Advanced stats are a tool. So use it as such. People who use one stat to try to prove something are like trying to say whether a car is good or bad based on if it has heated seats or not. Some are gonna say it NEEDS the heated seats to be any good while others are gonna say that the heated seats are useless. Where the truth is somewhere between those two extremes.
talking about the shots not hitting the net reminds me of Janne Ninimaa. He would wind up for a slap shot and the guys in the corners would hit the deck, the goalie would get a drink.
I think people miss the point of advanced stats. They're not meant to be perfect. They are just more information. Of course sometimes the 2nd assist is the more important pass. But that is usually not the case. Part of the goal(intention) of advanced stats is to PREDICT who will have success in the future, and 1st assists are better at predicting that. For example, I remember when the Leafs got Tim Connolly and I was excited because he had so many points the previous year. But it turned out they were mainly PP second assists, and he didn't have any success with the Leafs because he wasn't actually the useful player who was driving the offense in BUF.
When looking at single game or a few games then I can somewhat agree that goals and assists mean the most... but over a long period of time and when building a team I’m not looking at goals and assists - I’m looking at it that production is repeatable which is where xG, WAR, GAR, HDSC, xDef etc all come in handy. A guy who’s shooting percentage is really high is most likely not going to score 20 goals again (Morgan Rielly). That’s not to say Rielly is any worse offensively this season even though he scored significantly less. In the end you need to use a healthy mix of the eye test, box score statistics and advanced numbers.
@sennadasilva11 ik i was just referring to him saying that rielly wouldnt hit 20 goals again. And plus Rielly's d partners over the years Ceci/Hainsey/Hunwick. Yikes
@hockeyguy just doesnt want To understand how adv. stats work (sample size etc.). Every NHL team has analytics teams whos report to GM and coaches, so they do it just for fun I guess. 🤷♀️
@@aneetabelowjob976 even if Rielly was healthy, he doubled his career shooting percentage that season which is majority luck. But my main point was that even tho Rielly won’t put up elite goal scoring numbers again, he’s I’d say a top 3 OFFENSIVE defenceman.
Solid take Shannon. One area where advanced stats do help considerably is evaluating defensive play. There are some GREAT defensemen in the league who get overlooked for Norris because they don’t score a bunch of goals but are more valuable to their teams than (some) guys who do score a bunch of points.
Just at look at the Avalanche that one year where they overachieved but lost to the Wild. The stats did predict they would crash back down to earth, their PDO was insanely high all year. Advanced stats are a good measurement for puck possession, whether or not a player drives play or isn’t a detriment to their line. Just not the be all and end all.
Dude...even if it is a very short one, you need to showcase that Ash/S-mart sweater. Never had the full logo in frame and my girl and I are dying to see it. Love your vids, the combination of sheer knowledge (from watching so much hockey) and the dry humor/sarcasm in your delivery is wonderfull. I look forward to your vids every day.
You can't use single game as an example of giveaway / takeaway. Yes of course its person dependent to interpret which is a hit and which is a giveaway, but in the long run the person dependency disappears, since there are several people looking at the stats. There might be an slight error in some player giveaway vs. takeaway ratio, but it still gives a good picture on how the player plays and is useful in comparison. Stone is always leading the takeaway charts, it shows on takeaway charts, Subban is master of losing the puck and it shows there too...
"Oh crap theyre not buzzing are they?" Ive never related to a statement more about the canucks than i have with that one LMAO when the canucks are "buzzing" it means we've passed it 17 too many times in the o-zone LOL
Obviously the eye test and analytics need to be used in combination for the best results. But if I'm drafting a fantasy team or a GM building a real team, advanced stats are the best way imo to help predict the long term effect a player will have. A player can play poorly and be a +2, does that me he had a bad game? No. Does that make him a player I want on my team, most likely not. I think what you said about fit is really important though, since the goal of all these stats is to isolate the impact of individual players as much as possible. Fit is also key like you said and is where the "watch the game" side of the brain is just as important.
Advanced stats are great to predict future outcomes. The big issue is that people will use 1 advanced stats without using the other ones in context. A big one you missed was strength of competition, and strength of linemates, both of which have a huuuuuge impact on outcomes
Totally agree with everything you said here. I usually avoid advanced stats... I understand them all, and don't mind when people use them to help support a point. But some guys with their graphs sound like they don't even watch the games sometimes.
Do you think any casual fan should be interested in them if given the choice?, because I always see them on Twitter and Reddit but at the same time adopt a player's mentality I guess?.....where I think watching players putting up goals and assists win games and are the only stats that matter and should be paid attention to.
We all have favorites and we value qualities in a players differently. Obviously, the best way to get an idea is to watch games often and from a variety of teams. The bias will always be there, even in stats. I love the discussion even if it never leads anywhere.
Corsi and Fenwick are meant to be stats that predict goals. Not to make players look good. Someone with 55% CF and 15 goals in 2019 is more likely to score more goals in 2020 than someone with 45% CF and 20 goals. They are meant to negate the chance/lucky nature of scoring goals. And yes its just a tool. If you have a guy who scores 20 goals every year, vs a guy with a career high of 15, maybe thats better to go off of than Corsi to predict the player’s goal numbers next year.
6:10 - 6:45 “You think because you scored two goals today, you played a good game?”... “Hockey is played at both ends of the ice. You, you coast in your own zone. You skate like a figure skater. And when you get the puck, you don't pass.” -Walter Burns
Plus-minus is only bad because of how it's calculated. Normalizing something like on ice goal differential instead of plus-minus would be valuable. Also people on bad teams are guaranteed to have a bad plus-minus if they play big minutes, even if they play well
Jeff Schultz is the poster boy for this. He had something like +52 on a stacked Caps team during the Young Guns era, but he wasn't that good despite that
Yeah, I realized this when I started playing Be a pro in NHL games. I played as a winger, defended my guy and always had like - 30 despite 100 ponits. The defenders were just dumb and let goals in.
@@barbarian92 +/- just can't be looked at in a vacuum. You need to consider how good that team is and weigh it against the teammates. When I played defence in hockey I took a lot of pride in my +/-. I wasn't a very offensive guy, but I was good at shutting down scoring chances. So even in games where we lost 4-0, I might have still been an even because I did my job when I was on the ice. My points were also low, so that was a negative on me. But if you looked at my +/-, you knew you were getting a good shutdown defender. All these stats can be used to paint a picture. But you should really watch the game first to figure out how that player plays. Then stats can be used to compare effectiveness.
Analytics are a great tool and are important in todays NHL but with the nature of the randomness of Hockey and the way that goalies can have such a huge impact on the game analytics can’t predict things as easily as in a sport like Baseball. However I feel like they can only really help you if you use them correctly.
On top of all this there's also the impact a player makes on their team or the opponent. There are certain goalies and veterans that obviously boost their team's confidence by the little things they do and say when things are down to the wire. People are complex and there's more to the game than stats, but like you said they're part of the picture. I guess the debate is how big a part of the conversation they should be
Corsi doesn’t care about quality of the specific shot, Corsi is a proxy for a teams possession of the puck. Teams that shoot the puck more suggest they possess the puck more. You’re comment on it suggests you don’t understand what corsi is and why people value it
I think your criticism is based on not contextualising the stats properly. Certainly if you take only one advanced stat like Corsi (or even something basic like assists) by itself, it won't tell you much. But in combination with other factors, it will definitely tell you things. Like your example of Corsi - it's not meant to say that a shot attempt is just as good as a shot - but you could for example use it along with shots on goal to find out what percentage of a player's Corsi is actually hitting the net, which can tell you things about the player. Corsi For % can be an indicator of possession time for the team. Looking at xG can tell you how good a goalie is by accounting for shot quality. By using statistical analysis on these and other advanced stats, you can learn all kinds of things about a team or a player, like how good the defence is, how good a particular defenceman is (which isn't really captured by +/- because that also depends on who else is on the ice with you), and so on. Advanced stats are not meant to replace the basic stats, but are additional pieces of information necessary to assess the various aspects of the performance of a team or player, and to separate out one aspect of the game from others.
the issue is the use of the stats rather than the stats themselves, all they are is the collective record of what happened on the ice, not much to argue against. The employment of the stats is what causes issues. Particularly when factoring competition of opponents and line ups, because the analytics to measure those areas is sadly straight piss, esp for cross which is just the opponents corsi% (rel QoC)
See you say these aren’t good stats, but obviously they are Steve Yzerman built the Lightning using advanced stats, look who just won the cup and set a record for the best season in the past two years Not to mention that McCrimmon(Vegas GM), Dubas(Leafs GM), and Sakic(Avalanche GM), are all building their teams using advanced stats and all three of those teams have a great chance of winning the 2020-21 season
fwiw, i think LA/CHI/BOS were all good advanced stats wise when they all won out in the first half of the last decade. also, the Avs were at the tail end of the spectrum, and floundered throughout that entire decade in shit-supershit territory until 2018 when their metrics took an upturn.
What I have learned from this video is that advanced stats, apart from Points Per 60 Minutes, are complete dinkum as they are down to an analysts opinion of every single event in a game. Shannon explains the flaws in the compilation of these stats well and I think his opinion of these stats are well founded.
I feel like advanced stats should be used as a tiebreaker if you have decide between players but not as the main argument. Then again, the last heavy-analytics coach we had in Edmonton was Eakins so I'm probably a bit biased against analytics.
I remember those games, rumor was that the majority of his practice was looking at x's and o's on a whiteboard. I think the biggest issue with Eakins was his lack of experience, he did not know how to find the balance between actual play and using the stats to elevate the production.
I like how Bill Armstrong put it... paraphrasing a bunch here but it's used to ask questions and you use your eyes and, in the player's case, evaluating their own actions to arrive at a conclusion regarding said question. If it is used to create a summary statement about a player you have used it incorrectly. It's a way to improve the development track of a player or to aid in a scout arriving to a conclusion about where a certain facet of a player's skills are at.
A lot of times the so-called secondary assist is the more important one, because that's the pass that gets the puck out of your zone and puts your team on the offense.
As a goalie, I think there is value in advanced stats. Traditional stats like SV% and GAA don’t have much meaning without the context. And Shannon mentions that there is little difference between guys using stats. That’s not really true. For example, the traditional stats comparing Dubnyk to say Price don’t show much difference over an 82 game season. There are situations and circumstances behind every stat that add value to an assessment.
gonna steal a line from Tim and Sid, "Analytics are a lot like a Niki Minaj music video; they show a lot, but not everything." This is a great video breaking down that statement while not getting too into the weeds
Not sure if you've done this but you could easily have someone on to talk about advanced stats. Maybe they could talk about some of your concerns, your skepticism on how advanced stats apply to individuals is fair, but most of your questions have answers.
Eye test will forever be my method of evaluation. A guy could be on a bad line/team and have bad metrics but is better than a guy on a great team whose deficiencies get masked by other players.
Summed it up nicely Shannon. For how much +/- is maligned, I like to look at a player's +/- relative to his team's goal differential and his teamates'.
Some of the things you say, I absolutely disbelieve. But yeh. I remember seeing a Habs Leafs game in Toronto, where the Habs were clearly laying the body. And somehow, the Leafs had like 8x more hits at the breaks. From what I gathered from that game (I was lucky enough to sit behind the scenes on that one), is that the home team racks up the hits, takeaways, etc. So they are biased towards the home team. Just what I heard and saw. But you're totally right. You can't just look at stats and take them literally. I think puck tracking will actually make all of this way less "eye-test", and make fantasy hockey way less fun!
I've said this before, but i figured I'd say it again since you brought it up again, there is a diminishing return on points. What's more likely to be a created goal, a quaternary assist or a goal? obviously a goal, anyone will tell you that. There are certainly going to be a few quaternary assists that create a goal, but it's going to be incredibly rare. A goal is usually going to be a shot or some kind of movement into a scoring area that produces a point. I'd argue same goes for tertiary assists, a goal is much more valuable than those. So my point is, the closer you are to the goal scorer, the more likely you're seeing a goal-creating stat thus goals>primary assists (though i think these 2 are close)>secondary assists as a general rule. And personally, I think it's tough to argue otherwise. Even greasy deflection goals are adding a variable (the player's body) that causes a deflection that might never have occurred otherwise meaning that player still created a goal even if he didn't entirely mean to.
What do you think about WAR (Wins above replacement) and xGA I really like xGA to have some idea of how the player plays defensively. I personally like advanced stats for defensive play and Points or P60 for offensive capabilities
never had time for fake stats, i determine who i like/want on my team by effort which is why hyman has always been my favourite leaf of this era followed by freddy. i don't mind +/- maybe because it's always been there. tells you real quick who can't play defence or possibly who's on a bad team.
An interesting career to look at and do a video on, with all the involvement of advanced stats is Douglas Murray. Advanced stats kinda killed his career a bit. It would be nice to see why he wasn't looked at so fondly once he left San Jose. Was it a reason why the Sharks couldn't get over the hump? He was a top D-man along with Dan Boyle for awhile.
+/- the original advanced stat that now nearly everyone says is garbage. Interestingly, growing up looking at stats on hockey cards, I loved +/- because I felt it added a dimension to players game that couldn’t be garnered from points alone.
Best example of the power of (defensive) linemates I can think of is Scott Stevens / Scott Nidermeyer. Stevens made Nieds better on offense by removing defensive responsibility. Nieds made Stevens better on defense by removing offensive obligation. Both incredible talents, but Nieds got the media attention because of his numbers, while Devils fans knew that it was Stevens who ran the team.
Okay, can someone explain why secondary assist are a thing in hockey? In basketball, the only player that gets the assist is the player that passed it to the player that scored. In soccer, an assist only goes to the player that passes to the one that scored the goal. So, why do secondary assist count in hockey?
I think they should count in other sports as well, so why don’t they? It’s not unusual that the secondary assist is the most important or at least as important for creating the scoring chance.
Advanced stats will never ever be able to take into account hockey players being actual human beings playing with and under other human beings with lives outside of hockey which is sometimes one of the biggest factors in how they actually play
I haven't seen anyone mention corsi for a while... Why not talk about GF60, GA60, xGA60, xGF60 and especially xGA-GA OF GOALIES. Those are useful stats and good to understand for all.
What jersey is that and who did the Seattle Kraken steal their jersey from? Edit: I just saw the sleeve, S-mart, I'm assuming it's a Army of the Darkness special jersey.
Adjusted plus/minus makes sense to me. Adjust it for time on ice and expected plus/minus for the team in that same amount of ES TOI. It’s just a fairer way of identifying good players on bad teams
What's more valuable, a shot that goes a quarter inch wide of the post or one that hits the goalie in the chest? I can see a case for either and I'd personally probably argue they're about the same.
I really believe leaning one way is a mistake, whether it be heavily relying on advanced stats or the opposite in eye test / "basic" stats. in my opinion the best way to get an complete overview or opinion of a player or team is just use all the resources, since there's many flaws with all sides. For example; you can't convince me that you can get a decent opinion on defensive play without using advanced stats
Players with a high percentage of secondary assists are less likely to replicate their assist total compared to those with a high percentage of primary assists.Its a historically proven fact.Take Defencemen for example.Generally speaking Dmen accumulate more secondary assists by throwing the puck up the ice and letting the play unfold. That action is important at it generates an offensive opportunity however it more heavily relies upon the skillset of the forwards in this case.Hence it is typically more difficult for a dman to replicate his 5v5 assist totals than a forward. Thats not to be confused with the importance of the assist however.
I think the biggest problem with advanced stats is the people interpreting them. They are tools after all. However in order to understand what happened you have to tell the story as well. Using both of them hand in hand is the way to go.
Stats like Corsi only as good as the way you view them. Obviously if a player starts in the defensive zone the majority of the time, their Corsi Against is going to increase at a much faster than their Corsi For over the course of a season. You have to take their zone start percentages into account as well to get a true evaluation for the value they bring. You can directly use zone start percentages to alter the corsi data If you assume that any time a faceoff occurs there is a 50/50 chance a team wins and that when a player enters the ice during ongoing play there is an even chance of them enter their team's respective offensive or defensive zone. If you're taking that into account you can directly divide a player's corsi for by their offensive zone start percentage for stat A, and follow the same process for corsi against with defensive zone start percentage for a stat B. If you take use the formula A / (A+B) you would receive a much more accurate Corsi For % stat. By using this formula and accounting for my previous assumptions, you would essentially be receiving a player's Corsi For % if they played equal amounts of time playing offense and defense. It isn't 100% accurate obviously since the faceoff percentages will not always be 50/50 in either zone and a player will not have a 50/50 chance of entering ongoing play in the offensive zone, but as a player plays more shifts throughout a year, the numbers will approach an even result. As a Penguins fan, a guy like Dominik Simon is a great example as to why Corsi stats as it stands are flawed. Simon often gets defended by some penguins fans due to his corsi stats seeming good despite his lack of scoring, myself included for some time. For this past season according to hockey reference, Simon's Corsi for % was 53.1% which is good on the surface, but he started in the offensive zone 59.3% of the time. If you use the formula I mentioned to adjust the stats as if Simon spent 50% of his time on ice in the offensive and defensive zones, his adjusted Corsi For % drops to a measly 43.75%, indicating that his defensive flaws outweigh his offensive production. However if you apply this same method to Teddy Blueger you find a much different result. Again according to hockey reference, Blueger's Corsi for % was 50.4% despite him only starting in the offensive zone 30.2% of the time. His adjusted numbers using that formula skyrocket him to an adjusted corsi for % of 70.16%, which along with his current data indicates that he is exceptional at providing offense and is one of the best players in the league at limiting chances on defense (due to his relatively low corsi against despite a high defensive zone start%). The corsi stats as is are only as good as how you utilize them to look at a player
been saying this forever. people tell me I'm not "with the times" - old school stats don't tell enough of the story. I see those graphs and crease chance stats and my eyes glaze over. Not because I can't follow them, but because as someone who played the game they are false flags that do not tell the story. PLaying, watching, that's how you can tell the value of things that often aren't measured.
I’m pretty sure advanced stats are very useful if used the right way. One thing about shots on goal which is very weird to me is that if you hit the post or crossbar it doesn’t count. Truly unfair, so you at least have to agree that Corsi makes more sense in that situation. 😁
a fourth liner might have decent stats against other 4th liners -----and never touch the puck when he is on against a 1st line crew---stats dont take relative matchups into calculations --for example points per 60 minutes doesnt tell you against whom they are playing
Well of 'corsi', if a player puts pucks in the net at least 20 times a year consistently, and/or gets 30+ assists consistently, of 'corsi' he's a good player, no matter what his corsi-horsy says! And above all, corsi doesn't measure heart.
Yeah but is there a stat that tells me a player's goals per minute while shorthanded in the defensive zone during an overtime period as the home team in a January game during a full moon?
This Shannon. Exactly this! You watch more hockey than all of us, and being able to watch the players so often is more of a judge of what's really happening.
It would be interesting if there actually was a stat good vs bad penalties. I would think that bad penalties are in the majority, usually a penalty is a result of you being overplayed or being out of position.
It depends on what you are looking at, goaltending benefits somewhat from them over sp and gaa, quality of shot against is of value, Use Price as a example, he will often have less then stellar traditional numbers at times, but when you take into account back door plays and deflections you find him top 5 pretty much always, now do these stats tell us any more about Price then when the players rank their peers, idk. For shooters it matters if all the goals they score are scored in games that are already out of reach for the opponent, or if they are protected from the other teams top players , they should excel in that situation. I am not sure how the advanced stats really represent all the factors that matter. I do think they are worth tracking however, its useful when combined with the eye test, It has the advantage of no biases, you either have the numbers or you don't, you just need to look to find the reason why or why not
Hockey is impossible to measure as everything is situational . How many O-zone starts , TOI , what line you play on , How much pp time , Was it a bad line change , how good are your linemates , how good is your team , what style is being played and on & on..... +/- is very useful on one team ( ex: Andreas anthanasiou is so much worse than rest of team ) but it cannot measure team to team accurately . Baseball is the only sport where stats paint an accurate picture (yet pitcher's won lost are useless because of situations-quality of team , do you pitch against other teams best pitchers ?) Basketball is somewhat measurable- they measure performance with 2 or 3 or more , man groups . There are no line changes , bad bounces etc....
I remember when Trouba was being shopped and I said Trouba for Pionk and 1st and that Winnipeg would end up winning the deal. People said Pionk sucked and he wasn't anywhere near Trouba's level based on advanced stats. Well, Pionk has been great for WPG and Heinola looks good as well. Trouba has been good for the Rangers, but the eye test of Pionk was better than the stats showed and GmKC saw this.
At the end of the day I care more about what I see when a player is on the ice than what the numbers say. I have a better opinion of some fourth line players who get a bunch of PIM for making the top line of the other team miserable than I do of a second line player who gets a point every other game, even if the second line guy is statistically better for the team.
Advanced stats are a tool. So use it as such. People who use one stat to try to prove something are like trying to say whether a car is good or bad based on if it has heated seats or not.
Some are gonna say it NEEDS the heated seats to be any good while others are gonna say that the heated seats are useless. Where the truth is somewhere between those two extremes.
That’s a great analogy.
talking about the shots not hitting the net reminds me of Janne Ninimaa. He would wind up for a slap shot and the guys in the corners would hit the deck, the goalie would get a drink.
Then there's all those "third assist" brilliant passes that doesn't land the dude any credit except for knockup up the +/- a notch.
There was a time with 3 A per goal.
Facts, so it all evens itself out in the end
The defensive defence-man assist
+/- is the worst stat to judge a player's performance on. Good lord. The amount of people who think it's the holy stat frustrates me.
The secondary assist can also be even more important then the one who passes to the goalscorer.
I think people miss the point of advanced stats. They're not meant to be perfect. They are just more information. Of course sometimes the 2nd assist is the more important pass. But that is usually not the case. Part of the goal(intention) of advanced stats is to PREDICT who will have success in the future, and 1st assists are better at predicting that. For example, I remember when the Leafs got Tim Connolly and I was excited because he had so many points the previous year. But it turned out they were mainly PP second assists, and he didn't have any success with the Leafs because he wasn't actually the useful player who was driving the offense in BUF.
@@NHL94Communityunfortunately for advanced stats people, trying is the first step towards failure.
it can be but the majority of the time its not and thats just the way it is
@@AC-xq4hh 🗿
Advanced stats have shown to be a net negative.. relying on them hinders teams.
When looking at single game or a few games then I can somewhat agree that goals and assists mean the most... but over a long period of time and when building a team I’m not looking at goals and assists - I’m looking at it that production is repeatable which is where xG, WAR, GAR, HDSC, xDef etc all come in handy. A guy who’s shooting percentage is really high is most likely not going to score 20 goals again (Morgan Rielly). That’s not to say Rielly is any worse offensively this season even though he scored significantly less. In the end you need to use a healthy mix of the eye test, box score statistics and advanced numbers.
morgan rielly was playing with injuries though? If he was playing healthy theres a good chance he hits 20 goals again
@sennadasilva11 ik i was just referring to him saying that rielly wouldnt hit 20 goals again. And plus Rielly's d partners over the years Ceci/Hainsey/Hunwick. Yikes
I think people need to understand that points doesn’t always mean you are driving play you might just have some great linemates
@hockeyguy just doesnt want To understand how adv. stats work (sample size etc.). Every NHL team has analytics teams whos report to GM and coaches, so they do it just for fun I guess. 🤷♀️
@@aneetabelowjob976 even if Rielly was healthy, he doubled his career shooting percentage that season which is majority luck. But my main point was that even tho Rielly won’t put up elite goal scoring numbers again, he’s I’d say a top 3 OFFENSIVE defenceman.
This will surly not upset some people.
yeah, I can’t see how anyone could disagree with this
and yeah, I am of the same opinion with Shannon
Corsi is about getting pucks AT the net, not always ON the net; it correlates to more pressure and hence, more chances.
Solid take Shannon.
One area where advanced stats do help considerably is evaluating defensive play. There are some GREAT defensemen in the league who get overlooked for Norris because they don’t score a bunch of goals but are more valuable to their teams than (some) guys who do score a bunch of points.
Just at look at the Avalanche that one year where they overachieved but lost to the Wild. The stats did predict they would crash back down to earth, their PDO was insanely high all year.
Advanced stats are a good measurement for puck possession, whether or not a player drives play or isn’t a detriment to their line. Just not the be all and end all.
I'm a fan of simple stats yet find analyzers fascinating. How they compute these number's amazing.
Dude...even if it is a very short one, you need to showcase that Ash/S-mart sweater. Never had the full logo in frame and my girl and I are dying to see it.
Love your vids, the combination of sheer knowledge (from watching so much hockey) and the dry humor/sarcasm in your delivery is wonderfull. I look forward to your vids every day.
You can't use single game as an example of giveaway / takeaway. Yes of course its person dependent to interpret which is a hit and which is a giveaway, but in the long run the person dependency disappears, since there are several people looking at the stats. There might be an slight error in some player giveaway vs. takeaway ratio, but it still gives a good picture on how the player plays and is useful in comparison. Stone is always leading the takeaway charts, it shows on takeaway charts, Subban is master of losing the puck and it shows there too...
"Oh crap theyre not buzzing are they?" Ive never related to a statement more about the canucks than i have with that one LMAO when the canucks are "buzzing" it means we've passed it 17 too many times in the o-zone LOL
If you torture the numbers long enough, they will confess to anything!
Great quote, had to look it up: businessday.ng/opinion/article/if-you-torture-the-data-long-enough-it-will-confess-to-anything/
Well said.
Obviously the eye test and analytics need to be used in combination for the best results. But if I'm drafting a fantasy team or a GM building a real team, advanced stats are the best way imo to help predict the long term effect a player will have.
A player can play poorly and be a +2, does that me he had a bad game? No. Does that make him a player I want on my team, most likely not.
I think what you said about fit is really important though, since the goal of all these stats is to isolate the impact of individual players as much as possible. Fit is also key like you said and is where the "watch the game" side of the brain is just as important.
Advanced stats are stupid
I’ve been watching hockey for over 50 years and in my opinion the eye test always rules over the stat sheets!
Advanced stats are great to predict future outcomes. The big issue is that people will use 1 advanced stats without using the other ones in context. A big one you missed was strength of competition, and strength of linemates, both of which have a huuuuuge impact on outcomes
Totally agree with everything you said here. I usually avoid advanced stats... I understand them all, and don't mind when people use them to help support a point. But some guys with their graphs sound like they don't even watch the games sometimes.
Do you think any casual fan should be interested in them if given the choice?, because I always see them on Twitter and Reddit but at the same time adopt a player's mentality I guess?.....where I think watching players putting up goals and assists win games and are the only stats that matter and should be paid attention to.
We all have favorites and we value qualities in a players differently. Obviously, the best way to get an idea is to watch games often and from a variety of teams. The bias will always be there, even in stats. I love the discussion even if it never leads anywhere.
Corsi and Fenwick are meant to be stats that predict goals. Not to make players look good. Someone with 55% CF and 15 goals in 2019 is more likely to score more goals in 2020 than someone with 45% CF and 20 goals. They are meant to negate the chance/lucky nature of scoring goals.
And yes its just a tool. If you have a guy who scores 20 goals every year, vs a guy with a career high of 15, maybe thats better to go off of than Corsi to predict the player’s goal numbers next year.
6:10 - 6:45
“You think because you scored two goals today, you played a good game?”... “Hockey is played at both ends of the ice. You, you coast in your own zone. You skate like a figure skater. And when you get the puck, you don't pass.”
-Walter Burns
Love the hat... Getting a bit nostalgic, grew up watching them in the 80s/90s.
Plus-minus is only bad because of how it's calculated. Normalizing something like on ice goal differential instead of plus-minus would be valuable. Also people on bad teams are guaranteed to have a bad plus-minus if they play big minutes, even if they play well
Jeff Schultz is the poster boy for this. He had something like +52 on a stacked Caps team during the Young Guns era, but he wasn't that good despite that
Yeah, I realized this when I started playing Be a pro in NHL games. I played as a winger, defended my guy and always had like - 30 despite 100 ponits.
The defenders were just dumb and let goals in.
@@barbarian92 +/- just can't be looked at in a vacuum. You need to consider how good that team is and weigh it against the teammates.
When I played defence in hockey I took a lot of pride in my +/-. I wasn't a very offensive guy, but I was good at shutting down scoring chances. So even in games where we lost 4-0, I might have still been an even because I did my job when I was on the ice.
My points were also low, so that was a negative on me. But if you looked at my +/-, you knew you were getting a good shutdown defender.
All these stats can be used to paint a picture. But you should really watch the game first to figure out how that player plays. Then stats can be used to compare effectiveness.
Analytics are a great tool and are important in todays NHL but with the nature of the randomness of Hockey and the way that goalies can have such a huge impact on the game analytics can’t predict things as easily as in a sport like Baseball. However I feel like they can only really help you if you use them correctly.
On top of all this there's also the impact a player makes on their team or the opponent. There are certain goalies and veterans that obviously boost their team's confidence by the little things they do and say when things are down to the wire. People are complex and there's more to the game than stats, but like you said they're part of the picture. I guess the debate is how big a part of the conversation they should be
Corsi doesn’t care about quality of the specific shot, Corsi is a proxy for a teams possession of the puck. Teams that shoot the puck more suggest they possess the puck more. You’re comment on it suggests you don’t understand what corsi is and why people value it
I think your criticism is based on not contextualising the stats properly. Certainly if you take only one advanced stat like Corsi (or even something basic like assists) by itself, it won't tell you much. But in combination with other factors, it will definitely tell you things.
Like your example of Corsi - it's not meant to say that a shot attempt is just as good as a shot - but you could for example use it along with shots on goal to find out what percentage of a player's Corsi is actually hitting the net, which can tell you things about the player.
Corsi For % can be an indicator of possession time for the team. Looking at xG can tell you how good a goalie is by accounting for shot quality. By using statistical analysis on these and other advanced stats, you can learn all kinds of things about a team or a player, like how good the defence is, how good a particular defenceman is (which isn't really captured by +/- because that also depends on who else is on the ice with you), and so on. Advanced stats are not meant to replace the basic stats, but are additional pieces of information necessary to assess the various aspects of the performance of a team or player, and to separate out one aspect of the game from others.
the issue is the use of the stats rather than the stats themselves, all they are is the collective record of what happened on the ice, not much to argue against. The employment of the stats is what causes issues. Particularly when factoring competition of opponents and line ups, because the analytics to measure those areas is sadly straight piss, esp for cross which is just the opponents corsi% (rel QoC)
*Esp for corsi
See you say these aren’t good stats, but obviously they are
Steve Yzerman built the Lightning using advanced stats, look who just won the cup and set a record for the best season in the past two years
Not to mention that McCrimmon(Vegas GM), Dubas(Leafs GM), and Sakic(Avalanche GM), are all building their teams using advanced stats and all three of those teams have a great chance of winning the 2020-21 season
Not really they still watch all the games to. They use advance stats as part of the picture aswell.
And I think NHL teams use much more complicated stats systems than those available to public.
fwiw, i think LA/CHI/BOS were all good advanced stats wise when they all won out in the first half of the last decade.
also, the Avs were at the tail end of the spectrum, and floundered throughout that entire decade in shit-supershit territory until 2018 when their metrics took an upturn.
What I have learned from this video is that advanced stats, apart from Points Per 60 Minutes, are complete dinkum as they are down to an analysts opinion of every single event in a game. Shannon explains the flaws in the compilation of these stats well and I think his opinion of these stats are well founded.
15:26 it’s so true that you can make an argument for why an player is good or bad.
I miss an stat measuring howe many errors results in a goal against. Like in tennis.
I feel like advanced stats should be used as a tiebreaker if you have decide between players but not as the main argument. Then again, the last heavy-analytics coach we had in Edmonton was Eakins so I'm probably a bit biased against analytics.
I remember those games, rumor was that the majority of his practice was looking at x's and o's on a whiteboard. I think the biggest issue with Eakins was his lack of experience, he did not know how to find the balance between actual play and using the stats to elevate the production.
I like how Bill Armstrong put it... paraphrasing a bunch here but it's used to ask questions and you use your eyes and, in the player's case, evaluating their own actions to arrive at a conclusion regarding said question. If it is used to create a summary statement about a player you have used it incorrectly. It's a way to improve the development track of a player or to aid in a scout arriving to a conclusion about where a certain facet of a player's skills are at.
LOOOOVE the Cape Breton Oilers hat!!!
I almost spilt my food over at horsey Corsi 😂😂😂
A lot of times the so-called secondary assist is the more important one, because that's the pass that gets the puck out of your zone and puts your team on the offense.
Love this, Shannon!
As a goalie, I think there is value in advanced stats. Traditional stats like SV% and GAA don’t have much meaning without the context.
And Shannon mentions that there is little difference between guys using stats. That’s not really true. For example, the traditional stats comparing Dubnyk to say Price don’t show much difference over an 82 game season. There are situations and circumstances behind every stat that add value to an assessment.
gonna steal a line from Tim and Sid, "Analytics are a lot like a Niki Minaj music video; they show a lot, but not everything." This is a great video breaking down that statement while not getting too into the weeds
Not sure if you've done this but you could easily have someone on to talk about advanced stats. Maybe they could talk about some of your concerns, your skepticism on how advanced stats apply to individuals is fair, but most of your questions have answers.
Eye test will forever be my method of evaluation. A guy could be on a bad line/team and have bad metrics but is better than a guy on a great team whose deficiencies get masked by other players.
Agreed. That's the paragon way.
I love seeing the humorous side of shannon
Summed it up nicely Shannon. For how much +/- is maligned, I like to look at a player's +/- relative to his team's goal differential and his teamates'.
Yeah it's actually pretty valuable for comparing people/lines within the same team.
I fully support you going full Happy Gilmore from time to time.
Some of the things you say, I absolutely disbelieve. But yeh. I remember seeing a Habs Leafs game in Toronto, where the Habs were clearly laying the body. And somehow, the Leafs had like 8x more hits at the breaks. From what I gathered from that game (I was lucky enough to sit behind the scenes on that one), is that the home team racks up the hits, takeaways, etc. So they are biased towards the home team. Just what I heard and saw. But you're totally right. You can't just look at stats and take them literally. I think puck tracking will actually make all of this way less "eye-test", and make fantasy hockey way less fun!
I've said this before, but i figured I'd say it again since you brought it up again, there is a diminishing return on points. What's more likely to be a created goal, a quaternary assist or a goal? obviously a goal, anyone will tell you that. There are certainly going to be a few quaternary assists that create a goal, but it's going to be incredibly rare. A goal is usually going to be a shot or some kind of movement into a scoring area that produces a point. I'd argue same goes for tertiary assists, a goal is much more valuable than those. So my point is, the closer you are to the goal scorer, the more likely you're seeing a goal-creating stat thus goals>primary assists (though i think these 2 are close)>secondary assists as a general rule. And personally, I think it's tough to argue otherwise. Even greasy deflection goals are adding a variable (the player's body) that causes a deflection that might never have occurred otherwise meaning that player still created a goal even if he didn't entirely mean to.
What do you think about WAR (Wins above replacement) and xGA I really like xGA to have some idea of how the player plays defensively. I personally like advanced stats for defensive play and Points or P60 for offensive capabilities
never had time for fake stats, i determine who i like/want on my team by effort which is why hyman has always been my favourite leaf of this era followed by freddy. i don't mind +/- maybe because it's always been there. tells you real quick who can't play defence or possibly who's on a bad team.
I agree with your thoughts on this.
I Really Liked this episode. THANKS
loving the Cape Breton Oilers hat. Been hunting for a Jersey for a long time
An interesting career to look at and do a video on, with all the involvement of advanced stats is Douglas Murray. Advanced stats kinda killed his career a bit. It would be nice to see why he wasn't looked at so fondly once he left San Jose. Was it a reason why the Sharks couldn't get over the hump? He was a top D-man along with Dan Boyle for awhile.
There was one game in the playoffs where the lightning only missed the net once which was impressive.
Damn jimmy killed the mascot
😂😂
There’s so much misinformation and intended + unintended misunderstanding going on that I don’t even know where to begin
Not gonna lie, when you said tap it in I was so busy quoting Happy Gilmore I barely heard your own reference lol
I'm just happy with the Cape Bretton Oilers hat. That's going back...
+/- the original advanced stat that now nearly everyone says is garbage. Interestingly, growing up looking at stats on hockey cards, I loved +/- because I felt it added a dimension to players game that couldn’t be garnered from points alone.
This video raises the possibility of another video:
“Stats vs the eye test”
2:07, just taaaaaaap it in, tap, tap, taperoo.
Best example of the power of (defensive) linemates I can think of is Scott Stevens / Scott Nidermeyer. Stevens made Nieds better on offense by removing defensive responsibility. Nieds made Stevens better on defense by removing offensive obligation. Both incredible talents, but Nieds got the media attention because of his numbers, while Devils fans knew that it was Stevens who ran the team.
You need to use both analytics and the eye test to get the best opinion of a player/team imo
Okay, can someone explain why secondary assist are a thing in hockey?
In basketball, the only player that gets the assist is the player that passed it to the player that scored.
In soccer, an assist only goes to the player that passes to the one that scored the goal.
So, why do secondary assist count in hockey?
I think they should count in other sports as well, so why don’t they? It’s not unusual that the secondary assist is the most important or at least as important for creating the scoring chance.
Wasn't ryan graves like a +46. Crazy advanced stats from him !!!!
Love the hat Shannon! Reppin Cape Breton love to see it
Advanced stats will never ever be able to take into account hockey players being actual human beings playing with and under other human beings with lives outside of hockey which is sometimes one of the biggest factors in how they actually play
I haven't seen anyone mention corsi for a while... Why not talk about GF60, GA60, xGA60, xGF60 and especially xGA-GA OF GOALIES. Those are useful stats and good to understand for all.
What jersey is that and who did the Seattle Kraken steal their jersey from?
Edit: I just saw the sleeve, S-mart, I'm assuming it's a Army of the Darkness special jersey.
Adjusted plus/minus makes sense to me. Adjust it for time on ice and expected plus/minus for the team in that same amount of ES TOI. It’s just a fairer way of identifying good players on bad teams
What's more valuable, a shot that goes a quarter inch wide of the post or one that hits the goalie in the chest? I can see a case for either and I'd personally probably argue they're about the same.
For the 2019-2020 Playoffs, Stamkos had 22.1 Goals per 60. Stats can be fun
I would use advance stats for defences
It would show how good are they in tough situations and in soft situations
but it would be one tool you would use, not the only tool you would use.
@@jeremydanchuk1897 off course especially if don't see them play as often
I really believe leaning one way is a mistake, whether it be heavily relying on advanced stats or the opposite in eye test / "basic" stats. in my opinion the best way to get an complete overview or opinion of a player or team is just use all the resources, since there's many flaws with all sides. For example; you can't convince me that you can get a decent opinion on defensive play without using advanced stats
Players with a high percentage of secondary assists are less likely to replicate their assist total compared to those with a high percentage of primary assists.Its a historically proven fact.Take Defencemen for example.Generally speaking Dmen accumulate more secondary assists by throwing the puck up the ice and letting the play unfold.
That action is important at it generates an offensive opportunity however it more heavily relies upon the skillset of the forwards in this case.Hence it is typically more difficult for a dman to replicate his 5v5 assist totals than a forward.
Thats not to be confused with the importance of the assist however.
In the liga they track how far you skate every game id like to see that in the NHL
Hedman and McDavid’s stats🤭
Another advanced stat was " war"
Then theres money puck too
I think the biggest problem with advanced stats is the people interpreting them. They are tools after all. However in order to understand what happened you have to tell the story as well. Using both of them hand in hand is the way to go.
Stats like Corsi only as good as the way you view them. Obviously if a player starts in the defensive zone the majority of the time, their Corsi Against is going to increase at a much faster than their Corsi For over the course of a season. You have to take their zone start percentages into account as well to get a true evaluation for the value they bring. You can directly use zone start percentages to alter the corsi data If you assume that any time a faceoff occurs there is a 50/50 chance a team wins and that when a player enters the ice during ongoing play there is an even chance of them enter their team's respective offensive or defensive zone.
If you're taking that into account you can directly divide a player's corsi for by their offensive zone start percentage for stat A, and follow the same process for corsi against with defensive zone start percentage for a stat B. If you take use the formula A / (A+B) you would receive a much more accurate Corsi For % stat. By using this formula and accounting for my previous assumptions, you would essentially be receiving a player's Corsi For % if they played equal amounts of time playing offense and defense. It isn't 100% accurate obviously since the faceoff percentages will not always be 50/50 in either zone and a player will not have a 50/50 chance of entering ongoing play in the offensive zone, but as a player plays more shifts throughout a year, the numbers will approach an even result.
As a Penguins fan, a guy like Dominik Simon is a great example as to why Corsi stats as it stands are flawed. Simon often gets defended by some penguins fans due to his corsi stats seeming good despite his lack of scoring, myself included for some time. For this past season according to hockey reference, Simon's Corsi for % was 53.1% which is good on the surface, but he started in the offensive zone 59.3% of the time. If you use the formula I mentioned to adjust the stats as if Simon spent 50% of his time on ice in the offensive and defensive zones, his adjusted Corsi For % drops to a measly 43.75%, indicating that his defensive flaws outweigh his offensive production. However if you apply this same method to Teddy Blueger you find a much different result. Again according to hockey reference, Blueger's Corsi for % was 50.4% despite him only starting in the offensive zone 30.2% of the time. His adjusted numbers using that formula skyrocket him to an adjusted corsi for % of 70.16%, which along with his current data indicates that he is exceptional at providing offense and is one of the best players in the league at limiting chances on defense (due to his relatively low corsi against despite a high defensive zone start%). The corsi stats as is are only as good as how you utilize them to look at a player
been saying this forever. people tell me I'm not "with the times" - old school stats don't tell enough of the story. I see those graphs and crease chance stats and my eyes glaze over. Not because I can't follow them, but because as someone who played the game they are false flags that do not tell the story. PLaying, watching, that's how you can tell the value of things that often aren't measured.
What is the best stat for the defensive game? - like defensive D, penalty killers...
Plus minus is an excellent stat within context of the team and line
I’m pretty sure advanced stats are very useful if used the right way. One thing about shots on goal which is very weird to me is that if you hit the post or crossbar it doesn’t count. Truly unfair, so you at least have to agree that Corsi makes more sense in that situation. 😁
a fourth liner might have decent stats against other 4th liners -----and never touch the puck when he is on against a 1st line crew---stats dont take relative matchups into calculations --for example points per 60 minutes doesnt tell you against whom they are playing
Well of 'corsi', if a player puts pucks in the net at least 20 times a year consistently, and/or gets 30+ assists consistently, of 'corsi' he's a good player, no matter what his corsi-horsy says! And above all, corsi doesn't measure heart.
Yeah but is there a stat that tells me a player's goals per minute while shorthanded in the defensive zone during an overtime period as the home team in a January game during a full moon?
horsey-corsi??? bleeding eyes??? where do I sign up?
On StatsDingus dot com!
Great video! 👍🇸🇪
This Shannon. Exactly this! You watch more hockey than all of us, and being able to watch the players so often is more of a judge of what's really happening.
In the end hockey is a team game, win% is what really matters.
If you want shots that either hit the goalie in the crest or miss, miles wood is your man.
If a player draws a lot of penalties is it because he's weak and falls a lot, or because he's skilled and makes good moves.
Where can we see the P60 stats of each players?
It would be interesting if there actually was a stat good vs bad penalties. I would think that bad penalties are in the majority, usually a penalty is a result of you being overplayed or being out of position.
It depends on what you are looking at, goaltending benefits somewhat from them over sp and gaa, quality of shot against is of value, Use Price as a example, he will often have less then stellar traditional numbers at times, but when you take into account back door plays and deflections you find him top 5 pretty much always, now do these stats tell us any more about Price then when the players rank their peers, idk. For shooters it matters if all the goals they score are scored in games that are already out of reach for the opponent, or if they are protected from the other teams top players , they should excel in that situation. I am not sure how the advanced stats really represent all the factors that matter. I do think they are worth tracking however, its useful when combined with the eye test, It has the advantage of no biases, you either have the numbers or you don't, you just need to look to find the reason why or why not
Hockey is impossible to measure as everything is situational . How many O-zone starts , TOI , what line you play on , How much pp time , Was it a bad line change , how good are your linemates , how good is your team , what style is being played and on & on..... +/- is very useful on one team ( ex: Andreas anthanasiou is so much worse than rest of team ) but it cannot measure team to team accurately . Baseball is the only sport where stats paint an accurate picture (yet pitcher's won lost are useless because of situations-quality of team , do you pitch against other teams best pitchers ?) Basketball is somewhat measurable- they measure performance with 2 or 3 or more , man groups . There are no line changes , bad bounces etc....
A lot of people who are bad at math love advanced stats! Go figure. "I trust that smart stuff."
I had a good laugh watching this one.
I remember when Trouba was being shopped and I said Trouba for Pionk and 1st and that Winnipeg would end up winning the deal. People said Pionk sucked and he wasn't anywhere near Trouba's level based on advanced stats. Well, Pionk has been great for WPG and Heinola looks good as well. Trouba has been good for the Rangers, but the eye test of Pionk was better than the stats showed and GmKC saw this.
Pionk is better analytically though
@@michaeljordan6239 Pionk was last year but not the year prior. But I dont care about the analytics as much as I care about the eye test.
At the end of the day I care more about what I see when a player is on the ice than what the numbers say. I have a better opinion of some fourth line players who get a bunch of PIM for making the top line of the other team miserable than I do of a second line player who gets a point every other game, even if the second line guy is statistically better for the team.