JFresh Profile picture
Making hockey analytics make sense (and look pretty). Writer on stats, the eye test, and hockey in general at @EPRinkside and https://t.co/3WNmfjEOTq
Dec 21, 2023 7 tweets 2 min read
Here are the results of our Third Annual Airing of Grievances Survey.

Which team has the most annoying fanbase?

You voted.

🥇 #LeafsForever
🥈 #GoSensGo
🥉 #LetsGoOilers Image Which team has the most delusional fanbase?

You voted.

🥇 #GoSensGo
🥈 #LeafsForever
🥉 #Canucks Image
Nov 6, 2022 5 tweets 1 min read
- bruins clearly think/thought that they were trading a couple days of uncomfortable PR for a good prospect

- miller still seems to show zero signs that he considers any of this to be anything other than a pain in the ass he has to get through to be a pro hockey player the point was made at the time but somehow still relevant - the tormenting occurred when Miller was a child but the barely-masked indifference and resentment of having to do anything to fix it has happened as an adult.
Nov 6, 2022 7 tweets 2 min read
Some clarification.

On-ice shooting percentage is different from regular (individual) shooting percentage. It is a very useful stat to work out who is producing at unsustainably high or low levels. Image Early in a season, if someone's on-ice shooting % is at an extreme high or low (say, 17% or 2%), you can be basically certain that it will regress towards (but not to) the mean, and so will their production rate.
Nov 6, 2022 4 tweets 1 min read
a tip: if you're going to take a victory lap on a player's increased production early in a season, make sure their on-ice shooting percentage isn't obviously unsustainable last season, Tom Wilson had 24 points in 26 games to start with a 16% oish%. i pointed out it was unsustainable, to a lot of backlash. it was. he ended up with 52 in 78.

suzuki and caufield are rolling at 17% and 16% respectively and it's been 12 games.

Nov 6, 2022 4 tweets 1 min read
I see a whole lot of "but quality of competition" in the replies to this.

Understandable. But remember that the model accounts for competition/linemates. More relevant is barely playing with anyone who's not Savard. It's early. This isn't saying he sucks at hockey. There's no way around it - Guhle/Savard have been caved a lot so far this season. 12% on-ice shooting covers up some things.

As Guhle gets to play with other partners and the season goes on the model will learn more about who to assign credit/blame to.
Aug 7, 2021 5 tweets 1 min read
One of my biggest pet peeves in hockey analysis is "goals against personally responsible for" as a rebuttal to a player's bad defensive numbers (where you go through the GA and pick out whether the guy directly caused it with a turnover or missed coverage) Shockingly, it almost always concludes that the player (especially if they're a forward) actually isn't bad defensively and the numbers are lying to you and blaming him for things outside of his control
Aug 7, 2021 8 tweets 3 min read
It is in fact not misleading.

First of all, on the shooting percentage bit. The shooting percentage is clearly unsustainable. Nurse did not get better scoring chances this season on average but doubled his highest full-season scoring rate and sh%. If you watch the goals themselves, it's pretty clear from the number of five hole tricklers (and even an own goal) why this is the case
Aug 6, 2021 5 tweets 2 min read
Darnell Nurse, signed 8x$9.25M by EDM, is an offensive D who scored a lot of goals this season but who has consistently surrendered quality scoring chances when on the ice. Don't think he's nearly as bad as the overall projection suggests, but this is a BIG number. #LetsGoOilers This is a really tricky player to evaluate. First of all, we know that special teams WAR is tricky, and the Oilers for the past two seasons have ranked highly on the PK while giving up a lot of xG. Could be goaltending, could just be something the model is missing.
Aug 5, 2021 4 tweets 2 min read
NEW PATREON PERK: The WAR Roster Builder 4.0 is LIVE for $5+ subscribers

Build the team of your dreams and see how it would be projected to perform based on Wins Above Replacement. Image Or if you just want to see how your favourite team is projected to perform, use the NHL Depth Chart projector to check out their lineup and how they're projected to rank in every aspect of the game. Image
Aug 5, 2021 5 tweets 2 min read
Tomas Tatar, signed 2x$4.5M by NJ, has simply been one of the league's most effective 5v5 offence creators in the past three seasons. This is a major discount considering his recent body of work. #NJDevils Image If you're not one for analytics and expected goal driving, his production in the past three seasons tells the same story. Image
Jul 12, 2021 5 tweets 2 min read
who cares. nothing wrong with scouting, nobody thinks there shouldn't be scouts. drafting a player with 5 points in 77 games because he showed fearlessness or whatever is I think the issue people take with overvaluing intangibles i have fearlessness when I play beer league. i have heart. i stink

Patric Hornqvist and Brendan Gallagher have fearlessness and heart. they win puck battles, forecheck, and get dangerous scoring chances in front of the net. the intangibles lead to something tangible
Jul 11, 2021 4 tweets 2 min read
NEW PATREON PERK:

Prospect Cards are finally LIVE for $5+ Patreon supporters.

These are built using @TopDownHockey's new NHLe model and include all 2021 draft prospects and all NHL prospects.

Such as 2021 top prospect Owen Power: Image NHLe is a model that translates data from other leagues to project an NHL "equivalency" of point production. From there, it can be used to estimate a player's chance of becoming a star or full-time NHLer

topdownhockey.medium.com/nhl-equivalenc…
Mar 3, 2021 7 tweets 3 min read
My thoughts on PDO (since it's come up):

- Flawed because not all teams have equally talented finisher/goaltenders
- Useful because luck is by far a larger driver than talent
- Useful because it's very hard to just "guess" goaltending and finishing talent for a team... - Useful because while goaltending and finishing do have a talent component (obviously) they are also extremely variable, especially from season to season.
- Useful because it has been shown that season-to-season and mid-season to mid-season it does regress towards 100
Mar 2, 2021 7 tweets 3 min read
A reminder of some of the 2020-21 season visualizations available to $5+ Patreon supporters - we've added a whole lot!

Like these team cards, which show off a team's rank in the most important on-ice stats: And these head-to-head cards, which let you compare two teams directly based on their underlying numbers (offence- and defence-focused cards also available)
Mar 2, 2021 4 tweets 1 min read
My sources are telling me that Eichel to the Sens is certainly a possibility, but the Sabres would want Shane Pinto as the main piece going back. That's most likely a no-go for Ottawa Could things change in the future? Never say never. But the Sens organization considers Pinto to be huge part of their middle six moving forward and it would take a better piece than Eichel to pry him out of Ottawa. #GoSensGo
Mar 1, 2021 8 tweets 7 min read
Weekly 5v5 Team Charts Thread (March 1st): Expected Goals For vs Against (5v5) - March 1

The #GoHabsGo are still the class of the league when it comes to scoring chance share at 5v5. The #Canucks are crazy and high-event. The #GoStars and #LGRW are playing low-event defensive hockey. Image
Feb 28, 2021 4 tweets 1 min read
Fun fact: the Edmonton Oilers rank first in the NHL in the percentage of their shots, shot attempts, and unblocked shot attempts that are taken by defencemen. #LetsGoOilers Evan Bouchard leads all defencemen in personal share of on-ice shots taken. By far. Of the Oilers' 80 5v5 shots on goal with him on the ice, he has taken 31 (39%).

2nd place is Nick Holden with 28%. #LetsGoOilers
Feb 25, 2020 9 tweets 2 min read
Goodrow for a 1st is the most controversial trade of the day, for good reason. I have some thoughts on the deal, which are somewhat contradictory but stay with me. #GoBolts Goodrow has a decently strong analytical profile that places him roughly in the 60th percentile of NHL skaters in terms of wins above replacement. He's also done well enough in bigger minutes this season, and will be cheap next year for cap-strapped TB.