predict NHL futures

predict NHL futures

Choosing a sport

While Pythagorean Expectation was initially developed to predict baseball win totals, it has also be successfully applied to a number of different sports.

Being January we are restricted to only winter sports in the Northern Hemisphere, however we may revisit the case study with baseball in the future. Of the winter sports three were identified as possibilities for the case study; football, ice hockey and basketball.

With the draw being a distinct possibility in each match, and a number of international fixtures planned throughout January football was excluded. We do however plan on having a football based case study in the short term.

Left with ice hockey (NHL) and basketball (NBA), it was decided that the NHL would be the best candidate for the case study. At approximately half way through the season there is already a large amount of information about teams and their ability available making it ideal for our study. At the time of writing teams have played anywhere between 36 and 39 matches, giving an extra dimension to our modelling and testing of predictive power.

NHL futures are available at most major sports books. For this case study we will mostly look at divisional markets rather than overall or conference champions (unless there is an outstanding betting proposition), as these markets progress through until the post season.

One limitation for this study (especially compared to football markets), is the lack of liquidity on betting exchanges for NHL futures markets. Currently the NHL championship market is at 247%, while there are no back or lay bets available in either the conference or divisional markets. Unfortunately this is also the case with NBA markets meaning it is not a better alternative. Seemingly the US  sports are not popular with the majority of Betfair’s punters.

Identifying betting opportunities

Unlike a single match or proposition bet it is not possible to convert the Pythagorean Expectation into an implied probability to find positive expected value bets. Because of this only teams with a significant difference in current wins and expected wins, or expected end of season wins will be considered as betting options.

Additional analysis will also supplement the Pythagorean Expectation, primarily the use of Jeff Sagarin’s NHL Elo Power Rankings which will be used to determine if there is a scheduling reason for a teams current Pythagorean Expectation ratings, and the teams strength moving forward.

Bankroll management

Without the ability to calculate expected value it is not possible to use a Kelly or similar proportional staking system for the case study. Because of this a flat $100 will be used for any bets places, while any laying opportunities will have an exposure of $100.

Initial analysis

As at 2 January the NHL current divisional standings and there corresponding Pythagorean Expectations are as follows (prices quoted are for divisional winner):

Eastern Conference Atlantic

Games PlayedPoints ScoredPoints ConcededPredicted WinsActual WinsDiffExpected Season WinsCentrebet
Tampa Bay3912710024240512.75
Montreal371008621243503.25
Detroit38108952220-2456.00
Toronto3812811421210469.00
Boston3810110319190416.00
Florida358293151613617.00
Ottawa3697991815-33781.00
Buffalo3876128914525251.00

According to our Pythagorean Expectation there should be little change in the Atlantic Division through the remainder of the regular season. Current leaders Tampa Bay are currently exactly where they are expected to be with 24 wins from their 39 matches. Second placed Montreal are according to our model over performing with their 24 wins while Detroit and Toronto look set to fight out for the guaranteed playoff place with 3rd position.

Interestingly Buffalo with their 9 wins are over performing by an amazing 5 matches, however will fall well short of the playoffs.

There seems to be little value currently available in the market for the Atlantic Division.

Eastern Conference Metropolitan

Games PlayedPoints ScoredPoints ConcededPredicted WinsActual WinsDiffExpected Season WinsCentrebet
Pittsburgh371118723230511.58
NY Islanders3711710321254513.75
Washington37108962119-2449.00
NY Rangers35107892120-1487.50
Columbus3589110141623451.00
Philadelphia371031131714-33451.00
New Jersey39831111413-128101.00
Carolina37731001210-225201.00

The Metropolitan Division is the first to really peak our interest. Firstly we will closely be watching the currently 4th placed New York Rangers, our model shows that they will fall only 3 matches short of the divisional title. Pittsburgh are currently leading the division from the NY Islanders who despite seeming to be over achieving are predicted to finish on 51 victories with Pittsburgh.

Sagarin’s ratings currently have the Islanders as the strongest team in the division, and have had a much tougher schedule to date (9th toughest in the league), compared with Pittsburgh (23rd strongest).

At $3.75 the Islanders seem value and are the first of our bets. We will also be watching the NY Rangers.

Western Conference Central

Games PlayedPoints ScoredPoints ConcededPredicted WinsActual WinsDiffExpected Season WinsCentrebet
Chicago38119812625-1562.35
Nashville361067824240543.50
St Louis371089321221483.25
Winnipeg3896922019-14215.00
Dallas36108118161713834.00
Minnesota35100981817-14123.00
Colorado37961121514-13367.00

The Central Division contains the two best teams according to both our model and the Sagarin ratings, Chicago and Nashville. All teams are performing as expected by our model with only a match difference for all, or in the case of the Predators’ no difference.

Interestingly the bookmakers have St Louis as a $3.25 chance to take the division. Our model has them falling 8 matches short of Chicago, while the Sagarin ratings has them a half a goal worse team than the Blackhawks. Despite the possibility of there not being a liquid enough market for the bet, for the purpose of the case study we will lay St Louis.

Western Conference Pacific

Games PlayedPoints ScoredPoints ConcededPredicted WinsActual WinsDiffExpected Season WinsCentrebet
Anaheim3910710420244461.95
Vancouver351039419212474.75
San Jose38104962120-1444.75
Calgary391141032221-14517.00
Los Angeles38103942118-3425.50
Arizona37861211214229251.00
Edmonton3882131108-220501.00

With an additional four matches played, and out performing their Pythagorean Expectation by four wins the Anaheim Ducks are currently leading the NHL’s weakest division. With four games in hand Vancouver are expected to overhaul the Mighty Ducks by season end according to our model, and are the better team according to the Sagarin ratings. At $4.75 the Canucks are a bet for us however this is a close division which could change many times between now and the end of the season.

Current bets

TeamDivisionBack/LayPricePotential PayoutResult
NY IslandersMetroBack3.75375TBC
St LouisWesternLay3.25100TBC
VancouverPacificBack4.75475TBC
Scroll to Top