How accurate or predicative is PDO?
Id like to start a discussion about understanding what these analytics are all about. If possible, lets try and keep this so that the noobs (like myself) can understand what the hell Corsi is all about. So with that being said, what is Corsi? How does it relate to fantasy hockey?
What should i be looking for in analytics to determine whom i should target for trade?
How can i use the new Player Usage Chart to my advantage? I have tons of questions so let this discussion begin!
WHL - World Hockey League (24 Team - Daily H2H)
Maine Moose 2021-2022
2019 WHL GM of the Year #Supersoft
How accurate or predicative is PDO?
WHL - World Hockey League (24 Team - Daily H2H)
Maine Moose 2021-2022
2019 WHL GM of the Year #Supersoft
The basis of most hockey analytics stems from a recent (sort of obvious) realization.
i) Teams need goals to win games
ii) Goals come from shots
iii) A certain % of shots typically (go in the net) score goals
iv) Therefore, more shots = more goals... less shots = less goals
Simple, right!
Yes - it is! It really is!
[And if you look at the last 10 years of stats, you'll see that SHOTS per game is on the rise! Teams are really pushing to do anything to get shots.]
Jim Corsi recognized (& pushed) the plus-minus value of shots (and not so much "goals", which incorporates goalie quality for-or-against).
Coaches & GMs can institute player combinations or tactical strategies to improve SHOTS FOR or minimize SHOTS AGAINST.
[Goalies are left to their own world]
So - basically, almost all key hockey analytics start from this one realization - that teams should really try to outSHOOT their opponent... by as much as possible.
How do we OUTSHOOT our opponents, because that should lead to us OUTSCORING our opponents, and thus, winning!
Most player charts & advanced stats are looking at how players perform in different situations (5-on-4, 5-on-5) with different teammates and different situations (up a goal, down a goal).
Some stats look at possession - since possession leads to shots for... or against.
So "zone starts" and "zone entries" and "zone exits" are also stats that look at POSSESSION.
Again... because POSSESSION (or Dispossessing the other team) leads back to SHOTS.
Good Offensive players should be able to outshoot opponents through generating a surplus of shots, vs. average shots against, per minute.
Good Defensive players should be able to limit opponent shots, creating advantage if their team generates an average shots for, per minute.
Easy, peezy.
SHOTS!
It's all about SHOTS!
Thanks Peng. I was reading somewhere that all shots are not equal though. A shot from the blue line with no traffic in front is as good as a giveaway. How does this jive with Corsi? Or maybe it doesn’t?
Im trying to figure out if this plays into Fantasy hockey at all.
WHL - World Hockey League (24 Team - Daily H2H)
Maine Moose 2021-2022
2019 WHL GM of the Year #Supersoft
^Well, yes. All shots are not equal.
But separating shot quality does murky up most hockey analytics.
There is "dangerous area" shots, I believe.
The overlying point is that most teams have similar distributions of shot (ranges).
Most wingers might have similar distribution of shot (ranges).
Most defensemen might have similar distribution of shot (ranges).
So... typically - shot quality averages out when looking at several shots (whether from one team or one player, over a reasonable range of games)
IIRC, forwards typically shoot about 12% per year.
Most defenseman typically shoot about 5% per year.
I think average, overall, is about 10% shooting.
Typical hockey game over past 20 years has been a 30-shot night (per team) with 3-goals (per team).
Goalie sv% = 27/30 = 9/10 = 0.900
So goalies that do this are NHL-OK.
[FWIW - this is why I set WHL scoring at 0.1p/SV and -1/GAA. Goalies actually need 10 saves per goal against, .909sv%, to level out scoring on facing shots.]
Yeah - only the eye test will reveal that some forwards fling useless shots on net that aren't high-danger shots.
Michael Frolik is one I've noticed over the years.
He flings garbage on net from all over, and that's why his shooting rate (8%) over his career is one of the lower forward shooting percentages of veteran forwards.
[*Interestingly! Frolik is actually shooting an unsustainable 21% this year... like, WTF!... in 17gp, he has 7g... and... ZERO assists. A bizzarro world year for Frolik!]
I've found, though, that it is easier to weed-out that "crap shooter" from the typical forwards - who do shoot closer to 12%.
Dobber does a good job explaining them in his intro to his Fantasy Guide and how he uses them in his guide. I dont know if you purchase the guide or not (I recommend it if you do not) but there is lots of depth and it's explained nicely for regular folk to understand it.
From the Guide:
CF% (Corsi-For %) - Corsi is defined in the above NHL.com link - they call Corsi ‗SAT‘. This takes the total shot
attempts by a player‘s team when he is on the ice… and divides it by the total shot attempts when he is on the ice (both for
and against) and multiplies this by 100. So, a number more than 50 indicates that he drives possession.
SPSV% (PDO) - I get into this more in the article below, but this is a good statistic to quickly draw your attention to
any player with an abundance of luck (good or bad). It takes the 5on5 SH%+5on5 SV% and how far it deviates from 1000
tells you if he was the victim (or beneficiary) of some lucky bounces.
10 Team, Points Only, Cash League
25 Man Roster (no position), top 20 point getters count at end of month
Keep 20/25 at seasons end, Cut 5 to FA for redrafting
Goalie points W=2pt L=-1pt SHO=2pt
Stamkos, Tavares, Eichel, Mercer, JRobertson, RThomas, Kucherov, Nugent-Hopkins, Tuch, KConnor, Necas, Point, Konecny, SJarvis, Cozenz, Morrissey, Bouchard, Josi, Novak, Tolvanen, Peterka, Brink
G- Vasilevskiy, Sorokin, Oettinger
"Cleavage is like the sun. You can look, but dont stare.. Unless you're wearing sunglasses."
PDO is pretty much useless for individual players (unless you're trying to figure out if their plus-minus is going to improve). PDO takes both SH% and SV% but for fantasy hockey purposes, you only care about one of those stats: SH%. So just ignore PDO entirely and focus on 5on5 SH% instead. That's the best advice I can give to anyone because most people make the mistake of following PDO blindly without realizing how useless it is for players (just to clarify, it's a good stat to predict team success but terrible for individual players).
I have been in your very shoes. I wish I had this community to come to to get awesome answers when I was trying to learn.
However, I found a book that uses real life examples and goes in depth to explain it. Very interesting and informative read. Of course, it is not the easiest book to read as it is partially like a university math text-book. But if you are willing to study and learn it is very helpful tool.
http://www.hockeyabstract.com/statshot
Is PDO useful? Is it even real? Since there is no relationship between skater's ability to shoot and a goalie's ability to save (on the same team), there isn't really any reason for PDO to be 1000. Unless I've been misunderstanding what people have been trying to sell as PDO all these years. I guess it could useful if you had a team-specific PDO baseline that you were measuring it against.
Also on the subject of shots, stats people (as seen in articles here) are getting better at identifying which players are taking more shots from more dangerous areas.
For me, what I find most useful are individual and team SH% (particularly 5v5) for identifying whether current production is sustainable or not; and ATOI and PPTOI (and %PP) for telling me if the opportunity is there.
Not all shots are equal. That's where analytics are going. It's too simple to say shots are everything. There's players in the NHL right now playing as though the goal of hockey is to win the shot attempt battle.
I really like the work Steve Valiquette has done in recent years. He is in the process of gathering data to evaluate the probability of different plays resulting in goals. Breakaways (obviously) are great, as are double screens (shooter's teammate plus a defender) and plays where the goalie has to move across the middle of his net (ie he's not set in his stance).
That said, shot attempts still do matter. Even if some plays are low probability you can still generate goals through volume. But there's a reason why one player can score more than another and it's not just shot attempts.
For those with access to The Athletic: https://theathletic.com/626729/2018/...on-in-the-nhl/
Contact me for Frozen Tools bug reports and inquiries
Follow Frozen Tools on Twitter @FrozenTools
Follow me on Twitter @DH_EricDaoust
HTH Daily League, 8 teams, (goals, ppp, +/-, assists, SOG, W, GAA, SV%) keepers in Bold.
C (s3) EP40, Trocheck, Marchessault LW, Horvat
LW (s3) Johnny Hockey, Parise, Terivinen RW, Meier RW
RW(s3) Wheeler C, Taresenko, Atkinson, Gallagher
Until(s1)
D (s4) Carlson, Yandle, Jones, Trouba
G Dubnyk, Rittich, Markstrom, Binnington
IR Ehlers, Hall