin of victory (so if you win by 1 point, it's treated the same as a 7-point win). The same logic is applied to losses, as well. One other wrinkle for CFB is that all non-major opponents are included as one team for the sake of the ratings.
College Basketball SRS: SRS for college hoops is straight forward (no HFA & no adjusted MOV), but one item to note is that games against non-major opponents are not counted in our calculations.
MLB, NBA & NHL: All of these SRS calculations are straight forward with no adjustments for HFA and no capping of MOV. It should be noted, however, that no special consideration is given for extra-innings, overtimes or shootouts, either.
We'll close with a quick rundown of the various merits and weaknesses of SRS, from Drinen's original 2006 post. These bullet points were created to describe the system used for NFL SRS, but many of the strengths and weaknesses can applied to the other sports, as well:
- The numbers it spits out are easy to interpret - if Team A's rating is 3 bigger than Team B's, this means that the system thinks Team A is 3 points better than Team B. With most ranking algorithms, the numbers that come out have no real meaning that can be translated into an English sentence. With this system, the units are easy to understand.
- It is a predictive system rather than a retrodictive system - this is a very important distinction. You can use these ratings to answer the question: which team is stronger? I.e. which team is more likely to win a game tomorrow? Or you can use them to answer the question: which of these teams accomplished more in the past? Some systems answer the first questions more accurately; they are called predictive systems. Others answer the latter question more accurately; they are called retrodictive systems. As it turns out, this is a pretty good predictive system. For the reasons described below, it is not a good retrodictive system.
- It weights all games equally - every football fan knows that the Colts' week 17 game against Arizona was a meaningless exhibition, but the algorithm gives it the same weight as all the rest of the games.
- It weights all points equally, and therefore ignores wins and losses - take a look at the Colts season. If you take away 10 points in week 3 and give them back 10 points in week 4, you've just changed their record, but you haven't changed their rating at all. If you take away 10 points in week 3 and give back 20 points in week 4, you have made their record worse but their rating better. Most football fans put a high premium on the few points that move you from a 3-point loss to a 3-point win and almost no weight on the many points that move you from a 20-point win to a 50-point win.
- It is easily impressed by blowout victories - this system thinks a 50-point win and a 10-point loss is preferable to two 14-point wins. Most fans would disagree with that assessment.
- It is slightly biased toward offensive-minded teams - because it considers point margins instead of point ratios, it treats a 50-30 win as more impressive than a 17-0 win. Again, this is an assessment that most fans would disagree with.
- This should go without saying, but - I'll say it anyway. The system does not take into account injuries, weather conditions, yardage gained, the importance of the game, whether it was a Monday Night game or not, whether the quarterback's grandmother was sick, or anything else besides points scored and points allowed.
Posted in Announcement, Baseball-Reference.com, Basketball-Reference.com, CBB at Sports Reference, CFB at Sports Reference, Data, FAQ, Features, Hockey-Reference.com, Pro-Football-Reference.com, SRS, Stat Questions, Statgeekery, Uncategorized | 2 Comments »
3rd March 2015
We have added a process to our database that correctly determines NHL Rookie status, for the purposes of using the Player Finder tools in the Play Index. For example, see the Player Game Finder, and Player Season Finder.
Now you can use those tools to do a search for true rookies, in addition to the variety of other data points you can search on.
Hockey-Reference continues to make improvements to these tools. We welcome your feedback.
Posted in Announcement, Hockey-Reference.com, Play Index | Comments Off on Rookie Designation Added
27th February 2015
Posted in Announcement, Baseball-Reference.com, Basketball-Reference.com, CBB at Sports Reference, CFB at Sports Reference, expire21d, Hockey-Reference.com, Olympics at S-R, Pro-Football-Reference.com, Statgeekery | Comments Off on Hiring: Analytics Consultant, Posted Feb 27, 2015
9th February 2015
Adam Wodon has joined Sports Reference today as a Managing Director for Hockey Reference. Adam will be working out of our Philadelphia office as our staff size has now risen to six. Adam brings a great deal of development experience and hockey knowledge to Sports Reference. Adam is the founder and managing editor of College Hockey News. Adam is an Isles fan through and through and also supports the Mets, Jets, and Nets making him the first diehard National League fan on staff. Adam is also on twitter at (@chn_AdamWodon).
Hans VanSlooten (@CantPitch) who had been working on the hockey site for the last 14 months will be taking over primary day-to-day development of Baseball-Reference.com.
Posted in Announcement, Baseball-Reference.com, Basketball-Reference.com, CBB at Sports Reference, CFB at Sports Reference, Expire30d, Hockey-Reference.com, Olympics at S-R, Pro-Football-Reference.com | 2 Comments »
9th December 2014
We've added four new measures to goalies on Hockey-Reference.com, Games Started, Quality Starts, Quality Start Percentage, Really Bad Starts. These should be available anywhere you see goalie stats on the site currently, including the Play Indexes.
Games Started (GS) should be fairly obvious, this is the number of games that a goalie started (i.e. did not come in for another pulled goalie).
Quality Starts and Quality Start Percentage were developed by Rob Vollman, and are described in his Hockey Abstract. Here is a brief description, but we encourage you to pick up a copy of his book for additional discussion of this statistic (and many others):
A Quality Start (QS) is when the goalie achieves at least the mean save percentage (for the season) in a game. For the 2013-2014 season that percentage is 91.5%. So, if Tuukka Rask allows only 2 goals on 28 shots (a 92.9% save percentage), that is considered a Quality Start. There is an additional criteria for low shooting games: if a goalie faces 20 or fewer shots, he only needs to get an 88.5% save percentage. The relationship between save percentage and winning percentage (and hence the definition of Quality Start) is shown here (prior to 2009-2010):
|0.913 or better
|0.900 to 0.912
|0.885 to 0.899
|0.884 or worse
Quality Start Percentage (QS%) is simply the number of Quality Starts / Games Started. This gives you a sense of how often the goalie has a Quality Start. A good rule of thumb for this stat is that anything less than 50% is bad, anything over 60% is among the league leaders, and the league average for an NHL regular is about 53.4%. Also, according to Vollman: "Based on the average of every goalie with fewer than ten starts in a season, the average for replacement -level goalies is 42.8% but, in fairness, there is some selectino bias involved in this since playing that poorly will generally limit you to ten start in the first place."¹
Really Bad Starts (RBS) is another stat coined by Vollman that is "awarded" whenever a goalie has a save percentage in a game less than 85%. A team only has a 10% chance of winning when the goalie has save percentage that low.
Where to find these new stats
"So," you are asking, "where do I find these new stats?" They are available anywhere goalie stats have been listed on the site. Due to a limitation of our data source, they are only available going back to the 2007-08 season. So, if you go to a goalie page, you'll see the stats listed in the NHL Standard and NHL Playoffs sections.
Additionally, you can see the league-wide stats on the yearly page for a league. For instance, the 2015 stats are available here.
Finally, the stats are available on the Player Season Finder for goalies and the Player Playoff Finder for goalies.
As always, we welcome any feedback.
Posted in Advanced Stats, Announcement, Hockey-Reference.com | 2 Comments »
8th October 2014
Since we've launched the various Play Index tools on our sites, we've struggled with the best way to monetize what we feel is the deepest and most powerful set of sports data tools anywhere. We know they are useful, our users, the teams, broadcast networks, and newspaper reporters tell us so.
On baseball, we charge an annual fee of $36, but our best guess is that the traffic and audience of the other sites would not be sufficient to make a subscription model worth our while. We have been running traditional banner ads on those pages, but they pay so little and the play index content is niche content (valuable, but niche content) so the traffic will never rise that high on the play index pages. So we have decided to remove all banner advertising and instead run Google consumer surveys on the non-baseball play index pages.
Here's how it works. The first time you run a report each day your results will be obscured and you will be asked to answer a 1-5 question survey. After you answer the survey, your requested report will be shown to you and you won't see another survey for that site on that browser for 24 hours (multi-question surveys are good for 48 hours). These surveys pay us around 5 cents per survey which, while it doesn't sound like much, is about 15 times what we would get from banner ads on those pages. Here is an FAQ if you are wondering how Google uses this information.
Also, if you find these gateway surveys a step too far, consider subscribing to our Ad-Free Access. Starting at $20/year you can surf every Sports Reference page ad and survey free.
If you are wondering what our play index tools are, give them a try. We think you'll enjoy them.
Play Index at Basketball-Reference.com
Play Index at Pro-Football-Reference.com
Play Index at Hockey-Reference.com
Play Index at Sports-Reference.com/cbb
Play Index at Sports-Reference.com/cfb
Posted in Advanced Stats, Announcement, Basketball-Reference.com, CBB at Sports Reference, CFB at Sports Reference, Hockey-Reference.com, Pro-Football-Reference.com | Comments Off on Play Index Tools Now with Short Surveys & no Ads
25th September 2014
As part of our plan to continue to add "advanced" stats to the site, we've included new rate stats on the "Additional Stats" pages for players. These new stats are primarily "per 60 minutes" versions of already existing stats, but we also have included a break out of assists to include both 1st and 2nd assists per 60 minutes. The advantage of the "per 60" versions of these stats is that it makes it easier to compare players with wildly varying amounts of ice time.
These stats are available by clicking on the "Additional Stats" link on the player pages:
As always, we appreciate any feedback or questions.
Posted in Advanced Stats, Announcement, Hockey-Reference.com | 4 Comments »
3rd September 2014
It's been a while since we last announced new "advanced stats" features to Hockey-Reference, but with the latest shake ups in the hockey stats world, we felt that this was a good time to begin rolling out new features for the more statistically-minded fans out there. This is by no means the only roll out, as there are numerous features to come, but we feel it is better to roll out features as we complete them rather than waiting until they are all done.
That said, here's what we've added:
1) Additional Possession Stats to the Player Additional Stats Page
We've added Situational TOI, Relative Corsi For %, Relative Fenwick For %, On-Ice Save and Shooting %, and Offensive and Defensive Zone Start %. You can find these on any player page after 2006 by clicking on Additional Stats at the top of their regular stats.
2) A Play Index Finder for Team Advanced Stats
Until we can add additional stats to all of the league and team pages, we've created a simple Finder that allows you to run queries against the team advanced stats. To find this tool, just go to the top-left menu and click play index, then Team Advanced Stats Finder. This report can be run on any or all teams from 2007-2014.
3) Finally, as a small addition to the play-by-play stats, we've included Faceoff Wins and Losses.
We hope you enjoy these new features and keep checking back often as we add more #FancyStat features as we approach the new season. As always, we appreciate any questions and feedback.
Posted in Advanced Stats, Announcement, Hockey-Reference.com, Play Index | 8 Comments »
17th June 2014
We like to keep our sites really, really fast and obsess over as many performance details as we can. Unfortunately one area where we have little control is in the performance of our advertisements as companies of our size typically outsource this task. We rely on ads for 85% of our revenue, but we are aware that some of these ads degrade our service.
We have decided to launch a new ad-free service. It's available on all of our sites and for $20 you can surf all of our sites ad-free. To sign up see the "AD FREE" link that now appears at the top of every page.
Using this service will make your surfing experience much faster and will continue to support our site.
Sign up via Baseball-Reference.com
Read the rest of this entry
Posted in Announcement, Baseball-Reference.com, Basketball-Reference.com, CBB at Sports Reference, CFB at Sports Reference, Hockey-Reference.com, Olympics at S-R, Pro-Football-Reference.com | 9 Comments »
17th June 2014
As I mentioned a couple of weeks ago, we continue to work on our possession and play-by-play stats, and we have released the next enhancement, additional situations.
As a brief primer on what the new situations mean:
- All Situations, Even Strength, Power Play, and Short-Handed are the traditional situations and mean what you expect them to.
- 5-on-5 is situations where there were five skaters on the ice for both teams. It is similar to Even Strength, but doesn't include 4-on-4.
- 5-on-5 Close is situations where there were five skaters on the ice for both teams and:
- The score is within 1 point in the 1st or 2nd Period, or
- The score is tied in the 3rd Period.
- 5-on-5 Tied is situations where there were five skaters on the ice for both teams and the score was tied.
When you click on a situational tab, all of the stats on that page are for that situation (Blocks, Hits, Corsi For, etc.).
Where are they?
To find these stats, go to a player's page (one who's played in a game since the 2007-08 season) and you will see a link to "Additional Stats":
Clicking on that will take you to the Additional Stats pages where you can see these new enhancements.
We welcome any feedback or questions you have about these new features.
Posted in Advanced Stats, Announcement, Features, Hockey-Reference.com | Comments Off on Additional Possession and Play-by-Play Enhancements on Hockey-Reference