I built a fun model that shows, for every team, which individual player and player-grouping metrics has driven their successes and not-so-successes. The output for UNC MLAX is below as an example. When Kelly and Cameron were in the mix, the Heels tended to have better outcomes last year.
Full post is here.
Lacrosse Analytics
-
- Posts: 1128
- Joined: Wed Aug 29, 2018 3:54 pm
- Contact:
Re: Lacrosse Analytics
Data Engineer/Lacrosse Fan --- Twitter: @laxreference --- Informed fans get Expected Goals, the new daily newsletter from LacrosseReference
-
- Posts: 1128
- Joined: Wed Aug 29, 2018 3:54 pm
- Contact:
Re: Lacrosse Analytics
One of the big missing pieces in lacrosse analytics has always been player efficiency. Teams are starting to move towards using efficiency as the measuring stick for an offense or a defense, but there really isn't a great way to apply efficiency principles to individual players.
So I'm introducing a stat called usage-adjusted-EGA, which takes my traditional expected goals stat (overall production) and adjusts it to account for how many touches a player gets (play share). The result is a metric that tells you what a player was able to accomplish (good or bad) with the opportunity he or she was given.
If you've traditionally used the more obvious measures of production (i.e. points) to help make personnel and roster decisions, adding uaEGA into that decision-making process can have a real impact on overall team efficiency.
That matters.
The full post is here.
So I'm introducing a stat called usage-adjusted-EGA, which takes my traditional expected goals stat (overall production) and adjusts it to account for how many touches a player gets (play share). The result is a metric that tells you what a player was able to accomplish (good or bad) with the opportunity he or she was given.
If you've traditionally used the more obvious measures of production (i.e. points) to help make personnel and roster decisions, adding uaEGA into that decision-making process can have a real impact on overall team efficiency.
That matters.
The full post is here.
Data Engineer/Lacrosse Fan --- Twitter: @laxreference --- Informed fans get Expected Goals, the new daily newsletter from LacrosseReference
-
- Posts: 1128
- Joined: Wed Aug 29, 2018 3:54 pm
- Contact:
Re: Lacrosse Analytics
It only took 5 years, but I finally put together a central listing of all the stats that I calculate.
The idea behind this is to do 4 things:
1. List the stats that I have come up with (or just calculated)
2. Give a quick overview of each (with the ability to learn more easily)
3. Give an indication of the "normal" range for each stat
4. Show which player or team is best
The idea behind this is to do 4 things:
1. List the stats that I have come up with (or just calculated)
2. Give a quick overview of each (with the ability to learn more easily)
3. Give an indication of the "normal" range for each stat
4. Show which player or team is best
Data Engineer/Lacrosse Fan --- Twitter: @laxreference --- Informed fans get Expected Goals, the new daily newsletter from LacrosseReference
-
- Posts: 1128
- Joined: Wed Aug 29, 2018 3:54 pm
- Contact:
Re: Lacrosse Analytics
As I've had more discussions about stats with coaches, the question that comes up, probably more than any other, is whether I keep stats on faceoffs that don't end up making it into the offensive zone. So things like faceoffs that are picked up by the FOGO but turned over before the offense can really get started.
I did the analysis and two things stood out. First, the gap between the best and worst faceoff-converting teams is not that big. The worst team in 2021 converted 88% of their faceoffs into possessions. It's much more akin to clearing/riding rates.
Second, there is really no correlation between a team's faceoff win rate and their faceoff conversion rate. I was expecting that since it's the same unit, you'd see similar rates. But it seems as if it's a different skill.
Regardless, it was an interesting concept. Full post is here.
I did the analysis and two things stood out. First, the gap between the best and worst faceoff-converting teams is not that big. The worst team in 2021 converted 88% of their faceoffs into possessions. It's much more akin to clearing/riding rates.
Second, there is really no correlation between a team's faceoff win rate and their faceoff conversion rate. I was expecting that since it's the same unit, you'd see similar rates. But it seems as if it's a different skill.
Regardless, it was an interesting concept. Full post is here.
Data Engineer/Lacrosse Fan --- Twitter: @laxreference --- Informed fans get Expected Goals, the new daily newsletter from LacrosseReference
-
- Posts: 1231
- Joined: Wed Aug 01, 2018 1:08 pm
Re: Lacrosse Analytics
Surprised that the conversion rate is 88%. I'm sure a lot of fans (myself included) thought it would be a bit lower. It seems that way when your team needs a possession the most. Thank you for the data!!laxreference wrote: ↑Tue Dec 07, 2021 7:08 pm As I've had more discussions about stats with coaches, the question that comes up, probably more than any other, is whether I keep stats on faceoffs that don't end up making it into the offensive zone. So things like faceoffs that are picked up by the FOGO but turned over before the offense can really get started.
I did the analysis and two things stood out. First, the gap between the best and worst faceoff-converting teams is not that big. The worst team in 2021 converted 88% of their faceoffs into possessions. It's much more akin to clearing/riding rates.
Second, there is really no correlation between a team's faceoff win rate and their faceoff conversion rate. I was expecting that since it's the same unit, you'd see similar rates. But it seems as if it's a different skill.
Regardless, it was an interesting concept. Full post is here.
-
- Posts: 1128
- Joined: Wed Aug 29, 2018 3:54 pm
- Contact:
Re: Lacrosse Analytics
Yeah, I'm right there with you. It's like most stuff I'm sure; nobody remembers when everything goes as planned. But when it doesn't, it's memorable.oldbartman wrote: ↑Tue Dec 07, 2021 8:51 pmSurprised that the conversion rate is 88%. I'm sure a lot of fans (myself included) thought it would be a bit lower. It seems that way when your team needs a possession the most. Thank you for the data!!laxreference wrote: ↑Tue Dec 07, 2021 7:08 pm As I've had more discussions about stats with coaches, the question that comes up, probably more than any other, is whether I keep stats on faceoffs that don't end up making it into the offensive zone. So things like faceoffs that are picked up by the FOGO but turned over before the offense can really get started.
I did the analysis and two things stood out. First, the gap between the best and worst faceoff-converting teams is not that big. The worst team in 2021 converted 88% of their faceoffs into possessions. It's much more akin to clearing/riding rates.
Second, there is really no correlation between a team's faceoff win rate and their faceoff conversion rate. I was expecting that since it's the same unit, you'd see similar rates. But it seems as if it's a different skill.
Regardless, it was an interesting concept. Full post is here.
Data Engineer/Lacrosse Fan --- Twitter: @laxreference --- Informed fans get Expected Goals, the new daily newsletter from LacrosseReference
-
- Posts: 1128
- Joined: Wed Aug 29, 2018 3:54 pm
- Contact:
Re: Lacrosse Analytics
We often look at shooting percentages, but uniquely in lacrosse, shots that don't go in are not all the same. A shot that is saved by the goalie is in the same bucket as a turnover. A shot that sails wide generally means another bite at the apple for the offense.
So I wanted to take a quick dive into saved-shot percentages. Just take the number of shots a player takes that are saved and divide by the total number of shots they've taken.
For some context, across D1 MLAX in 2021, there were 30,869 total shots taken. Of those, 29.0% were goals and 30.2% were saved.
As with anything, that obscures what ind. players did. Of the MLAX players with at least 50 shots, these were the lowest saved-shot percentages:
Timmy Ley (BU) - 14.3%
Dylan Hess (GTWN) - 15.1%
Stephen Rehfuss (SYR) - 15.6%
Nickolas Edinger (ARMY) - 16.4%
Reid Bowering (DREX) - 17.3%
Aidan Coll (DREX) - 17.3%
Jonathan Huber (STJ) - 18.2%
Adam Charalambides (RU) - 19.0%
Peter Garno (UVA) - 19.0%
Jack Bowie (NJIT) - 19.2%
Mac OKeefe (PSU) - 19.5%
Jackson Morrill (DEN) - 20.0%
Logan McGOVERN (BRY) - 20.7%
Travis Ford (FFL) - 20.7%
Nicky Petkevich (GTWN) - 20.8%
Matt DeMEO (STNY) - 21.2%
Bubba Fairman (MD) - 21.3%
The interesting thing about saved shot percentages is that they aren't necessarily a good or bad thing. I can have a really low saved-shot percentage if I never put anything on cage. On the flip side, you'd expect higher SOG rates to be correlated with higher saved-shot rates.
As with a lot of things, there is a sweet spot somewhere in the middle. Your ideal shooter is taking shots that have a high chance of becoming goals and a low chance of becoming saves. There is a optimal curve where you maximize goals by finding the right trade-off.
I'll keep an eye on this stat throughout the year to see if that trade-off curve can be firmly established. Best case scenario, it's another tool to put shooting prowess in context.
So I wanted to take a quick dive into saved-shot percentages. Just take the number of shots a player takes that are saved and divide by the total number of shots they've taken.
For some context, across D1 MLAX in 2021, there were 30,869 total shots taken. Of those, 29.0% were goals and 30.2% were saved.
As with anything, that obscures what ind. players did. Of the MLAX players with at least 50 shots, these were the lowest saved-shot percentages:
Timmy Ley (BU) - 14.3%
Dylan Hess (GTWN) - 15.1%
Stephen Rehfuss (SYR) - 15.6%
Nickolas Edinger (ARMY) - 16.4%
Reid Bowering (DREX) - 17.3%
Aidan Coll (DREX) - 17.3%
Jonathan Huber (STJ) - 18.2%
Adam Charalambides (RU) - 19.0%
Peter Garno (UVA) - 19.0%
Jack Bowie (NJIT) - 19.2%
Mac OKeefe (PSU) - 19.5%
Jackson Morrill (DEN) - 20.0%
Logan McGOVERN (BRY) - 20.7%
Travis Ford (FFL) - 20.7%
Nicky Petkevich (GTWN) - 20.8%
Matt DeMEO (STNY) - 21.2%
Bubba Fairman (MD) - 21.3%
The interesting thing about saved shot percentages is that they aren't necessarily a good or bad thing. I can have a really low saved-shot percentage if I never put anything on cage. On the flip side, you'd expect higher SOG rates to be correlated with higher saved-shot rates.
As with a lot of things, there is a sweet spot somewhere in the middle. Your ideal shooter is taking shots that have a high chance of becoming goals and a low chance of becoming saves. There is a optimal curve where you maximize goals by finding the right trade-off.
I'll keep an eye on this stat throughout the year to see if that trade-off curve can be firmly established. Best case scenario, it's another tool to put shooting prowess in context.
Data Engineer/Lacrosse Fan --- Twitter: @laxreference --- Informed fans get Expected Goals, the new daily newsletter from LacrosseReference
Re: Lacrosse Analytics
A priori assumption: “in a way based on theoretical deduction rather than empirical observation.”
“A shot that sails wide generally means another bite at the apple for the offense.“ What % of shots don’t have a man back? Taken at the end of the shot clock? Don’t even make it to the crease?
“There is a optimal curve where you maximize goals by finding the right trade-off.“ Is there? How do you know? This is Boolean?-
“I can have a really low saved-shot percentage if I never put anything on cage.” No. A saved-shot has to be at least some what on cage for the goalie to touch it for it to become a saved shot. From your statistics 40.8% of shots were not even close enough for a goalie save much less a shot on goal than the goalie did not touch (pipe) or a goal. Oh, and if I score on all of my shots, the saved shot % is low too? Right?
“A shot that sails wide generally means another bite at the apple for the offense.“ What % of shots don’t have a man back? Taken at the end of the shot clock? Don’t even make it to the crease?
“There is a optimal curve where you maximize goals by finding the right trade-off.“ Is there? How do you know? This is Boolean?-
“I can have a really low saved-shot percentage if I never put anything on cage.” No. A saved-shot has to be at least some what on cage for the goalie to touch it for it to become a saved shot. From your statistics 40.8% of shots were not even close enough for a goalie save much less a shot on goal than the goalie did not touch (pipe) or a goal. Oh, and if I score on all of my shots, the saved shot % is low too? Right?
-
- Posts: 1128
- Joined: Wed Aug 29, 2018 3:54 pm
- Contact:
Re: Lacrosse Analytics
Virginia is playing 14 games this year. Line is 10.5 wins prior to the NCAA tournament should they make it in. Think they'll go over or under?
Here's the schedule:
Air Force - Feb 5
High Point - Feb 13
Towson - Feb 19
Syracuse - Feb 26
Johns Hopkins - Mar 5
North Carolina - Mar 12
Maryland - Mar 19
Notre Dame - Mar 26
Richmond - Apr 2
North Carolina - Apr 9
Duke - Apr 14
Quinnipiac - Apr 16
Syracuse - Apr 23
Lafayette - Apr 28
Here's the schedule:
Air Force - Feb 5
High Point - Feb 13
Towson - Feb 19
Syracuse - Feb 26
Johns Hopkins - Mar 5
North Carolina - Mar 12
Maryland - Mar 19
Notre Dame - Mar 26
Richmond - Apr 2
North Carolina - Apr 9
Duke - Apr 14
Quinnipiac - Apr 16
Syracuse - Apr 23
Lafayette - Apr 28
Data Engineer/Lacrosse Fan --- Twitter: @laxreference --- Informed fans get Expected Goals, the new daily newsletter from LacrosseReference
-
- Posts: 1128
- Joined: Wed Aug 29, 2018 3:54 pm
- Contact:
Re: Lacrosse Analytics
All ACC teams play two of the other ACC teams twice and the other two once. UVA has Cuse and UNC twice this year. That's in-lieu of having a conference tournament.
Data Engineer/Lacrosse Fan --- Twitter: @laxreference --- Informed fans get Expected Goals, the new daily newsletter from LacrosseReference
-
- Posts: 1128
- Joined: Wed Aug 29, 2018 3:54 pm
- Contact:
Re: Lacrosse Analytics
Tricky, they've got 3 losses to work with to hit the over. If you told me they would sweep Cuse, I'm probably with you. Lose one of the early games and the path to 11 wins gets a lot murkier though.
Data Engineer/Lacrosse Fan --- Twitter: @laxreference --- Informed fans get Expected Goals, the new daily newsletter from LacrosseReference
-
- Posts: 1128
- Joined: Wed Aug 29, 2018 3:54 pm
- Contact:
Re: Lacrosse Analytics
Data Engineer/Lacrosse Fan --- Twitter: @laxreference --- Informed fans get Expected Goals, the new daily newsletter from LacrosseReference
-
- Posts: 1128
- Joined: Wed Aug 29, 2018 3:54 pm
- Contact:
Re: Lacrosse Analytics
I'm running a Win Totals Contest this season. 100% free to play. Right now, you've got the options to pick individual games each week OR just play the Bracket Challenge for 3 weeks at the end of the season. I thought that it would be good to add a middle option: pick whether a bunch of teams will go over or under a certain win total and then follow along all year to see where your entry ranks.
I wrote a bit about how it will work and why I thought it should exist here.
I wrote a bit about how it will work and why I thought it should exist here.
Data Engineer/Lacrosse Fan --- Twitter: @laxreference --- Informed fans get Expected Goals, the new daily newsletter from LacrosseReference
-
- Posts: 1128
- Joined: Wed Aug 29, 2018 3:54 pm
- Contact:
Re: Lacrosse Analytics
Duke looked good. The 34.4% offensive efficiency just edged out the mark they put up against RMU last year in their first game (32%). The difference is that the Blue Devils never made us nervous that the game was ever actually in doubt. In last year's matchup, RMU actually had a win probability above 50% with about 4 minutes left in the first half.
This year? No such luck. With a 10-3 first quarter, this game was over before it really ever got started. And you probably credit the Duke defense for that. Last year, they gave up 29.3% efficiency day to RMU. Tonight? 24.5%. And if you want to throw the ride in there too (9 failed clears), the defense never gave the Colonials a chance to put up a fight.
If you are on the Duke NCAA Champs 2022 train, I doubt you saw anything tonight to make you wonder whether you want to keep your seat.
This year? No such luck. With a 10-3 first quarter, this game was over before it really ever got started. And you probably credit the Duke defense for that. Last year, they gave up 29.3% efficiency day to RMU. Tonight? 24.5%. And if you want to throw the ride in there too (9 failed clears), the defense never gave the Colonials a chance to put up a fight.
If you are on the Duke NCAA Champs 2022 train, I doubt you saw anything tonight to make you wonder whether you want to keep your seat.
Data Engineer/Lacrosse Fan --- Twitter: @laxreference --- Informed fans get Expected Goals, the new daily newsletter from LacrosseReference
Re: Lacrosse Analytics
Good analysis. On the defense vs ride how does that work into the number? Thought the ride was strong (and/or ROMO poor on clears) but 6 on 6 defense less strong. Could also be a factor of ROMO missing some clear dunks.laxreference wrote: ↑Fri Feb 04, 2022 8:29 pm Duke looked good. The 34.4% offensive efficiency just edged out the mark they put up against RMU last year in their first game (32%). The difference is that the Blue Devils never made us nervous that the game was ever actually in doubt. In last year's matchup, RMU actually had a win probability above 50% with about 4 minutes left in the first half.
This year? No such luck. With a 10-3 first quarter, this game was over before it really ever got started. And you probably credit the Duke defense for that. Last year, they gave up 29.3% efficiency day to RMU. Tonight? 24.5%. And if you want to throw the ride in there too (9 failed clears), the defense never gave the Colonials a chance to put up a fight.
If you are on the Duke NCAA Champs 2022 train, I doubt you saw anything tonight to make you wonder whether you want to keep your seat.
-
- Posts: 1128
- Joined: Wed Aug 29, 2018 3:54 pm
- Contact:
Re: Lacrosse Analytics
Failed clears do not count against an offense (or for a defense) in terms of efficiency. The denominator in the efficiency calculation is (total times they gained possession in the play by play - failed clears). So no, the 24.5% efficiency for the Duke defense is not affected by the number of failed clears for ROMO.rolldodge wrote: ↑Fri Feb 04, 2022 8:50 pmGood analysis. On the defense vs ride how does that work into the number? Thought the ride was strong (and/or ROMO poor on clears) but 6 on 6 defense less strong. Could also be a factor of ROMO missing some clear dunks.laxreference wrote: ↑Fri Feb 04, 2022 8:29 pm Duke looked good. The 34.4% offensive efficiency just edged out the mark they put up against RMU last year in their first game (32%). The difference is that the Blue Devils never made us nervous that the game was ever actually in doubt. In last year's matchup, RMU actually had a win probability above 50% with about 4 minutes left in the first half.
This year? No such luck. With a 10-3 first quarter, this game was over before it really ever got started. And you probably credit the Duke defense for that. Last year, they gave up 29.3% efficiency day to RMU. Tonight? 24.5%. And if you want to throw the ride in there too (9 failed clears), the defense never gave the Colonials a chance to put up a fight.
If you are on the Duke NCAA Champs 2022 train, I doubt you saw anything tonight to make you wonder whether you want to keep your seat.
Data Engineer/Lacrosse Fan --- Twitter: @laxreference --- Informed fans get Expected Goals, the new daily newsletter from LacrosseReference
Re: Lacrosse Analytics
What happens when a team clear the ball over the mid-line and then gets stripped of the ball before getting it to the attack? With hard riding teams typically riding well into the offensive zone, I’m guessing there is grey area around a “successful clear” and a turnover in the offensive end.laxreference wrote: ↑Sat Feb 05, 2022 5:58 amFailed clears do not count against an offense (or for a defense) in terms of efficiency. The denominator in the efficiency calculation is (total times they gained possession in the play by play - failed clears). So no, the 24.5% efficiency for the Duke defense is not affected by the number of failed clears for ROMO.rolldodge wrote: ↑Fri Feb 04, 2022 8:50 pmGood analysis. On the defense vs ride how does that work into the number? Thought the ride was strong (and/or ROMO poor on clears) but 6 on 6 defense less strong. Could also be a factor of ROMO missing some clear dunks.laxreference wrote: ↑Fri Feb 04, 2022 8:29 pm Duke looked good. The 34.4% offensive efficiency just edged out the mark they put up against RMU last year in their first game (32%). The difference is that the Blue Devils never made us nervous that the game was ever actually in doubt. In last year's matchup, RMU actually had a win probability above 50% with about 4 minutes left in the first half.
This year? No such luck. With a 10-3 first quarter, this game was over before it really ever got started. And you probably credit the Duke defense for that. Last year, they gave up 29.3% efficiency day to RMU. Tonight? 24.5%. And if you want to throw the ride in there too (9 failed clears), the defense never gave the Colonials a chance to put up a fight.
If you are on the Duke NCAA Champs 2022 train, I doubt you saw anything tonight to make you wonder whether you want to keep your seat.
Re: Lacrosse Analytics
Thank you for your work, find your analytics very interesting despite my lean to subjective "eye test". The best coaches today incorporate both. Se. When coaching , I wanted the the explosive X factor who may measure as less efficient but can 'play up' to any level of completion. seen some talents play down to inferior opponents with efficient players ruling day only to see same X factor raise and assert themselves agaisnt superior opponents which their more efficient teammates cant do. Have noted that opponents views, from coaches and players when available tend to basically reinforce my assessments of own.. Who my opponent prepares for tells much.laxreference wrote: ↑Tue Nov 02, 2021 9:52 am One of the big missing pieces in lacrosse analytics has always been player efficiency. Teams are starting to move towards using efficiency as the measuring stick for an offense or a defense, but there really isn't a great way to apply efficiency principles to individual players.
So I'm introducing a stat called usage-adjusted-EGA, which takes my traditional expected goals stat (overall production) and adjusts it to account for how many touches a player gets (play share). The result is a metric that tells you what a player was able to accomplish (good or bad) with the opportunity he or she was given.
If you've traditionally used the more obvious measures of production (i.e. points) to help make personnel and roster decisions, adding uaEGA into that decision-making process can have a real impact on overall team efficiency.
That matters.
The full post is here.