Login

juke · 09-04-2020, 11:11 AM

Ready for grading

(I copied this intro from the SHL version, feel free not to include it in grading)

SMJHL playoffs are here, which means it’s once again time to run the numbers and report the percent chance that each team has of winning any given game, and subsequently their series. Just like my media series about this last playoffs, the graphs and series chance for each matchup will be updated after every sim or two, and some context and stats will be added for individual games, as well as the series as a whole. This year, @"Dexteria" and @”SDcore” will be joining me in covering some of the specific matchups for the championship - a divisional showdown. As always, before just going straight into the numbers, I’d like to clarify some of the background and theory behind how this breakdown works.The methodology behind the numbers and calculations are the same as last year, so feel free to skip straight to the breakdowns if you’re familiar with how they’re calculated, or just don’t care.

Numbers only: The most important thing for me to clarify, is that these aren’t my ‘predictions’ of who’s going to win each series. It’s just reporting the probabilities of team wins, based on that team’s scoring trends. There’s no context behind the probabilities, it’s literally just statistics. Other members of the SJHN have/will put out their own spins of the specific breakdown of each matchup, and those pieces are where you’ll find more context and predictions about each series.

Tactics: This is the biggest flaw in all these numbers. They’re just using season averages, where a lot of teams likely found the best tactics for their team in a broad, general sense, and went with them to try and win as many games as possible. However, come playoffs with 4-7 games against the same opponent, teams will be trying to counter the other, or maybe have new line combinations or tactic settings to give them an advantage. This can’t really be reported in my numbers (other than head to head factor which I talk about in a second), so certain teams are going to match up better or worse then the numbers suggest, based on how well they can adjust their strategies.

Estimated score: This is the entire foundation of all the probabilities in this series. For each matchup, each team is given an estimated number of goals they’ll score in any given game, which is then used for all the rest of the math in these posts. This number is a function of each team’s scoring trends, both goals for and against, during the 50 game season, relative to league average. The number generated won’t be a real score, it’ll be something impossible to actually score, like 3.41 as a random example. But we use that value to calculate the probability of each team scoring a real number of goals, ranging from 0 to 8 in a single game.

Game chances: Using this estimated score and the probabilities that each team scores a certain amount of goals, we can estimate the percent chance that each team will win any given game.

Series chances: Once we have the percent chance that each team will win in either location, we can extrapolate to calculate the probability of each team winning at least 4 out of the 7 games for the series win.

No momentum: I know that momentum is a thing in FHM, but these numbers won't take that into account directly. However, I’ve changed how the expected goals are calculated this season, and the matchup values will be weighted more heavily towards the team’s head-to-head record on the season then I weighted them last playoffs. This means that if a team wins 2 games in a row in the series, although the actual momentum modifier from FHM won’t be considered, that team’s odds will still increase more significantly, because they’ll also have performed better in the head-to-head modifier on my end.

Home vs. Away: Also a new change from last playoffs, I decided to scrap the home vs. away distinction in the code. Although some teams really did perform much differently based on location, I decided 25 games for each was too small of a sample size, and things like a few bad games, or playing backup goalies more in one location could really skew the numbers, and as a result, the entire probabilities. Team scoring stats are based on their 50 game season stats, as well as updated for playoff stats as each game is played.

Scoring graphs: I also decided to not make the individual scoring probability graphs, which show the chances of all possible score outcomes for games. I figured no one really looked at them or cared too much. They’re automatically generated when I make the rest of the figures, so I have them if anyone wants them, but I really doubt anyone does.

Matchup breakdown: This is the first of the graphs that will be updated after each game. It compares the two teams in each matchup, and how well each of them are performing relative to each other in a few key stats. Before game 1, the matchup breakdown graphs will just be using season stats, but after game 1, it will be representative of the current playoff matchup. The numbers plotted are per game, and the exact value for each team is also on the graph.

Series tracker: This final graph represents each team’s series chances after each game. Before game 1, it will be a very boring graph with just two points, but after each game it will track how the series chances change with each team’s win or loss.

Ok lets get on to the probabilities.

The championship version of this series kind of got derailed because a) I didn’t have the index from the end of last round so I don’t have g1 or g2 stats, or overall series stats, b) games were simmed back to back so I can’t get individual game graphs, and c) I shifted my focus to trying to help win the series for NL. So I’m just gonna throw a few graphs down for this one but that’s it.

MET vs. NL Berserkers

Pre Series

In a rare showing in sports, we have a finals matchup between two division opponents, so the two teams have had plenty of time to see what the other team is bringing to the table. MET’s stellar playoff run continues to the finals, as they knocked off the West’s two best teams from the regular season, but now they have to face the league’s best team in NL. On the other side of the ice, NL was able to exorcise their CAR demons from last season, beating the strong Kraken team in 7 competitive games, and move on to their first finals in franchise history. Both teams have very high powered offenses, but NL let in almost 0.9 less goals per game than the Timber in the regular season. Despite relatively small advantages in shots and corsi, the two teams match up pretty evenly across the board in other stats. NL decisively won the series matchups 6-0-1 this year, but head-to-head results from the regular season don’t mean anything when it comes to the finals, especially with a red hot team like MET. Because of the much lower goals against, NL will start the series with a 57.3% chance of winning any game, leading to a 66.2% chance at bringing home their first cup. Regardless, this series has all the makings for a high-offense back and forth battle, and the winner of this will ultimately be decided by whoever can find out a way to slow down the other’s attack, even if it’s just by a little.

General Series Stats