Login

juke · 02-23-2020, 08:09 PM

Edit: This is just for round 1. Won't let me change the title to specify

Ready for grading
1802 words (I'm not sure if it's relevant enough for the RMB2, but if it is I would like to use the bonus)

SMJHL playoffs get started tomorrow, so I though I’d make a quick breakdown of the two different match ups, and the expected outcomes of the games and series from a statistical point of view. These breakdowns aren’t for the purpose of team analyses or breaking down team playstyles and rankings, and a few people have put out great articles about that already. This article is purely for looking at certain stats about the game, and using them to predict the most likely outcomes.

Methods:
Just to briefly explain where the numbers in this breakdown come from, they all originate from an ‘expected goals scored’ value, determined for each team in a series match up. The expected goals scored value comes from each team’s goals scored and goals conceded per game statistics relative to the league average. The stats are corrected for home and away stats as opposed to just general season averages. I didn’t think it would actually matter, but some teams weirdly happen to have much different away vs. home stats, as you’ll see below. It still probably doesn’t matter, but oh well.

One we have the expected goals value for each team (which will be an arbitrary value, something like 3.57, not an actual amount of goals they could score), we can use that value to predict the probability that they’ll score an actual real amount of goals. For these breakdowns, I predicted the chances of each team scoring anywhere from 0 to 8 goals in a game. Once we know the probabilities of each team scoring anywhere from 0 to 8 goals a game, we can estimate the probability of all 81 combinations of those scores occurring during a game. Once we get those probabilities, we can add them all up and get a percent chance that each team wins the game. Knowing the probability of a home win vs. away win, we can predict the percent chance that each team wins the 7 game series, once again correcting for which team is home or away.

For each match up, I provide 6 graphs to try and shed some light on who might win.The first two graphs are general statistics about the match up, and the last 4 graphs are expected outcomes of the games for both home and away locations. The first graph simply reports a few team statistics for each team, this time not separated by home or away, but rather general season numbers. The length of the bars dictates which team has more of that value, and how much more than the other team. Keep in mind, that things like goals against, shots against, and PIM, the bar literally means which team has more, so a big bar on these stats is a bad thing.

The second graph isn't exciting at the start of each series at all, in fact it will just be two lone points on the graph before the first game. But it will slowly become more relevant and meaningful as the series progresses. It just shows the percent chance that each team has to win the whole series, coming in to each specific game. I'll continue to update these graphs after every game of the series, as the chance of a series win for each team will obviously change depending on who wins the previous games, and who has the current series lead.

The third graph(s) that has dots all over is the likelihood that the game ends in a certain score. Like I said, I calculated the probability of 81 different score combinations, and the size and color of these dots represent that probability. Bigger and red-er dots mean that particular score is more likely to happen. Bigger dots above the diagonal line mean the away team is favored, and under the line means the home team is favored. This is just regulation, so games can actually end in ties, which as you’ll can be fairly likely. I have an overtime modifier that predicts an overtime winner for the actual game/series “percent win” chances, but it’s behind the scenes and not on the graphs. This graph is repeated twice for each match up, once for each team in the series being the home/away team.

The last graph(s), the bar graph, makes predictions for the margin of victory of each game. Higher bars on the right of the vertical line means the home team is favored, and the left side of the line means the away team is favored. The height of the bars is the probability that each score differential occurs, and bars far away from the center line in either direction indicate the probability of larger and larger goal differentials/blowout games. This graph is also repeated twice for each match up depending on the location of the game.

Hopefully that was all easy enough to follow, and now on to the actual results.

Series 1: Anchorage Armada vs. Newfoundland Berserkers

[Image: matchup1_breakdown.png]

Anchorage Armada @ Newfoundland Berserkers

This matchup has the potential to be so close. You can see from the stat breakdowns, that these teams are very evenly matched. ‘Serks have the slightest advantage in the special teams and overall team corsi. This series is interesting, because neither team is good at home, but Anchorage has a massive difference between home and away, 7-15-3 vs. 14-8-3 records respectively. So Newfoundland may actually have a home ice disadvantage throughout the series. Looking at the graphs confirms this, Anchorage has a huge advantage in scoring when they’re away.

Expected goals scored: 3.55 to 2.46 to Anchorage
Winner: Anchorage has a 61% chance of an away win

Newfoundland Berserkers @ Anchorage Armada Armada

Once again, the away team has the advantage. This time it’s the Berserkers, and it’s even slightly more advantageous than Anchorage’s away advantage.

Expected goals scored: 4.17 to 3.00 to Newfoundland
Winner: Newfoundland has a 64.5% chance of an away win

Series winner: Armada have the absolute slightest of advantages in the series, coming away with a 50.2% chance of a series win, based on the single game outcomes. This is slated to be a tight series, but it might be a tight series by trading blowout games. I personally think home and away stats differentials are fake and a random byproduct of Simon, so I’m not sure how it will hold up.
After game 1: The Armada took game 1, increasing their odds of a series win to 62.9%, compared to Newfoundland's now 37.1% chance of a series win.
After game 2: The Berserkers evened the series up in game two, and now Anchorage has home ice this series. Normally this would be bad, but based on the scoring breakdowns this is a benefit for Newfoundland. With the game 2 win, they now have a 57.8% chance of winning the series.
After game 3: A great late-game heroic effort by Newfoundland gives them the 2-1 series lead. This skyrockets their series chances up to 72.2%. However, winning game 4 in Anchorage is crucial. If the series is tied 2-2 coming back to Newfoundland having home ice, it will drop their chances significantly due to the Armada's away record.
After game 4: We expected this series to be close, and it's lived up to the hype so far. This one might just go 7. The Armada tied it all up, and it's anyone's series. Anchorage now has a 53.1% chance of a series win as we head into a crucial game 5.
After game 5: What a series this has been, back and forth with no team ever holding onto more than a 1 game lead. I mentioned it might be a close series by trading blowout games, and the last two games have been wins of 4 and 3 goals respectively, and only 1 of the 5 games have been a one goal game. Newfoundland now has a 78.3% chance to win the series, and they'll look to do it tonight in game 6 in Anchorage, where the Armada have struggled all year.
After game 6: The Berserkers finish off a great series, and go on to face the Falcons in round 2.

Series 2: Halifax Raiders vs. Kelowna Knights

[Image: matchup2_breakdown.png]

Halifax Raiders @ Kelowna Knights Knights

This series is not predicted to be as close. The Raiders have had a rough season, and I’ve wrote before about how I think the Knights are underrated based on their team stats. You can see from the match up stats that Kelowna has the advantage in almost every category, but the physical game looks pretty even. Looking at the graphs, a home win for the Knights is easily the most likely outcome during these match ups, and a blowout wouldn’t be unlikely either.

Expected goals scored: 3.99 to 2.51 to Kelowna
Winner: Kelowna has a 63.7% chance of a home win

Knights

Kelowna Knights @ Halifax Raiders raiders

No surprise but it’s the same story for the games at Halifax as well, and even a little unbalanced as Halifax had a much worse record at home than they did on the road this season.

Expected goals scored: 3.62 to 2.50 Kelowna
Winner: Kelona has a 66.9% chance of an away win

Series winner: Kelowna has an 80.1% chance of coming away with a series win. It’s not surprising, but a 1/5 chance of a big upset from the last seed Raiders are better odds than I would’ve guessed. An underdog series win certainly isn’t out of the realm of possibility.
After game 1: The Knights took a game 1 win home, increasing their series odds to 88.6% chance of a victory. Halifax's underdog run becomes harder, with only an 11.4% chance of upsetting the Knights now.
After game 2: Amazing effort by the Raiders to end up splitting the first 2 games in Kelowna. They still have an uphill battle to climb if they want to get the series upset, but their chances rose substantially to 22.5% after the win in game 2.
After game 3: Kelowna went back to taking care of business. A solid win on the road puts their chances of a series win back to a comfortable 87.7%.
After game 4: The Knights took care of business in a bizarre game 4, and took both of the games in Halifax. They have a 95.6% chance of winning the series, and will look to do it in front of their home fans in game 5.
After game 5: Halifax lives another day. Down 3-2 they still have a huge uphill battle with a 12% chance of winning the next two games, but I think their performance so far has already outperformed their pre-series expectations, and they could continue to do so with another huge win in game 6.
After game 6: Kelowna meets their first round expectations and takes care of the Raiders in 6.

Nhamlet · 02-24-2020, 06:28 PM

Simon is biased against home teams.

**Lime** · 02-24-2020, 07:33 PM

simon hates newfoundland

Navigation

Extra Menu

About us