1. 4.

Wednesday, October 8, 2008

Scoring In The Average NHL Game by The Colonel


Here at OGA, we set out to define just what an NHL game looks like in terms of average performance. We wanted to know how teams score in that game and how it plays out so we can share that assessment with you. In the process, we found several, interesting definitions of what we call “The Average NHL Game.”

Just how did we do this? At OGA, we believe the game changed in order of magnitude with the new rules that followed the Lockout year. Then we counted. Heck. AND Dang. There have been 3690 regular season games played in those three years. That’s a lot to study in such detail when we are busy, frantically preparing to launch this site and evaluate the new season. So we chose to take a representative sampling.

For that sample, we looked at 12 games per team (360 total contests), randomly drawn from the October/November 2005, December 2006/January 2007 and March 2008 timeframes with no contests repeated. This is because we have found play is characteristically different, for different teams, in different seasons, and at different times of the year. (That’s a lot of difference...)

And then we further broke down each of those 360 games into five-minute blocks of time within each period, all on the whim of our lead researcher. You know the kind – he sits eating his donut in his cubicle and hollers great ideas over the divider but somehow doesn’t get the research done himself. In my military world, we call him 'The Good Idea Fairy.'

Except in this case, our researcher was on to something (good gut feeling, Tex!)…

Some Interesting Findings

The first observation discovered that defines the average NHL game actually comes from the sum of all 3690 games. It is the fact that NHL teams net an average of almost six goals in each game. This is not a huge revelation but relates to the overall model.

Past the above statement, when charting all goals scored for our 360-game sample in relation to their five-minute time hacks, we discovered an interesting trend as displayed in Figure 1 below:

Figure 1

That’s right – there are two, prominent spikes in scoring over the course of a typical game. Based on the averages from the sample, the first peak is 10.19% of all goals that are scored between 5:01 and 10 minutes of the second period. The second spike comes in the last five minutes of regulation where 10.77% of goals were registered. Over the remainder of the five-minute blocks, the percentage of goals scored averages 7.9% of the total every five minutes, so scoring is relatively ‘even’ for more than 80% of an average game. In mostly minor, measurable amounts, the overall scoring effort fluctuates up and down throughout a game, just like waves crashing on the shore. This is seen by all as those times where teams seem to be playing at an even effort with no scoring (the valleys), punctuated by periods where one team holds a momentum advantage over another that leads to a goal (the peaks). As you can see by the percentages, however, a game’s overall effort – and the result it generates – is still a very close thing.

Even more interesting was what we found at a second glance. Of the 360-game sample, 253 games (70.28%) displayed what OGA calls The Alternate Scoring Response (ASR). In those games, one team responded within the next five-minute block of an opponent’s goal at least once during the course of a game. While not every one of those goals either tied the score or changed who was in the lead, we at OGA believe the ASR is indicative of the intense will to win that resides in the heart of all Hockey players.

But what about that end-of-game spike? It’s all Empty-Netters from pulling the goalie, right? Our sample says that in games where any scoring occurred in the last five minutes of the game, the team that was behind on the scoreboard netted a game-tying goal 81 of 172 times (47.1%). Another 52 times (30.2% of total time), the score going into the last five minutes had more than two goals separating teams, so very rarely, if at all, involved pulling a goalie. That left only 39 instances / 22.7% of all games where empty net goals were scored. (This 360-game sample percentage is higher than the actual NHL statistics for empty net goals scored over the last three seasons where only 16.2% of all 3690 games ended with an empty net goal.)

The Effort As Boxscore

Taking into account the 360-Game Sample, the ASR and Figure 2 above, the ‘average’ game’s scoring as represented in a modified boxscore occurs as in Figure 2 below:

1st Period: In our model, one goal (16.67% of scoring) occurs in the first period between 10 and 15 minutes into the game by Team A. This figure seems to run contrary to a report by NHL.com columnist John Kreiser (see John’s article dated 21 MARCH 2008). Kreiser’s column, however, does not take into account either empty net goals, an undeniable characteristic in about 16% of all games, or statistics from the 2007/8 season. It is likely these statistics added into his calculations would not display much variance. But it is also not lost on OGA that empty net scoring was up in 2007/8 while at the same time scoring was down an overall 10.03% if 2005/6 is used as a baseline measurement.

2nd Period: The first goal scored by Team A – based on the ASR – is countered in the first five minutes of the second period by a goal from Team B. The second period spike indicates another goal is scored toward the end of the 5:01-10:00 block, again based on the ASR, by the opposing Team. These two goals equal 33.33% of all scores in this model.

3rd Period: Team A will add the forth goal of the game to start this stanza. This allows for 29.72% of the time when the ASR does not yield an alternating score. Team B will fight for and tally a score – goal number five – in the 10:00 to 15:00 minute block and will likely be looking to pull their goalie in the last 1:30 of the game in an attempt to tie and force extra periods. By the OGA model, a final goal will be scored in the last five minutes of the game, totaling three in the third (or 50% of the total). The model above shows the goal being scored by Team A and does not indicate if it is an empty net goal. The 360-Game model tells us this will be the case 52.92% of the time. For the rest of the time, it is either not scored, or is a game-tying goal, sending a contest into bonus Hockey time.


This is a model only, and, as we all know, does not occur as depicted in every game. Such is the constraint of an average model. But the model highlighting the ASR and spikes in scoring is of interest in how we view games. But the revelations of the second period and end-of-game scoring spikes and the Alternate Scoring Response are interesting to contemplate. Even in looking at 2008 NHL Pre-Season games ending on 30 SEPTEMBER, the scoring indicated 40-of-60 games with reportable stats / 66.7% displayed spikes in either the second, and/or end-of-the-third, period, and 29-of-60 / 48.3% of games displayed the ASR.

A study to contemplate direct to you from OGA. If any of you have played and/or coached hockey and know why the scoring spikes or ASR occur, write us and we will present the best responses at a later date.

No comments: