Insider 8/26/2005 Six gunners among free agents

[sunsfn]

Updated: Aug. 26, 2005, 12:12 PM ET
Six gunners among summer's free agents


By John Hollinger
ESPN Insider

Naming the game's best 3-point shooter isn't as easy as that little contest at All-Star Weekend would have us believe.

Take last season's Phoenix Suns, for instance. Joe Johnson shot 48 percent on 3-pointers last year, which would seem to make him one of the game's best. On the other hand, he didn't shoot that many on a per-minute basis (just two made 3s per 40 minutes). Was he really a better 3-point shooter than his teammate Quentin Richardson (the 3-point champ), who made a much larger number of attempts but shot a lower percentage? And is either that great, when one considers that neither had done much the year before?

To resolve these questions, I developed my "Top Gun" ranking of the game's 10 deadliest 3-point shooters. I wanted to find out which players had done the most to increase their team's scoring. And I wanted to find out at what rate their 3-point shooting helped the team, so I measured every player's output per 40 minutes of playing time.

For each player, I determined the number of points a team gained from a player's 3-point shooting expertise. To do this, I figured out the the difference between the player's shooting accuracy and what a mediocre (or "replacement level") player would shoot. I computed the "replacement level" figure by taking 90 percent of the league average in 3-point shooting, ending up with 32.0 percent in 2004-05.

Then I took the difference between the player's percentage and "replacement level" and multiplied it by the player's 3-point attempts. That resulted in the number of points the player added to (or subtracted from) his team's total with his 3-point shooting, compared to the "replacement level." Finally, I divided that result by the player's minutes and multiplied by 40. That produced a per-40-minute rating for the value of each player's 3-point shooting.

However, the examples of Johnson and Richardson show that 3-point percentages aren't as consistent from year to year as some other numbers. Thus, I looked at two seasons' worth of data. I weighed the 2004-05 season twice as heavily as the 2003-04 season, although in the case of rookies I could use only the 2004-05 data. Once that was done, I had a final score for each player.

Before I introduce the top 10, let's discuss a few surprising names who didn't make the list. Reggie Miller, for instance, had a horrendous 3-point year last season, making only 32.2 percent. Steve Nash is a great shooter (41.8 percent career) but hasn't shot the long ball with nearly the frequency of many other top shooters, so he didn't qualify either. And Johnson, despite his 47 percent mark a year ago, didn't come close. His low frequency of '04-05 attempts and his inaccuracy in '03-04 combined to keep him well behind the leaders.

With that said, let's take a look at the top 10:



10. Mike Miller, Memphis Grizzlies
Miller has one of the most gorgeous strokes in basketball, but until last season he hadn't found the net particularly often. Miller made 43.3 percent in 2004-05, however, to rank fourth in the NBA. That helped make up for an unspectacular showing the previous year, when he barely made one 3-pointer a game while hitting at a 37 percent clip.

9. Eric Piatkowski, Houston Rockets/Chicago Bulls
A lot of the players on this list aren't stars in the traditional sense, because they're guys who are in the league only because of their shooting. The Polish Rifle is a good example, hitting 42.5 percent from downtown last year but still rarely playing because his other weaknesses were so pronounced.

8. Ben Gordon, Chicago Bulls
The only rookie to make the list, Gordon was nearly as accurate on 3s (40.5 percent) as he was on 2s (41.4). Some might argue that Gordon should have been excluded from the study because he didn't have a two-year sample like the others, but based on his shooting exploits in college, his season hardly seems surprising. Incidentally, Gordon's inclusion bumped Jim Jackson out of the top 10.

7. Cuttino Mobley, Houston Rockets/Orlando Magic/Sacramento Kings (signed with Los Angeles Clippers)
Only two players finished in the top 10 in both 3-pointers per game and 3-point percentage last season. One was Damon Jones, and the other was Mobley. Mobley's 43.9 percent shooting ranked third in the NBA, while his 2.3 bombs per game rated 10th. He wasn't quite as strong in 2003-04, hitting 39.3 percent, but his long-range shooting ability will be a big boost to a Clippers team that was last in the NBA in 3-pointers.

6. Peja Stojakovic, Sacramento Kings
Peja shot the lights out in 2003-04, hitting 43.3 percent while raining in three a game, but couldn't keep up the pace in 2004-05. His numbers still were outstanding (40.2 percent, 2.6 triples per game), and he had the best 2003-04 of any player on this list. Last year's "slump" kept him out of the top five.

5. Wesley Person, Memphis Grizzlies/Portland Trail Blazers/Atlanta Hawks/Miami Heat/Denver Nuggets
Person is similar to Piatkowski -- a veteran, 6-foot-7 swingman bouncing around the league, but using his shooting to stay afloat. Person got especially hot when he went to Denver last season, hitting 48.5 percent, and that performance should earn him another contract for this season. His career 41.8 percent mark is the best of anyone's in the top 10.

4. Fred Hoiberg, Minnesota Timberwolves (now a free agent)
The Mayor led the NBA in 3-point shooting last season at 48.3 percent and was no slouch the year before (44.2 percent). If we were just looking at percentages, he'd be in the top spot, but because Hoiberg doesn't shoot the rock as often as the top three guys, he falls short. Hopefully we'll get to see his sweet stroke again -- his career is in question after offseason heart surgery.

3. Damon Jones, Milwaukee Bucks/Miami Heat (now a free agent)
Jones exploded onto the scene last season by leading the NBA in 3-pointers -- of course, that might have something to do with the fact that he played alongside Shaquille O'Neal, after signing with the Heat in the offseason. But Jones also earned plenty of those 3s on his own, shooting off the dribble in transition or coming off a high screen. He became the first player since Larry Bird to lead the league in 3-pointers while finishing in the top five in percentage (43.2 percent).

2. Donyell Marshall, Chicago Bulls/Toronto Raptors (signed with Cleveland Cavaliers)
Surprise, surprise. Though nominally a 6-9 power forward, Marshall has become an assassin from the corners. He hit 41.6 percent on 3s last season and set an NBA record with 12 bombs in a game. Believe it or not, Marshall had the best rate of made 3-pointers in the NBA last year, making 2.3 per game in just over 25 minutes of action. His offseason move to the shooting-starved Cavaliers should be a perfect match.

1. Kyle Korver, Philadelphia 76ers
He doesn't rank No.1 in either percentage or frequency, but over the two seasons Korver has been the most consistent (39.1 and 40.5 percent) while hitting at a prolific rate (one every 11.4 minutes in his two seasons). Overall, he ranked fifth in 2003-04 and sixth last season, which doesn't seem like it should add up to No. 1. But most 3-point shooters see their percentages go up and down like yo-yos, which is why several of the top players from 2003-04 (Brent Barry, Anthony Peeler, Rasual Butler) and 2004-05 (Joe Johnson, Eddie House) don't cut the mustard when we combine the two seasons.

Plus, Korver should continue to be among the most deadly shooters for some time. The results are based on his first two seasons in the league, but most players need a year or two of adjustment to get used to the NBA's longer 3-point line. If this is how Korver's "adjustment period" panned out, I can only imagine how many nets he'll be ripping in future seasons.


"TOP GUN" RATINGS
Player - 2003-04 - 2004-05 - Overall
Kyle Korver .61 .71 .679
Donyell Marshall .34 .85 .678
Damon Jones .18 .90 .66
Fred Hoiberg .44 .74 .64
Wesley Person .39 .75 .63
Peja Stojakovic .73 .56 .62
Cuttino Mobley .34 .68 .57
Ben Gordon N/A .56 .56
Eric Piatkowski .18 .67 .51
Mike Miller .22 .64 .50


-

 

[elindholm]

Taking 90% of the average accuracy to represent "mediocre" was a terrible idea. If he didn't want to use the straight mean, he should have at least looked at the distribution of percentages. 32% has to be very close to the bottom of the league among players with a decent amount of 3-point attempts.

 

[SactownSunsFan]

Is it just me or does Hollinger really not like the Suns? He goes off in one article talking about how we got rid of our three-point specialists in JJ and Q, thus destroying the most explosive offense in the last ten years. Then when he writes an article about the deadliest 3 point shooters he excludes them both because Q shot too low a percentage and JJ didn't shoot enough threes??? One way or the other Hollinger, you jerk!

 

[thegrahamcrackr]

Taking 90% of the average accuracy to represent "mediocre" was a terrible idea. If he didn't want to use the straight mean, he should have at least looked at the distribution of percentages. 32% has to be very close to the bottom of the league among players with a decent amount of 3-point attempts.

And this is why 82games.com has problems selling its stuff to teams. While it looks neat and is fun to talk about, the value isn't really there.

 

[SactownSunsFan]

And this is why 82games.com has problems selling its stuff to teams. While it looks neat and is fun to talk about, the value isn't really there.

Exactly. If you look at their best/worst defenders article, the author, using some kind of mathematical formula, ranks Raja Bell as being one of the worst defenders at shooting guard. This obviously isn't true- he's one of the best defensive SG's in the game.

 

[thegrahamcrackr]

Exactly. If you look at their best/worst defenders article, the author, using some kind of mathematical formula, ranks Raja Bell as being one of the worst defenders at shooting guard. This obviously isn't true- he's one of the best defensive SG's in the game.

The author is Dr. Rosenbaum. He actually has some neat mathematical models, and I would give them more credit over anything Hollinger wrote.

The problem is he tried to create a formula for something that doesn't have enough data. Basketball is quite simply hard to get high level statistical information in. The pace is just to quick for most people.

For example, in football they track tackles and assisted tackles. Basketball only gives steals credit to 1 person, and doesn't track deflections. If someone wanted to front the money (a lot of it) it would be possible to deliver the numbers. Until that time these models just don't usually work.

 

[George O'Brien]

The multi-year format he used is a joke if he's going to include Gordon who was shooting college three pointers the year before.

Clearly Q doesn't deserve to be in the top ten, but JJax was a lot more effective than some of the other off the bench guys.

 

[hora]

Taking 90% of the average accuracy to represent "mediocre" was a terrible idea. If he didn't want to use the straight mean, he should have at least looked at the distribution of percentages. 32% has to be very close to the bottom of the league among players with a decent amount of 3-point attempts.

"Mediocre" was a poor choice of words on his part. "Replacement-level" typically means
the level of player that the average team can obtain with zero cost (which generally means
one of the worst players in the league, or for example, in baseball, about the level of
an average AAA player). The reason you generally use replacement-level ability as your baseline
instead of average player ability is that average and even below average players have positive value
to their teams.

 

[George O'Brien]

"Mediocre" was a poor choice of words on his part. "Replacement-level" typically means
the level of player that the average team can obtain with zero cost (which generally means
one of the worst players in the league, or for example, in baseball, about the level of
an average AAA player). The reason you generally use replacement-level ability as your baseline
instead of average player ability is that average and even below average players have positive value
to their teams.

Welcome aboard.

I'm not convinced this is a viable approach. Baseball stats don't require the rest of the team to cooperate except very much on the margin (a pitcher wil be hurt by a slow defense that doesn't get to line drives). Also, three point stats come as part of what the coach asks for or accepts. Using a minimum salary player as a baseline makes no sense if the coach is unlikely to him shoot.

 

[hora]

Welcome aboard.

I'm not convinced this is a viable approach. Baseball stats don't require the rest of the team to cooperate except very much on the margin (a pitcher wil be hurt by a slow defense that doesn't get to line drives). Also, three point stats come as part of what the coach asks for or accepts. Using a minimum salary player as a baseline makes no sense if the coach is unlikely to him shoot.

Thanks for the welcome!

I must admit that I am much more comfortable with analysis of baseball statistics than analysis of basketball statistics.
The idea of replacement-level arose in baseball analysis so that you could correctly answer questions like the
following: Which is the more valuable season, 8 games of Tracy McGrady, or 80 games of Morris Peterson (assuming for
the sake of argument that Peterson is an average SF). If we use an average player baseline, 8 games of Tracy
McGrady is more valuable. If we use a "replacement-level" baseline, 80 games of Morris Peterson is more valuable
(for any rational choice of replacement-level anyway). I agree that the degree to which players' contributions are
isolated in baseball means that many of the concepts from the baseball statistical world do not apply (or need to be
carefully applied) to basketball analysis. However, I think a concept like replacement-level has value in
basketball analysis (like I said, though, I'm more comfortable with baseball analysis than baskeball analysis).

As for the choice of a minimum-salary player as replacement-level, there is certainly room to argue that
replacement-level should be higher. Typically, the reason you choose something that low is because it
represents zero cost to all teams in the league (even though in reality, richer franchises may be able
to easily afford replacement players that their poorer cousins cannot). If a typical coach is unwilling
to allow his min players to shoot, well, that affects their intrinsic value (after all, we're measuring
value, not ability).

 

[elindholm]

"Mediocre" was a poor choice of words on his part. "Replacement-level" typically means the level of player that the average team can obtain with zero cost

Right, well argued. The problem is that 32% is still far too low. I don't know how one would arrive at the right figure, but just choosing 90% of the league average out of the blue is a total stab in the dark, and intuitively we can see that it's not particularly close.

 

[elindholm]

The problem is that 32% is still far too low.

Just to argue this a little further...

Among the 50 players with the most three-point attempts last season, only two shot worse than 32%: Allen Iverson (31%) and J. R. Smith (29%). Among the top 50 in makes, it's only Iverson. So should everyone on the list, even the poor three-point shooters -- like Antoine Walker, Tracy McGrady, Baron Davis, Jason Williams, and, yes, Shawn Marion, all in the 32-33% range -- get "bonus points" in the analysis for being above this arbitrary standard for "mediocrity"? I would think not.

 

[sunsfn]

"Mediocre" is a wonderful choice of words to use, to explain John Hollinger's articles he writes for ESPN.

"Replacement-level" typically means the level of writer that the average fan can hope to obtain, with the cost of paying for ESPN Insider, in the replacement of Chad Ford!



-

 

[hora]

The problem is that 32% is still far too low.

Just to argue this a little further...

Among the 50 players with the most three-point attempts last season, only two shot worse than 32%: Allen Iverson (31%) and J. R. Smith (29%). Among the top 50 in makes, it's only Iverson. So should everyone on the list, even the poor three-point shooters -- like Antoine Walker, Tracy McGrady, Baron Davis, Jason Williams, and, yes, Shawn Marion, all in the 32-33% range -- get "bonus points" in the analysis for being above this arbitrary standard for "mediocrity"? I would think not.

You've definitely hit upon one of the major problems with the replacement-level concept, which is
that at the end of the day, it's hard not to be completely arbitrary when setting it. Although
I haven't completely thought this through, I also think that the idea of replacement-level has
more of a place in discussions of a player's total value, rather than the value of one small
subset of a player's contributions.

 

[George O'Brien]

You've definitely hit upon one of the major problems with the replacement-level concept, which is
that at the end of the day, it's hard not to be completely arbitrary when setting it. Although
I haven't completely thought this through, I also think that the idea of replacement-level has
more of a place in discussions of a player's total value, rather than the value of one small
subset of a player's contributions.

You may be right, although I'd need to see more of how it is going to be applied.

The problem of basketball statistics is a common issue on this board. Graham has access to a lot more interesting stuff than we do, but I'm sure even he will tell you how difficult it is to use basketball stats witout actually watching the player play.

In baseball, stats can do a decent job of determining how a player is doing. A stat like runs batted in requires assistance of other players, but stats like "batting average with men on base or in scoring position"; on base percentage, etc. don't require others and tell a lot about how good a hitter is.

Nothing in basketball can be completely separated from what the rest of the team is doing. Scoring typically requires someone to pass the ball to the scorer. Shooting percentage is heavily effected by whether the player got an open look, which means that picks, screens, and decoys all may have been part of the proces.

Defense is even harder. There is no meaningful statistic for measuring how good a player defends. Steals and blocks are not good indicators because too often players end up having to gamble to make the play, thus leaving their guy open if the gamble fails. Another problem is that some guys get tougher defensive assignments than others, so having good defensive stats (shooting percentage by opposing player) is not necessarily indicative.

A third problem is that basketball is a sport that plays backup players on a regular basis. In basebal, non-pitchers will play typically play the entire game. In basketball, bakups will play at least 25% of the minutes and often more. This means that players may be playing against backups some of the time. Star players may typically face starters whenever they are on the court, but others will get some minutes against backups.

Does this mean using statistics is pointless? Not at all. It is just that they won't tell as much as baseball statistics. The games are just too different.

 

[thegrahamcrackr]

Graham has access to a lot more interesting stuff than we do, but I'm sure even he will tell you how difficult it is to use basketball stats witout actually watching the player play.

That is very very true.

In a few years I suspect that a company will try to undertake a massive stats program that will go even more in depth than we do (which is amazing in itself).

Until that point, stats are simply a pointer. You need the video to see the context of how things occur.

 

[Joe Mama]

That is very very true.

In a few years I suspect that a company will try to undertake a massive stats program that will go even more in depth than we do (which is amazing in itself).

Until that point, stats are simply a pointer. You need the video to see the context of how things occur.

thank God we provide that too!

Eric, I agree that the shooting percentage that Hollinger used for "mediocre" was not a great one. However since everybody was compared to the same number it really wouldn't change the order of the list, would it?

Joe

 

[elindholm]

However since everybody was compared to the same number it really wouldn't change the order of the list, would it?

It would, because it would throw off the weighting between accuracy and number of shots made. For example, if his line is at 32%, then shots by a 40% shooter are worth twice as much as those by a 36% shooter (not really, but you know what I mean), because he's interested in how much more accurate the 40% shooter is than this hypothetical "mediocre" person. But move that line up to 34% or down to 30%, and it changes the relative weight between the 40% and 36% shooters quite a bit.

 

[George O'Brien]

However since everybody was compared to the same number it really wouldn't change the order of the list, would it?

It would, because it would throw off the weighting between accuracy and number of shots made. For example, if his line is at 32%, then shots by a 40% shooter are worth twice as much as those by a 36% shooter (not really, but you know what I mean), because he's interested in how much more accurate the 40% shooter is than this hypothetical "mediocre" person. But move that line up to 34% or down to 30%, and it changes the relative weight between the 40% and 36% shooters quite a bit.

In my thread How Big A Deal Is Three Point Shooting? I actually pulled the combined three point shooting for the NBA last season and the mean is 35.3%. The team averages hover in the same trading range between between 34.0% and 37.0%.

Of the 30 teams, only 5 teams were outside this range. Five were lower and and three were higher. As a general rule, the five teams that shot less than 34% for three, only the Raptors shot substantially more than the average number of threes for a team. The Hornets shot just 24 over the average while the Cavs, Hawks, and Jazz rationally took substantially fewer three point shots.

The over 37% stats are terribly distorted by the Suns. The Kings shot 104 more three point shots than the average while the Heat actually took 31.4 fewer threes than the average. The Suns shot an amazing 728 more three point shots than the average.

50% of the teams were within 1% plus or minus of 35.5% and 83% were plus or minus 1.5%. Of the 5 teams that shot under 34% (16.6% of the teams), cummulately they shot 821 fewer shots than average

What does this mean? I think it supports Eric's point that setting the baseline at 32% is unreasonably low. I'm not sure what the correct number should be, but shooting at the average of 35.3% does not get the shooter into the top 100 qualified shooters (taking the minimum number of shots).