Uncategorized

# What Are The (Playoff) Odds?

Six NHL seasons have passed since the NHL lockout. Due to rule changes, NHL hockey fundamentally changed. On Goal Analysis has been prognosticating predictions of who has and has not earned a playoff berth at about a 90% accuracy rate since then.

But what have we given you for the playoffs? Can we tell who will win before the series is over, and if so, with what kind of accuracy?

There is no fanciful formulary here. Playoff hockey is blood, and guts and character. It is simpler math than OGA’s regular season Playoff Qualifying Curve (PQC). So there is no reason to hold back the recipe
for predicting what is and is not a good call on who will win a playoff series.

Below, and long in advance of the 2012 NHL Playoffs, we give you the numbers. Keep in mind we will differentiate between overall numbers and predictive numbers. (Predictive numbers are for results that do not end a playoff series and can therefore still be useful for future prognosticating.)

Overall

Just to put this study in perspective, here are some overall numbers:

There have been 514 playoff games since the Lockout, over 90 total series, for an average of 5.71 games per series.

Total possible number of W and L combinations for Game 1’s through Game 7’s = 138

Predictive W and L combinations for Game 1 through Game 6 = 72 (52.17%)

Combinations that have not occurred since the Lockout = 26 (18.84%)

Eastern, predictive W and L combinations for Game 1 through Game 6 that have not occurred = 1 (0.7%)

Eastern, predictive W and L combinations for Game 1 through Game 6 that have not occurred = 8 (5.8%)

Out of all of these, what the best calls to drop an anchor next to?

The Best Calls

What is the best indicator of when a team will win their playoff series? Here at OGA, we like to call it ‘The Rule Of 3’s.’ Be the first team to three W’s and you move on; first to three L’s will be going home. How
accurate is this rule?

Predictive NHL 3 x W’s = Wins Series: 45 of 49 / .918

Overall NHL 3 x W’s = Wins Series: 71 of 76 / .934

Predictive NHL 3 x L’s = Loses Series: 45 of 49 / .918

Overall NHL 3 x L’s = Loses Series: 71 of 76 / .934

Anything predictive in the playoffs with accuracy greater than 90% is a relatively safe call. Interestingly, there are several NHL-wide combinations of predictive W and L combinations where the six-year average is a perfect 100% for a team winning a series:

At Game 4 – W, W, W, L (11 – 0)

At Game 5 – L, W, L, W, W (8 – 0); W, W, L, L, W (7 – 0); L, L, W, W, W (5 – 0); W, L, W, W, L (4 – 0);  W, W, W, L, L (4 – 0); L, W, W, W, L (3 – 0); W, L, L, W, W (2 – 0);  and W, W, L, W, L (2 – 0)

At Game 6L, L, W, W, L, W (4 – 0); W, L, W, W, L, L (4 – 0); L, L, W, W, W, L (3 – 0); L, W, W, W, L, L (3 – 0); W, W, L, W, L, L (3 – 0); L, W, W, L, W, L (3 – 0); W, L, L, W, W, L (3 – 0); W, W, W, L, L, L (2 – 0); L, W, W, L, L, W (2 – 0); L, W, L, L, W, W (2 – 0); L, L, L, W, W, W (2 – 0)

Note there are four At Game 6 combinations in bold print where the Rule of 3’s would indicate the team was most likely to lose the series instead of win. This is why the rule is not 100% accurate and underscores how perseverance and character push teams on to the next round.

For pure numbers, however, most folks may point to the slightly less accurate but larger data set of going W, W, W to open a series with its 26 – 1 / 96.3% accuracy. At Game 3, it is the earliest call of victory that is greater than 90%. And it is the most accurate predictor of a series win you can get.

But is there any difference between the two conferences?

Predictions Back East

Actually, yes. The East’s predictability is not as prevalent as the West’s. They sport 16 predictive combinations of W’s and L’s in series which have been 100% accurate since the Lockout in forecasting victory:

At Game 4 – L, W, L, W (3 – 0)

At Game 5 – L, L, W, W, W (3 – 0); W, W, L, L, W (2 – 0); L, W, L, W, W (2 – 0); L, W, L, W, L (2 – 0);     L, W, W, L, W (1 – 0); L, W, L, L, W (1 – 0); and L, L, L, W, W (1 – 0)

At Game 6 –L, L, W, W, W, L (2 – 0); L, W, L, W, L, W (2 – 0); L, L, W, W, L, W (2 – 0); L, W, W, L, W, L (1 – 0); L, W, L, W, W, L (1 – 0); L, W, W, L, L, W (1 – 0); L, W, L, L, W, W (1 – 0); and L, L, L, W, W, W (1 – 0)

Note the low number of times each of these combinations has occurred. If a one-time occurrence is statistically insignificant, and twice is a mere coincidence, then all but two of these combinations need to be ignored as predictors of Eastern victory. Also key is the fact that half of the total indicating 100% victory consists of combinations that seem to violate the Rule of 3’s in terms of victory because they hit three L’s first.

As with the overall predictive numbers, a series that begins W, W, W in the East predicts a victory for the winning team with 12 – 1 / 92.3% accuracy. So in the East, to look less foolish around the Playoff water cooler or blogosphere, call victory at Game 3 with a W, W, W opening or the first combination above at Game 4 or 5 of a series. Making calls at Game 6 for a series means you will either be brassy or wrong.

Predictions Out West

The West’s predictability data set is larger than the East’s. There are 18 predictive W and L combinations which have called the series winner with 100% accuracy. They also consist of a total of 65 games to the
East’s 25 total data set:

At Game 3 – W, W, W (14 – 0)

At Game 4 – W, W, W, L (8 – 0); W, W, L, W (6 – 0); W, L, W, W (5 – 0); and L, W, W, W (2 – 0)

At Game 5 – L, W, L, W, W (7 – 0); W, L, W, W, L (5 – 0); W, W, W, L, L (4 – 0); W, L, L, W, W (3 – 0); W, W, L, W, L (1 – 0); L, W, W, W, L (1 – 0); and L, L, W, W, W (1 – 0)

At Game 6 – W, W, W, L, L, L (2 – 0); L, W, L, W, W, L (2 – 0); W, L, W, W, L, L (1 – 0); W, L, L, W, W, L (1 – 0); and L, W, L, W, L, W (1 – 0)

Here there is only one violator of the Rule of 3s which coincidentally also occurs in the Eastern Conference. And if removing from the equation all single or double occurrences of a combination is in vogue here, nine of the 18 (a similar 50%) can be discounted. This includes all Game 6 predictive combinations.

But with predictive numbers the champion is W, W and W at Game 3 in the West with its perfect 14 – 0 / 100% accuracy. You can also call the series with a measure of confidence for the first three Game 4, and first four Game 5 combinations above. Don’t make a call with Game 6 combinations above unless you know a lot more information than simply Ws and Ls.

Some Other Interesting Tidbits

In the predictive world, what other interesting predictions are there? I have a couple of significant ones.

How many times is a W, W, W in Game 3 followed by a Loss in Game 4? The answer is 13 of 26 times or 50%.  Only one of those times did the 3 – 0 team wind up losing in Game 7. This was Boston versus Philadelphia in 2010 where their first three wins were followed by four straight losses of 1, 4, 1 and 1 goals respectively. In other words, three times the Bruins could have ended the series by scoring two more goals and potentially been in the Finals two straight seasons.

On the flip side, teams at L, L, L after Game 3 won the next game that same 50% of the time. So an automatic call of an opposite result in Game 4 is either brilliance or perilous depending on the outcome. Flip a coin, here.

How many times has a team going into Game 7 with a series of W, W, W, L, L, L won the deciding game? The answer is twice in the West and zero times in the East. The converse is also true for an L, L, L, W, W, W series’ Game 7 where one Eastern team won and two Western teams lost because all three of these series were intra-Conference contests.

How much parity is there in the NHL? If more difficult competition is an indicator, try this on for size. Of 90 playoff series since the Lockout, 24 or 26.7% have gone to a Game 7. Only seven series went the distance from 2006 – 2008. Almost 71% of all Game 7’s have come in the last three playoff seasons.

At the same time, only 13 of 90, or 14.4%, have been four-game series. While five-game series have occurred 24 times just like seven-gamers, the most often played series is one decided by Game 6 (29 times / 32.3%).

Summary

In NHL playoffs, the Rule of 3’s is the best call you can make in a predictive sense as 91.8% of the time a team beginning a series W, W, W wins or L, L, L loses.

Predictions that are 100% correct are more prevalent in the Western Conference than the Eastern.

There is a 4:1 West to East advantage in 100% predictability for series that historically occur more than twice.

And there are no really strong 100% calls for either Conference after Game 5 in a series.

OGA pledges to keep you abreast of the odds next playoff season as the series play out. The odds as we describe them above for predicting series’ outcomes remain the same through Round 1 in 2012.

## Discussion

• ...This suggests that the LAK will defeat CHI in Game 6 to play the NYR for the Stanley Cup 4 years ago
• Since 2005/6, when a Round 3 team wins Game 5, they lose Game 6 66.7% of the time; also, Western series are determined by Game 5.889... 4 years ago
• ...So those stats point to the NYR claiming victory tonight and playing the winner from the Western Conference in the Finals 4 years ago
• Since 2005/6, when a Round 3 team wins Game 5, they lose Game 6 60% of the time; also, Eastern series are determined by Game 5.778... 4 years ago
• 4. Top 4 in each DIV play 2 rounds seeded within the DIV, 1 round versus other DIV champ and then the Finals between the Conference champs. 4 years ago