Sell me on Analytics... then sell me on ours

[Pookie]

Leafs have invested a lot of money in an Analytics department. This includes personnel and technology.

Dubas has gone on speaking tours about how the game has changed and how the Leafs are at the forefront.

The stat that I can't figure out is why they only outscored the Habs in 1 of 11 periods over the last 3 games.

Wouldn't the analysts have some idea as to why that was happening? And how to fix it?

If not... what value do they have?

So here's my challenge:

1. convince me that investing in analytics are important.
2. If so, convince me that the current leader of this analytics movement is good at his job. What other resources does he need to make analytics useful?

Click to read more...

 

[TMLegend]

My favorite analytical tool is the score at the end of the game. Always tells you who won/lost the game. I'm really tired of the xGF stat being thrown around like its gospel.

 

[TheDoldrums]

Every team in the league has some sort of analytical presence.

Sell me on ignoring data.

 

[keonsbitterness]

For some reason, the 1 in 100 scenarios that analytic models show us losing keep happening to us. Very strange.

Explanations, please?

 

[Pookie]

Sell me on ignoring data

easy if the data isn't correlated to winning, it's just noise.

if the data is important... why couldn't we use it to close out the series?

 

[93LEAFS]

Leafs have invested a lot of money in an Analytics department. This includes personnel and technology.

Dubas has gone on speaking tours about how the game has changed and how the Leafs are at the forefront.

The stat that I can't figure out is why they only outscored the Habs in 1 of 11 periods over the last 3 games.

Wouldn't the analysts have some idea as to why that was happening? And how to fix it?

If not... what value do they have?

So here's my challenge:

1. convince me that investing in analytics are important.
2. If so, convince me that the current leader of this analytics movement is good at his job. What other resources does he need to make analytics useful?

I have more questions about the person interpreting the data than its actual usage. Data is only as good as the person analyzing it.

 

[TheDoldrums]

easy if the data isn't correlated to winning, it's just noise.

if the data is important... why couldn't we use it to close out the series?

What if I told you other teams also use data. We don't have a monopoly on analytics, far from it.

We probably have more scouts than the vast majority of teams and I don't think our lack of success means scouting is bad. We have the resources to have top of the line departments in every area.

 

[Bomber0104]

1) Just because something can be measured doesn't mean it needs to be.

2) Just because something can be measured doesn't by default mean we should value or put significance in the measurement.

 

[hamzarocks]

Sell me on ignoring data

Ignore data when it's leading to poor conclusions. such as Mitch Marner being a 10.9M player, Sheldon Keefe being a top 5 coach in the league (people are spouting this due to his extremely exaggerate win% so far in his career)

Ignore data that is used to help support and serve as evidence for a narrative. Fans have used different variations of scoring chances and possession stats since 2019 to paint the picture the leaves are nearly there and are losing to bad luck/goaltending in the playoffs. People have ignored results and said the underlying information is more relevant and will lead to a reversing back to average eventually. They ignore that this isn't the case come playoffs which are a small sample, with different teams year to year who have different coaches/players and strategies being deployed from one series to the next.

Advanced stats are useful but they shouldn't be the calling card of our GM/management team. Dubas ability to build a team, draft/scout, negotiate and develop a team structure with a strong culture is more relevant and important than his ability to hire people who can gather, utilize and interpret the advanced stats to form a picture that can be used as justifications to support his decisions as to the GM

I'm not saying he is good or bad at the stuff I mentioned above, I'm just saying that people need to stop using advanced stats and associating them with a strong team, a team that can win, and a team that is developing/thriving. The team is more than an association of statistics and it can succeed in a series even if they have poor advanced stats (like the habs just did, though that type of play probably wont be sustainable enough to win a cup unless we have a 2018 Caps type run)

 

[koyvoo]

They are a tremendous tool to aid in valuations, but they are an aid, not the number 1 tool. Many times that directly conflict or even refute what is visually apparent.

The issue with us, is u think our GM is like many of our posters, where either they follow the analytics without watching games. Or, he does watch the games, but doesn’t understand the nuances of what he sees on the ice nearly as much as he understands graphs and charts. We have a lot posters with the very same issue.

 

[TheDoldrums]

Just one more point. "Analytics" have been pretty clear that Colorado is the best team in the league this year and they've looked great in the playoffs. They also use data to drive their decisions. The Devon Toews trade was immediately praised by the nerds for example.

This notion to call us the "Analytics" team (while making several moves analytics people hated) and pretend the rest of the league is a scouting team or whatever is just lazy and bad. If we were all in on data we would never have valued Foligno higher than Hall and would not have emphasized leadership in our depth to that extent. There was no analytical case for a guy like Wayne Simmonds, for example.

 

[egd27]

1) Just because something can be measured doesn't mean it needs to be.

2) Just because something can be measured doesn't by default mean we should value or put significance in the measurement.

I've done dozens of development projects where the stakeholders ask for all sorts of elaborate reports. My question to them is always the same, "What decisions or changes will the data enable you to make?"

Usually cuts the requests in about half.

 

[Jozay]

Sell you on analytics? Pick up a book dude, idk. If you got a brain, you'd realize why they're valuable and why its nice to have . Theres nothing to sell you on.

Who knows what the Leafs are looking at. They brought in a lot of guys that no analytically driven team would bring in. And as expected, it blew up in their face.

 

[Aus34Matt]

1) Just because something can be measured doesn't mean it needs to be.

2) Just because something can be measured doesn't by default mean we should value or put significance in the measurement.

But it's the geek thing to do. In case you didn't know, nerds now rule the world and this board is full of wanna-be's.

 

[Menzinger]

Colorado, the Hurricanes and Tampa have invested in them and they have likely some of the best front offices in the league.

Analytics are just a tool to help decision making, it's not a philosophy or a specific playing style of the game. too many people make broad assumptions about their use in sports (and especially hockey) which leads to a lot of straw men type arguments that get thrown around particularly against their use.

 

[Pookie]

What if I told you other teams also use data. We don't have a monopoly on analytics, far from it.

We probably have more scouts than the vast majority of teams and I don't think our lack of success means scouting is bad. We have the resources to have top of the line departments in every area.

Ok.. so if we agree that Data is important.

Do we have the right leader of the data department? I'm not seeing results.

 

[JT AM da real deal]

Even da great Dubie himself stated at Leaf training camp 2 years ago in Niagara when speaking to Brock sports mgmt class that analytics was a new field and he thought 30% of game was still based on pure luck were his words exactly .. he also said he expected da field to change significantly over coming years as we learn more about what really matters .. I came away, after over hearing it, thinking even he thought it was nowhere near close to explaining much but a tool to use to evaluate players and team play .. bottom line every team has different requirements in players based on a lot of factors and da true test of evaluating players comes right down to a mix of pure fundamental skills and hockey IQ .. analytics seems to miss both as it looks at things on a combined basis .. and it depends on whose hockey IQ you like and whose you don't prefer .. bottom line this is not a science it is pure artform at this stage in its development

 

[larueskee]

I believe in some analytics and not others so I can't sell you anything. What I will tell you is that there are great players that lead teams to wins and there are great players that are not great leaders. Central Scouting doesn't or hasn't figured that out yet as has the rest of the league. Just look at the first overall picks since Eric Johnson was drafted and see how many have led their teams to greatness.

 

[Gallagbi]

Ok.. so if we agree that Data is important.

Do we have the right leader of the data department? I'm not seeing results.

We're also not seeing the data, process, information or suggestion in fairness.

We have access to basic "Advanced" stats on the internet. NHL teams, and not just Toronto, are paying for more than what they can get online.

 

[Gabriel426]

The only stats that matters is the W.
Even in Moneyball, the key is to get more W and in order to get more W the team needs to score more Runs, Runs are generated by getting on base, getting a hit, a walk, hit by pitch, reach on error....is the same as long as the batter can get on base.
Using this concept, in order to get more W, it needs more goals, goals are generated by shots, whether someone deke pass 2 defenders or slap shot from the point, or shot through traffic, or tip in, or banging on rebounds...

 

[Menzinger]

We're also not seeing the data, process, information or suggestion in fairness.

We have access to basic "Advanced" stats on the internet. NHL teams, and not just Toronto, are paying for more than what they can get online.

And quite frankly the how\when\where they are being used as well. The way people act is like Dubas' or Keefe walk around with a calculator to decide every micro decision they make, which also obviously is not the case either

 

[Gallagbi]

And quite frankly the how\when\where they are being used as well. The way people act is like Dubas' or Keefe walk around with a calculator to decide every micro decision they make, which also obviously is not the case either

It's interesting, part of me wonders how much they regret moving away from their process this season. We added some obviously different players this year - ending the season with Nash, Foligno, Thornton, Simmonds, Bogosian on this roster. Outside of Bogo, who was great, I don't think any of them really helped the team. You have guys who were given first line opportunities down through healthy scratches.

I wonder if they went too much to be like the teams who beat them rather than focusing on what they need to do to succeed and break through. We heard comments about "nothing happening when Nash was on the ice" or the defensive responsibility in a guy like Foligno shutting down opposition. Truthfully there wasn't anyone to shutdown on the Habs and those type of players didn't end up bringing any value. At the same time, neither did our elite players outside Nylander

 

[justashadowof]

Analytics were at best a boutique scouting enhancement tool that for the most part have completely played out. Just think 3-4 years ago there were mobs of hockey fans who bought into this as a universal panacea for team building. Where are they now? And the scams adjacent to it have run their course.

 

[Stephen]

Analytics are a good supplement to a management and coaching staff that is well versed in hockey culture and playoff hockey to gain the edge.

But I often wonder if we’re seeing a Leafs game plan built to generate stats totals like number of passes attempted and % completed, total time with puck on stick and clock time in the offensive zone which results in good metrics while actually not particularly dangerous.

Sometimes you just need a chip and chase down the boards and a hard board battle and recovery but we curl back for the extra 3 steam boats and complete 2 extra touches of the puck which looks good on paper until that puck is turned over with momentum going the other way.

 

[Prominence]

Key things to note before reading:
- We have a s*** core. When a line that consists of 33 mil can't score during the playoffs, you will lose
- Muzzin and Tavares were injured, two guys who generate great underlying numbers, so losing them hurts quite a LOT
- We did not have the edge in terms of analytics once Tavares went down. The average value of the lines, with the exception of the top line, was actually lower than the average value of the habs lines or extremely close. Habs defense is also very good.

Leafs have invested a lot of money in an Analytics department. This includes personnel and technology.

Dubas has gone on speaking tours about how the game has changed and how the Leafs are at the forefront.

The stat that I can't figure out is why they only outscored the Habs in 1 of 11 periods over the last 3 games.

Wouldn't the analysts have some idea as to why that was happening? And how to fix it?
We didn't generate many good offensive opportunities. Most of the shots generated were easy saves that Price could make. The leafs also had a hard time getting to the net. Our pp was dogs*** as well.

If not... what value do they have?

So here's my challenge:

1. convince me that investing in analytics are important.
2. If so, convince me that the current leader of this analytics movement is good at his job. What other resources does he need to make analytics useful?

1. Let's look at the teams that barely or don't use analytics. Do you really want to be the Buffalo Sabres? How about Peter Chiarelli's Oilers? Brian Burke's leafs? The Nonis era leafs?

It's been proven that the organizations that use analytics well have gone on to win championships. Bill James, the guy who wrote sabremetrics, has won championships with the Red Sox. The blackhawks, kings have been noted to extensively use analytics during their cup runs. In basketball, even on the nba subreddit, every fan and manager embraces analytics (they look at true shooting percentage, lots of fans like rapm).

The reason why you want to pour more money into analytics is because it is also an ongoing process. Note the drastic decline of the blackhawks. The GM traded Hjalmarsson, Byfuglien, Nick Schmaltz, Saad, then traded panarin for saad away. Most of these guys, if not all, are analytics darlings. One of the best moves Chicago has done, from an analytics perspective, was draft debrincat. Since Kane's 3rd championship, chicago has made a lot of mistakes. My point is that there's positive correlation between positive results and the use of analytics so when a team decides to completely ignore the use of analytics, their decision-making absolutely falters and losing happens.

Analytics is a tool used, alongside the eye test, to help answer the question, "how do i build the best team possible or a championship winning team?" Dubas, from the looks of it, is using the RAPM model (me and him think alike except for the tavares signing, marner contract (i wanted to starve him), matthews contract. From his point of view, his idea of winning is to limit high danger scoring chances against the leafs, generate great opportunities against other teams and transitions (moving out of defensive zone or skating into the offensive zone). Therefore he looks for players who can do all that and in order to do that, you need metrics. For example, Dubas needed to find a dman who could play in the top 4, but could also limit scoring opportunities against the leafs so he traded for Muzzin. Dubas signed Jason Spezza (don't you dare talk s*** about Spezza) because he's good at transitions. Analytics has data for this and without this data, these signings wouldn't be possible.

2. I don't know if I could say he's good at his job, but there's not really a gm in the league I would want to replace Dubas. I don't think he should be fired though. Dubas' mistakes are actually fixable. Marner can be moved. Our prospect cupboard is pretty decent and can supplant any roster inefficiencies. He has done good things and his good things outweigh his bad things. He has used analytics to make great decisions in Muzzin (although injury history concerns me), Campbell (he wasn't good in game 7, but the leafs did ONLY score 1 goal in total), Cody Ceci was a decent 3rd pairing dman and Penguins fans love him, developed an ECHLer into a 2-3rd pairing dman in Holl, signed Spezza for cheap, established a good reputation with european free agents, somehow got a 15th overall pick for kapanen, signed TJ brodie. Vesey was a dumb signing that I was able to see a mile away. One area that dubas has done well in would be drafting and amateur scouting. Sandin's already seeing NHL games, Robertson is very close and if you have followed the 2020 draft, the leafs did VERY WELL. The draft alone is why dubas is good at his job. Robertson works hard. Niemela and ovechkinov play with effort. You're going to LOVE ryan tverberg.

What other resources does he need to make analytics useful? Experience and not get fired.

Moving forward, I think the leafs will most likely keep making the playoffs, perhaps make it to the 2nd round and then flame out until matthews and marner (the hometown kid) leave. I believe the players that dubas drafts will make the new and better core of the leafs and its only until then that the leafs win a cup.