Vorp vs Per vs TS

[jasonxxx102]

just use DARKO or EPM or RAPM. LEBRON is fine, RAPTOR is garbage, and the other AIO metrics are objectively worse. Why use something that is worse when you can get the better version easily?

[FluLikeSymptoms]

Feels like I’ve been crapping on PER for 20 years. Must be close.

[Jadoogar]

True shooting isn't like the other 2. It's not an all encompassing stat like PER, it just a better measure of efficiency than FG% because it accounts for 3s and free throws.

[maxpower8888]

"I like this stat that has LeBron over Jordan. All those other stats are bull."

[dhsilv2]

What is Vorp good for? I never really get these things. Like I get what it is, it just seems a bit arbitrary to judge players by someone directly replacing them when it's some defender and it's probably the last few minutes of garbage time. At least the other stats are actually trying to judge a player on their own merits.

https://www.basketball-reference.com/leaders/vorp_yearly.html

Imagine you judged MVP every year using it, Dirk, Giannis, Nash, Iverson, Hakeem, Moses, Walton all no MVPs.

Well you did list a couple bad MVP choices...

[dhsilv2]

Ts% only measures scoring ability VORP and PER are trying to encompass the entire stat sheet. All of them have flaws, including every advanced stats you can come up with.

Exactly, it is simple minded to just use one stat to evaluate players. Several advanced stats would say that Manu Ginobili was better than Kobe Bryant but is that really the truth? They are evaluating tools, and each advanced stat need to be put into context before just saying see this guy is better than this other guy.

There's a great debate to be had if Manu was per minute as good or better than Kobe. Kobe however played a lot more minutes.

[dhsilv2]

TS% (plus assisted percentage ) is the only one that does a good job when comparing players IMO. PER a black box hidden formula. Probably tells you how bad a player is more than how good they are. VORP is essentially plus minus which has obvious flaws.

Per is dependent on how good the avg player is in a given season. It is why you can’t compare pers season to season. It is only useful for that given season and even at that it has its shortcomings. It doesn’t properly evaluate defensive measures and leans more to offensive stats which are easier to quantify.

What makes PER great is that it adjusts to league average. That is why PER WORKS when comparing players from different seasons. If it didn't, THEN you couldn't compare players season to season. For example you can't compare points per game in different seasons. You can however compare a 30 PER to different seasons.

[JonFromVA]

TS% is the best of the bunch because it only tries to measure one thing (shooting efficiency). It uses an estimate for free-throws, but it's a smaller leap than the other stats make and hence less prone to error.

[nikster]

TS% is the best of the bunch because it only tries to measure one thing (shooting efficiency). It uses an estimate for free-throws, but it's a smaller leap than the other stats make and hence less prone to error.

Exactly. TS% does a great job of what's its supposed to be measuring. Can't say the same for a lot of other stats.

[JujitsuFlip]

As many pointed out, no one is using TS% to compare players across eras or for their whole careers, doesn't even make sense.

Also, who cares what some columnist prefers? Dude has a Juris Doctorate, he should just go be a lawyer lol

He has become known for his deep dives on the CBA and salary cap, he's not a mathematician or anything of the sort.

[Wooderson]

TS% (plus assisted percentage ) is the only one that does a good job when comparing players IMO. PER a black box hidden formula. Probably tells you how bad a player is more than how good they are. VORP is essentially plus minus which has obvious flaws.

VORP (or BPM) don't use individual +/- or lineup data, it's box score based with a team component.. The formula is derived/indexed off long term +/- related numbers, but for individual players it doesn't care about their +/-.

[dhsilv2]

TS% is the best of the bunch because it only tries to measure one thing (shooting efficiency). It uses an estimate for free-throws, but it's a smaller leap than the other stats make and hence less prone to error.

Exactly. TS% does a great job of what's its supposed to be measuring. Can't say the same for a lot of other stats.

This pretty much. I mean, I like BPM well enough, but it's limited and far from exact. TS% is about as good as you can get at it's job.

[dhsilv2]

As many pointed out, no one is using TS% to compare players across eras or for their whole careers, doesn't even make sense.

Well, you absolutely can use it across era's. You just have to modify it, and we've done that. TS%+ allows us to compare players between eras really well.

[dhsilv2]

TS% (plus assisted percentage ) is the only one that does a good job when comparing players IMO. PER a black box hidden formula. Probably tells you how bad a player is more than how good they are. VORP is essentially plus minus which has obvious flaws.

What? Someone else addressed the VORP which has zero plus minus data. But what?

uPER = (1 / MP) *
[ 3P
+ (2/3) * AST
+ (2 - factor * (team_AST / team_FG)) * FG
+ (FT *0.5 * (1 + (1 - (team_AST / team_FG)) + (2/3) * (team_AST / team_FG)))
- VOP * TOV
- VOP * DRB% * (FGA - FG)
- VOP * 0.44 * (0.44 + (0.56 * DRB%)) * (FTA - FT)
+ VOP * (1 - DRB%) * (TRB - ORB)
+ VOP * DRB% * ORB
+ VOP * STL
+ VOP * DRB% * BLK
- PF * ((lg_FT / lg_PF) - 0.44 * (lg_FTA / lg_PF) * VOP) ]

pace adjustment = lg_Pace / team_Pace

aPER = (pace adjustment) * uPER

PER = aPER * (15 / lg_aPER)

It's not a black box at all. It's a simple formula that you can do yourself with a little time and effort.

[kobe_vs_jordan]

TS% (plus assisted percentage ) is the only one that does a good job when comparing players IMO. PER a black box hidden formula. Probably tells you how bad a player is more than how good they are. VORP is essentially plus minus which has obvious flaws.

What? Someone else addressed the VORP which has zero plus minus data. But what?

uPER = (1 / MP) *
[ 3P
+ (2/3) * AST
+ (2 - factor * (team_AST / team_FG)) * FG
+ (FT *0.5 * (1 + (1 - (team_AST / team_FG)) + (2/3) * (team_AST / team_FG)))
- VOP * TOV
- VOP * DRB% * (FGA - FG)
- VOP * 0.44 * (0.44 + (0.56 * DRB%)) * (FTA - FT)
+ VOP * (1 - DRB%) * (TRB - ORB)
+ VOP * DRB% * ORB
+ VOP * STL
+ VOP * DRB% * BLK
- PF * ((lg_FT / lg_PF) - 0.44 * (lg_FTA / lg_PF) * VOP) ]

pace adjustment = lg_Pace / team_Pace

aPER = (pace adjustment) * uPER

PER = aPER * (15 / lg_aPER)

It's not a black box at all. It's a simple formula that you can do yourself with a little time and effort.

The inventor never released the formula. What you posting is the reverse engineered version that is close enough to the original version. Which is why different sites have different numbers.

The problem is you have to agree with the way to stats are weighted to agree with per. PER just combines a bunch of stats in a debatable manner.

Are steals really significantly more valuable than blocks? Are offensive rebounds 2.5 times more valuable than a defensive rebound? Stuff along those lines.

[kobe_vs_jordan]

TS% (plus assisted percentage ) is the only one that does a good job when comparing players IMO. PER a black box hidden formula. Probably tells you how bad a player is more than how good they are. VORP is essentially plus minus which has obvious flaws.

VORP (or BPM) don't use individual +/- or lineup data, it's box score based with a team component.. The formula is derived/indexed off long term +/- related numbers, but for individual players it doesn't care about their +/-.

Any plus minus variation sufferers from noise. Can mitigate that with doing a 3 season period of plus minus data.

At that point, the data matches the eye test better.

[dhsilv2]

TS% (plus assisted percentage ) is the only one that does a good job when comparing players IMO. PER a black box hidden formula. Probably tells you how bad a player is more than how good they are. VORP is essentially plus minus which has obvious flaws.

What? Someone else addressed the VORP which has zero plus minus data. But what?

uPER = (1 / MP) *
[ 3P
+ (2/3) * AST
+ (2 - factor * (team_AST / team_FG)) * FG
+ (FT *0.5 * (1 + (1 - (team_AST / team_FG)) + (2/3) * (team_AST / team_FG)))
- VOP * TOV
- VOP * DRB% * (FGA - FG)
- VOP * 0.44 * (0.44 + (0.56 * DRB%)) * (FTA - FT)
+ VOP * (1 - DRB%) * (TRB - ORB)
+ VOP * DRB% * ORB
+ VOP * STL
+ VOP * DRB% * BLK
- PF * ((lg_FT / lg_PF) - 0.44 * (lg_FTA / lg_PF) * VOP) ]

pace adjustment = lg_Pace / team_Pace

aPER = (pace adjustment) * uPER

PER = aPER * (15 / lg_aPER)

It's not a black box at all. It's a simple formula that you can do yourself with a little time and effort.

The inventor never released the formula. What you posting is the reverse engineered version that is close enough to the original version. Which is why different sites have different numbers.

The problem is you have to agree with the way to stats are weighted to agree with per. PER just combines a bunch of stats in a debatable manner.

Are steals really significantly more valuable than blocks? Are offensive rebounds 2.5 times more valuable than a defensive rebound? Stuff along those lines.

I'm pretty sure this will be dead on John's published values, been years but I tied my few tests back to John's numbers on ESPN. That said what's published everywhere today is the above. Obviously basketball reference is calculating it daily with the above. So to say it's a black box is wrong. A black box would be something you can't calculate yourself, such as say RPM which while we have a general knowledge of what it is and how that math works...there are some priors added that we don't know how they're done.

[kobe_vs_jordan]

What? Someone else addressed the VORP which has zero plus minus data. But what?

uPER = (1 / MP) *
[ 3P
+ (2/3) * AST
+ (2 - factor * (team_AST / team_FG)) * FG
+ (FT *0.5 * (1 + (1 - (team_AST / team_FG)) + (2/3) * (team_AST / team_FG)))
- VOP * TOV
- VOP * DRB% * (FGA - FG)
- VOP * 0.44 * (0.44 + (0.56 * DRB%)) * (FTA - FT)
+ VOP * (1 - DRB%) * (TRB - ORB)
+ VOP * DRB% * ORB
+ VOP * STL
+ VOP * DRB% * BLK
- PF * ((lg_FT / lg_PF) - 0.44 * (lg_FTA / lg_PF) * VOP) ]

pace adjustment = lg_Pace / team_Pace

aPER = (pace adjustment) * uPER

PER = aPER * (15 / lg_aPER)

It's not a black box at all. It's a simple formula that you can do yourself with a little time and effort.

The inventor never released the formula. What you posting is the reverse engineered version that is close enough to the original version. Which is why different sites have different numbers.

The problem is you have to agree with the way to stats are weighted to agree with per. PER just combines a bunch of stats in a debatable manner.

Are steals really significantly more valuable than blocks? Are offensive rebounds 2.5 times more valuable than a defensive rebound? Stuff along those lines.

I'm pretty sure this will be dead on John's published values, been years but I tied my few tests back to John's numbers on ESPN. That said what's published everywhere today is the above. Obviously basketball reference is calculating it daily with the above. So to say it's a black box is wrong. A black box would be something you can't calculate yourself, such as say RPM which while we have a general knowledge of what it is and how that math works...there are some priors added that we don't know how they're done.

the u in uPer stands for unadjusted Per . No way its dead on John Published values because it hasn't been adjusted and standarized lol Furthermore this is the equation posted is what basketball- reference uses . You can easily compare their values and ESPN values and see they are different .

[tsherkin]

It has been well-covered by now that TS doesn't belong on this list. PER is old, wasn't great when it was released, and in the nearly 20 years since its release, has been surpassed. Any unifying, single-number stat is going to have weaknesses, though. There will never be ONE number that just does it for everyone, and looking for such is a bit naive, really.

Looking at how given players perform across a selection of these different numbers gives us something to consider, but there is always going to be some subjective component. How things are arranged, what value is placed where, etc, etc.

[dhsilv2]

The inventor never released the formula. What you posting is the reverse engineered version that is close enough to the original version. Which is why different sites have different numbers.

The problem is you have to agree with the way to stats are weighted to agree with per. PER just combines a bunch of stats in a debatable manner.

Are steals really significantly more valuable than blocks? Are offensive rebounds 2.5 times more valuable than a defensive rebound? Stuff along those lines.

I'm pretty sure this will be dead on John's published values, been years but I tied my few tests back to John's numbers on ESPN. That said what's published everywhere today is the above. Obviously basketball reference is calculating it daily with the above. So to say it's a black box is wrong. A black box would be something you can't calculate yourself, such as say RPM which while we have a general knowledge of what it is and how that math works...there are some priors added that we don't know how they're done.

the u in uPer stands for unadjusted Per . No way its dead on John Published values because it hasn't been adjusted and standarized lol Furthermore this is the equation posted is what basketball- reference uses . You can easily compare their values and ESPN values and see they are different .

I'm pretty sure the difference is just that basketball reference and john use a different pace (I'd forgotten about that). Either way, you'll note I show the formula to convert uPER to PER above.