Welcome! Please take a few seconds to create your free account to post threads, make some friends, remove a few ads while surfing and much more. ClutchFans has been bringing fans together to talk Houston Sports since 1996. Join us!

Jeremy Lin out 2 weeks with grade one knee sprain (UPDATE: Will return vs. Portland)

Discussion in 'Houston Rockets: Game Action & Roster Moves' started by thesonofsam, Nov 29, 2013.

Page 41 of 45

pippendagimp Member

Joined:

Sep 1, 2000

Messages:

27,074

Likes Received:

21,343

real_egal said: ↑

Houston Rockets completed that 22-game winning streak without Yao Ming for the majority of those games.
Click to expand...

yao actually got injured after the 12th win against NO

#801 pippendagimp, Dec 11, 2013

1 person likes this.

meh Contributing Member

Joined:

Jun 16, 2002

Messages:

15,426

Likes Received:

2,312

JustAGuy said: ↑

Exactly how would you structure that query?

site:clutchfans.net torocan -"sample size"

?
Click to expand...

I believe it was something like "lin stat" or something like that. But that post did appear on the first page of whatever I tried using.

Honestly, I've read a lot of torocan's posts, and he does generally caveat the data he presents.
Click to expand...

Yes, but many times he uses the wrong caveat. I totally agree with him about the concept of hypothesis and using logic to form an educated opinion on the future.

But let's be honest here. When you read a post that says "Beverley is horrible on defense. He's overrated." The poster is not trying to form a hypothesis. And as someone who had taken many tests that involved hypothesis, I find it insulting to have the concept used in GARM posts.

I don't understand being annoyed at someone on the site who actually backs up his comments with something, instead of the standard "this is so because duh". If the data has problems, you can actually refute it.
Click to expand...

Actually, this started because he dismissed my use of stat. Not the other way around. I was trying to use data to show that Beverley seemed to have been a good defender so far this year. And apparently someone didn't like me using this data. But he did like people who use hypothesis(without data) to argue otherwise.

http://bbs.clutchfans.net/showpost.php?p=8500763&postcount=585
http://bbs.clutchfans.net/showpost.php?p=8501048&postcount=589

#802 meh, Dec 11, 2013
Rocketfan_Dave Contributing Member

Joined:

Dec 22, 2004

Messages:

117

Likes Received:

16

pippendagimp said: ↑

yao actually got injured after the 12th win against NO
Click to expand...

Thank you! There are so many people around here that don't give Yao credit for the streak. He was there for over half of it. And if I remember correctly t-Mac was actually injured during the first half of the streak. Yet every time the streak is mentioned, t-Mac seems to be the only one doing anything. If anything, the streak was testiment to the chemistry and team-ball of that great rocket team!

Rant over

Repped

#803 Rocketfan_Dave, Dec 11, 2013
torocan Member

Joined:

Oct 15, 2012

Messages:

4,228

Likes Received:

436

First, an apology as this will be quite long...

meh said: ↑

torocan is basically saying that all people bashing Beverley's defense in this thread are "making a hypothesis." Do you agree with him that these people people fit the criteria for a hypothesis you're proposing?
Click to expand...

No, I never said that. You're making up a quote or comment that I never made.

I made ONE comment regarding the misuse of a statistic that was being used to forward an argument. You then stretched that into an argument that was irrelevant to the point that I was making against the statistic...

Shall I remind you?

torocan said: ↑

To be fair, if you're using that stat it's a bit dodgy as it skews all over the place depending on your defensive assignment and what line ups you're on the floor with in terms of time. Also it swings wildly over small game samples as it will depend heavily on the opposing teams and results. Being on the floor during a blow out like the Clippers game can make you look terrible. While being on the floor against the Offense impaired Bulls can make you look like a superstar.

Without breaking it down further over a much longer sample of games, it's hard to say.
Click to expand...

To which you replied...

meh said: ↑

Not sure why you brought Lin into this. The question was whether Beverley should be playing starter minutes(30+mpg). I have never once said Lin should NOT be getting starter minutes.

If anything, I've liked Lin's defense so far this year, except those games at SG where freaking James Anderson went Kobe on him.
Click to expand...

To which I replied...

torocan said: ↑

I was merely pointing out that the stat is unreliable over small game sample sizes. IE, if you're pointing to it as proof that Beverley is All NBA defense off that stat, then Lin is as well. An assertion which is on the face of it, ludicrous. :grin:
Click to expand...

Whereupon you then switched it to an argument about Beverley, arguing that my criticism of the use of a single statistic was automatically an attack on Beverley...

meh said: ↑

I don't understand why you have to get annoyed at my use of small sample sizes, when the attacks on Beverley in this thread all use the same small sample sizes?

Unless, of course, you can point out to someone who makes a logical argument using large sample size which makes a convincing argument that Beverley has become a mediocre defender now that he plays more minutes?
Click to expand...

Which I pointed out was not the case...

torocan said: ↑

I'm not opposed to the use of small sample sizes per se, as long as they're considered with the usual caveats. IE, statistical variation/deviation, context, etc.

As for Beverley's defense, I think the sample size is still too small to determine if his defense is declining, improving, or remaining the same in terms of correlation to minutes played.
Click to expand...

I then forwarded a Hypothesis that it was possible that Beverley's defense *could* decline with increased minutes...

torocan said: ↑

However, I do think it's not an illogical hypothesis to argue that it is very difficult to maintain a high level of on ball defense for large numbers of minutes due to 2 simple factors.

1) Energy expenditure. There have been players that have maintained very high energy levels for 30+ minutes per night when playing aggressive guard defense, but that list is very, very short. It's simply physically exhausting to constantly chase guards and fight over screens. Is it possible that Beverley isn't suffering from a decline in overall per minute effectiveness? Sure. However, it would be unusual.

2) Fouls. Beverley's on ball defense relies heavily on being physically aggressive. This is fine in small minutes like 15 or 20 minutes when it's highly unlikely that he'll foul out. However, when playing 30+ minutes, the issue of fouls becomes much more significant. On a purely stylistic basis Beverley may be required to play less aggressive defense (less bodying up, and less "handsy") simply in order to stay in the game, especially against players with more favorable referee treatment (superstar calls).

So, while the sample size is too small to draw any conclusions, I think as a hypothesis that Beverley's defensive effectiveness will have a drop off with increasing minutes is not one without logical basis.

Whether that plays out in reality, we won't know for some time.
Click to expand...

Notice how I qualified that while it was possible, that it was not certain and could not be determined without sufficient data.

However, I guess that made you unhappy as your next response was...

meh said: ↑

So what you're saying is, you have no problem with people attacking Beverley's defense for reasons that are only supported by personal (non-professional) opinion and not by any factual events.

But when someone uses data to dispute that above opinion, then you need large sample size proof to ensure that it's 100% ironclad?

I'm curious to know how you feel about people attacking Lin's play this year. Do you feel you cannot defend Lin because, let's face it, it's only a small sample size?
Click to expand...

Thus demonstrating a desire to conflate a Hypothesis of future behavior with a discussion of the validity of a statistic used as evidence in proving the validity of a completely different hypothesis.

You then made an extreme exaggeration ("100% ironclad") in an attempt to invalidate a legitimate concern with the validity of a statistic that YOU introduced into a discussion that had not only a small sample size and high variation, but a statistic for which you provided no context, no qualifiers, and no corroborative statistics or models.

In other words, because you were unwilling or unable to refute my concern with introducing a statistic with legitimate concerns of validity, you instead chose to mischaracterise my intent and original statement.

I re-stated that concern to make sure you understood...

torocan said: ↑

No, what I'm saying is that someone can have an opinion (pose a hypothesis), but it's nowhere near factual until you have data that argues one way or the other in a meaningful way.

Whereupon you responded with an attack on my character...

meh said: ↑

I love the steps you'll go to support Beverley bashing.

Random Beverley haters(or perhaps support a certain other PG): "Beverley's defense has sucked. I have no proof. But he has to be sucking because he's play more minutes."
Click to expand...

And then intentionally mischaracterised my statement as an attack on Beverley...

meh said: ↑

meh say: "So far, that hasn't seemed to be the case. Beverley looks like he's defending fine based on the results so far."

torocan to meh: "Well, it's only been a few games. And although Beverley has been a good defender, let's wait some more games before we stop calling him a bad defender. The results aren't here yet. But logic would imply that he'll falter in the future."
Click to expand...

And then engaged in a character attack in an attempt to discredit my statement...

meh said: ↑

I don't want to bother digging up all your old Lin posts. But I'm quite sure I can find gazillion of them pointing to sample sizes of less than 19 games to support an argument for Lin. The fact that you expect more than a 1/4 of season of sample size for a PG different from Lin of course, is the epitome of Lin fandom.
Click to expand...

To which I noted the correct interpretation of my statement...

torocan said: ↑

No, the correct response is that the data is inconclusive in either direction, that he is improving or declining.

Whereupon you then attempted to attack my character again...

meh said: ↑

Not that I expect you LOFs to recall anything beyond your worshipping of Lin, nor do I expect you to actually respond to my post with anything resembling acceptance that torocan conveniently "forgets" sample size when the situation fits him...
Click to expand...

And then presented a link to data that was completely different in nature to the data you presented in that I was presenting factual data that was counter to someone else's hypothesis in the absence of ANY data.

meh said: ↑

Here's one for you.

http://bbs.clutchfans.net/showpost.php?p=7351691&postcount=201
Click to expand...

Where then you attacked my character again...

meh said: ↑

Look, I'm fine with fans being biased. I'm biased. But at least I admit my bias and think more of the Rockets than other fans would. Just admit you guys put on rose colored glasses when you watch Lin and we can all have nice discussions.
Click to expand...

And then you presented a false equivalence in an attempt to invalidate any attempt at discussion by mischaracterising the presentation of a hypothesis as a presentation of fact....

meh said: ↑

Ah, good ole hypothesis. When facts fail, go for making up stuff and pass it off as facts.
Click to expand...

Which you followed by a sarcastic remark...

meh said: ↑

I remember the good ole days when hypothesis are vetted through scientific process and only accepted if the Null Hypothesis is rejected. In this case, the Null Hypothesis would be "Beverley is a good defender." And so far hasn't been rejected.

But I can understand. If it can get rejected in the future, why not make the statement today when it hasn't happened yet? Isn't that how things suppose to work.
Click to expand...

And then presented another false equivalence in comparing an opinion without supporting evidence or logical basis with a hypothesis...

meh said: ↑

Here's a hypothesis from me then. "Lin Sucks. I don't have to prove that I'm correct. But you have to prove me wrong. And you cannot use stats from this season because there's not enough of a sample size." Yeah, that's really the definition of a "reasonable" argument.
Click to expand...

Whereupon you then changed tracks completely, making a vast generalization and presented a straw man claiming that Nobody *ever* presents Hypotheses...

meh said: ↑

I know what a hypothesis is. I also know that no one bashing Beverley here is making a hypothesis. Just like no one saying "Lin sucks" last year were making a hypothesis.

Which you followed with another false equivalence between a structured Hypothesis and a ridiculous statement...

meh said: ↑

If I say "Sky is purple.". And you say I'm wrong because you look up the sky and it's blue. And then I counter with "I'm making a hypothesis and you've only seen the sky for 2 seconds. Therefore you can't say I'm wrong yet." Then I'm full of bull****, not making a hypothesis.

Which you then used as justification to NOT have to adhere to any standard of debate, discourse or presentation of data...

If you truly know what a hypothesis is, then you should also know that comparing the concept of hypothesis to what's written on this message board is just stupid. It's like comparing Shakespeare to random fan-fiction on the internet.
Click to expand...

Where you then presented another false equivalence to justify your use of poorly structured arguments...

meh said: ↑

Do you believe "Lin sucks because the league has figured him out" to be a proper hypothesis and should never be countered without a mountain of evidence?
Click to expand...

Which brings us back to the original mischaracterisation of my original statement that I noted at the beginning of this exceedingly long post.

meh said: ↑

torocan is basically saying that all people bashing Beverley's defense in this thread are "making a hypothesis." Do you agree with him that these people people fit the criteria for a hypothesis you're proposing?
Click to expand...

Sadly, this all began because you completely misunderstand the nature of my original complaint. And this is evident in your last post....

meh said: ↑

Actually, this started because he dismissed my use of stat. Not the other way around. I was trying to use data to show that Beverley seemed to have been a good defender so far this year. And apparently someone didn't like me using this data. But he did like people who use hypothesis(without data) to argue otherwise.
Click to expand...

And this is the real tragedy. I have no issue with people using statistics. However, if you're going to use statistics then expect the following:

1) You will be expected to use the relevant statistics (do not use the wrong stat to argue a point)
2) Expect your statistic to be scrutinized if it is not scientifically or mathematically sound.

You can't just pull out a stat over a small sample size without having a mathematically or methodologically sound justification for the use of that statistic. If you are as familiar with statistics as you claim, then you would KNOW that not all statistics are created equal, especially statistics with small sample sizes...

torocan said: ↑

I'm not opposed to the use of small sample sizes per se, as long as they're considered with the usual caveats. IE, statistical variation/deviation, context, etc.
Click to expand...

IF you wish to present a statistic, then expect that it's validity may be challenged. Whereupon you can either acknowledge the limitations of the data set, or you can explain why you feel the statistic still hold some validity due to whatever factors you feel are relevant.

However, I somehow suspect that you're not really concerned in a debate over the validity of your statistic or the methodology that you use to frame that statistic...
Click to expand...

Click to expand...

Click to expand...

Click to expand...

Click to expand...

Click to expand...

Click to expand...

Click to expand...

Click to expand...

#804 torocan, Dec 11, 2013
Crunchyball Rookie

Joined:

Nov 20, 2013

Messages:

359

Likes Received:

5

Come on meh ! Save us from torocan , he is like an evil genuis

#805 Crunchyball, Dec 11, 2013
d12babymamas Member

Joined:

Oct 10, 2013

Messages:

1,608

Likes Received:

34

um wow something tells me this torocan dude is a stats professor and meh is fighting an uphill battle while he's chained to a tree. i would stop before it gets more embarrassing

#806 d12babymamas, Dec 11, 2013
Spiegel Member

Joined:

Feb 9, 2010

Messages:

5,403

Likes Received:

101

Rocketfan_Dave said: ↑

Thank you! There are so many people around here that don't give Yao credit for the streak. He was there for over half of it. And if I remember correctly t-Mac was actually injured during the first half of the streak. Yet every time the streak is mentioned, t-Mac seems to be the only one doing anything. If anything, the streak was testiment to the chemistry and team-ball of that great rocket team!

Rant over

Repped
Click to expand...

Tmac missed the 1st game of the streak but played in the next 21 and played as well as anyone in the league in that streak. Yao missed the second 11 but played in the first 11 or 10.

#807 Spiegel, Dec 12, 2013

highlander3128 Member

Joined:

May 21, 2008

Messages:

741

Likes Received:

15

Torocan is actually Daryl Morey posting on the GARM!!!

#808 highlander3128, Dec 12, 2013
Lihao Member

Joined:

Oct 21, 2012

Messages:

1,023

Likes Received:

28

that's a knock out there by Torocan , sending a clear message "don't mess with me" :grin:

#809 Lihao, Dec 12, 2013
scaramanga Member

Joined:

Nov 2, 2012

Messages:

282

Likes Received:

6

Damn.... I don't know how people got so much time to compose such lengthy arguments... I'm feeling like I'm watching Kill Bill 3.

#810 scaramanga, Dec 12, 2013
PatBeverleyFan Member

Joined:

Mar 10, 2013

Messages:

412

Likes Received:

39

Torocan dropping the hammer lol

#811 PatBeverleyFan, Dec 12, 2013
Panda23 Member

Joined:

Jul 2, 2012

Messages:

8,566

Likes Received:

619

MFW reading the above

#812 Panda23, Dec 12, 2013
ipuff Member

Joined:

Nov 8, 2013

Messages:

46

Likes Received:

1

to torocan

#813 ipuff, Dec 12, 2013

RedEyesKirby Member

Joined:

Oct 30, 2012

Messages:

1,025

Likes Received:

67

torocan said: ↑

First, an apology as this will be quite long...
...
Click to expand...

#814 RedEyesKirby, Dec 12, 2013
rhadamanthus Contributing Member

Joined:

Nov 20, 2002

Messages:

14,304

Likes Received:

596

wow....

#815 rhadamanthus, Dec 12, 2013
EnergyGuy Rookie

Joined:

Dec 2, 2013

Messages:

466

Likes Received:

18

rhadamanthus said: ↑

wow....

Click to expand...

That knockdown is like a heavyweight champion toying with a random guy on the street.

Torocan just trades a few jabs with meh for a couple of rounds, knowing full well he can deliver the knockout whenever and however he wishes ... and when the right time comes ... BAM!!!

That was embarrassing.

But meh will bounce back. He always does. No matter how many times he embarrasses himself, he'll always come back louder than ever. And yes, he'll blame this on LOFs as opposed to do some self reflection. It's always the LOFs' fault ...

#816 EnergyGuy, Dec 12, 2013
meh Contributing Member

Joined:

Jun 16, 2002

Messages:

15,426

Likes Received:

2,312

torocan said: ↑

1) You will be expected to use the relevant statistics (do not use the wrong stat to argue a point)
2) Expect your statistic to be scrutinized if it is not scientifically or mathematically sound.
Click to expand...

1. Was my stat not relevant? Do you disagree that opposing eFG% is a bad stat to use for PG defense?
2. Was my stat not mathematically sound? As in, do you believe 82games made up that stat and not actually calculated it correctly?

You can't just pull out a stat over a small sample size without having a mathematically or methodologically sound justification for the use of that statistic. If you are as familiar with statistics as you claim, then you would KNOW that not all statistics are created equal, especially statistics with small sample sizes...
Click to expand...

Okay. Then prove to me that my sample is statistically insignficant. After all. You questioned my methodologies. So provide proof. Proof that Beverley's sample size is small enough that you cannot make a valid conclusion.

If you do so, and have your conclusions validated multiple posters, I will personally donate $30 to the tipjar and write any signature you wish.

#817 meh, Dec 12, 2013
Lihao Member

Joined:

Oct 21, 2012

Messages:

1,023

Likes Received:

28

i'm surprise meh still going on about the opponents efg% thing after Torocan showed him the flaw of his own statement.

Patrick Beverley
Opposing EFG% - .430, Opposing PER - 12.5
(meh : All-NBA defense, couldn't careless if he averaged 5/2 )

Jeremy Lin
Opposing EFG% - .432, Opposing PER - 12.3
(meh : limited defensively , that's why demoted to bench)

btw i do think Bev is a better defender before you go on and call me delusional LOF who just want to bash Bev

#818 Lihao, Dec 12, 2013

torocan Member

Joined:

Oct 15, 2012

Messages:

4,228

Likes Received:

436

Okay. Then prove to me that my sample is statistically insignficant. After all. You questioned my methodologies. So provide proof. Proof that Beverley's sample size is small enough that you cannot make a valid conclusion.

If you do so, and have your conclusions validated multiple posters, I will personally donate $30 to the tipjar and write any signature you wish.
Click to expand...

Ok, personally I don't care about the tip jar, but it's nice of you to offer.

meh said: ↑

1. Was my stat not relevant? Do you disagree that opposing eFG% is a bad stat to use for PG defense?

2. Was my stat not mathematically sound? As in, do you believe 82games made up that stat and not actually calculated it correctly?

Ok, I'm not going to get into sample size within this post as I don't have the time right now to do a complete write up on statistical probability and the nature of sample size, confidence level and standard deviation.

I'll try to get into that in a later post. However, I will address your specific use of eFG% as a statistic in general.

eFG% is calculated as eFG% = 100 * (FG + 0.5*3PT) / FGA

Mathematically the calculation of the stat is sound and calculated correctly, however that's not the problem with using eFG% as a statistic.

The question regarding eFG% is whether the stat is relevant or a good stat for assessing PG defense.

While eFG% is significantly better stat to use to judge defense than FG% as it factors in the value of 3pt%, it makes zero effort to assess the impact of Free Throws. That it does not include Free Throws is a significant shortcoming as it does not differentiate between the denial of a shot/miss, from a complete lapse in defense like a blow by and desperation swipe foul, or a defender fouling a jump shooter or finisher.

For example, let's say that you're defending James Harden.

He takes 1 shots and misses it (0/1).

He then drives past you 4 times and in desperation swipe him from behind and you foul him 4 times. He then takes 2 x 4 (8) free throw attempts and sinks them at an 85% clip (6.8 but we'll say 7 points).

According to eFG%, you held James Harden to 0% eFG%, though he actually scored 7 points from 5 plays due to your "defense". On top of that you have accumulated 3 fouls AND put the opposing team into the bonus earlier than normal.

Now, let's say that Clutch the uber wing defender is also on Harden. Harden takes 5 jump shots and misses all 5 (0/5).

According to eFG%, Clutch also held James Harden to 0% eFG%, however unlike you Harden scored 0 points from 5 plays due to his defense. All while accumulating 0 personal or team fouls.

Similarly, eFG% does not account for bad fouls of jump shooters or players that finish at the basket (And1's).

Let's say James Harden drives to the basket 5 times, and he makes 3/5 but you foul him 3 times when he finishes. He then shoots 3 And1's at 85% (2.55 points, we'll say 2 to be kind). Harden has now scored 6 + 2.55 points, or 8.55 points on 5 attempts. However, Harden's eFG% will show 60%.

Now, Clutch the elite Wing defender is on Harden. Harden drives to the basket 5 times, and he makes 3/5 but he doesn't foul Harden at all. Harden has now scored 6 points, or 6 points on 5 attempts. Yet Harden's eFG% is also 60%.

Even though you gave up 2.55 more points AND accrued 3 fouls, according to eFG% you are the same defender as Clutch.

eFG% has made it look like you're some awesome defender on par with Clutch under BOTH scenarios. This is why eFG% is a terrible stat to determine defense, though in it's defense (no pun intended) eFG% is a somewhat better stat to judge *just* offensive jump shooting efficiency (as it excludes free throws).

So, in the direct scoring sense eFG% is actually a terrible stat to calculate scoring defense simply due to it excluding the impact of free throws. A superior statistic (and more commonly accepted statistic) to use would be TS% (true shooting percentage).

TS% = TS% = 100 * (1/2) * PTS / (FGA + 0.44*FTA)

TS% attempts to calculate the impact of FT% into over all scoring. I know the 0.44*FTA looks kind of wierd... it's an attempt to remove the impact of technical free throws. The 0.44 is a fairly arbitrary number based upon expected average ratio of FT's to shots (versus technical free throws). If you wanted to be more precise you would have to actually calculate every technical free throw. Of course that's incredibly cumbersome so most of us just accept that TS% will be a smidge off. That said, TS% isn't without it's own faults.

Aside from the lack of the consideration of Free Throws, eFG% does not factor in points that arise from outside of on ball defense. What I mean by this is it only factors when someone attempts a jump shot, it does NOT factor in defense that is related to how well you played the passing lanes (ball denial).

Imagine once again you're once again defending James Harden (I know, tough assignment). You're playing him somewhat close, but your body positioning is poor relative to other players on the floor.

So, because of how you're positioned (hands not extended to the sides) and at a poor angle, there is an easy passing lane towards the base line and the middle. Now, because your body position is poor, Chandler Parsons cuts along the baseline and James Harden does an easy chest pass to a cutting Parsons for a dunk.

He then does that 5 times in a row. Even though Harden did not take a single shot, the Rockets have now scored 10 points off those 5 plays. On the bright side, eFG% now shows that Harden shot 0%.

Now, imagine Clutch the elite wing defender. He not only plays Harden close, but his hands are extended to the side. He has positioned his body and hands in a way that denies him an easy passing lane along the baseline. Even though Parsons makes a baseline cut, Harden doesn't have a clear passing lane because your body is in the way. Instead, he does 5 swing passes outside so that the defense can try to find another way to exploit the defense.

According to eFG%, Clutch also has a eFG% of 0%.

Apparently now you are as elite a defender as Clutch according to eFG%, even though your poor defensive fundamentals have indirectly led to 10 points and Clutch's solid fundamentals have indirectly led to 0 points.

Once again, eFG% shows it's limitation in terms of determining over all defensive play. It makes zero attempt to judge fundamental defensive positioning and stance.

In it's defense, TS% also fails in this instance as it only judges direct scoring impact.

That said, there is no *single* stat that is able to decisively determine positional and passing lane defense. Those who try to assess defense have tried to come up with a universal defense, however given the complexities of team and individual defense within a team setting, the best that most of us have been able to do is to judge player defense is to instead view defense through a contextual tapestry.

In other words, instead of taking a single stat, we take a variety of statistical models and data and present them together, thus giving us an idea of likely differentials in defensive effectiveness.

Without getting into the nitty gritty of every single stat, here are *some* of the stats that would typically be included/examined in a detailed discussion of defensive prowess...

eFG% (non foul, jump shooting), TS% (fouls included), RAPM (regression adjusted plus/minus), team and individual PPP (points per possession), PPS (points per shot), DWS (defensive win shares), DWS/48 (defensive win shares/48), DRtg (defensive rating), DeFF (defensive efficiency), Steals% (steals per possession), SPG (steals per game), B/48 (blocks/48), BPG (blocks per game), Blocks% (blocks per possession), Opponent team points/pass (Opponent points earned for every pass), Opponent APG (opponent assists per game), Opponent AP/48 (Opponent assists per 48) ... and the list goes on.

The idea is that because it is hard to quantify defense in it's totality both on an individual and team level, that in order to reach any significant level of confidence it's best to assemble multiple statistical models that assess specific areas of defense with varying levels of accuracy.

This is what I mean when I say Contextual Tapestry. You are bringing together multiple statistical models with various strengths and weaknesses and looking for commonalities of direction, the idea being that as you add more and more data from various points of view you are reducing the margin of error in terms of the quality of your statistical evidence.

In addition to that you would combine that with the "eye" test to add context to that statistical picture. For example, having Dwight Howard or Omer Asik in the paint can mask the defensive deficiencies of many players versus having let's say Greg Smith patrolling the paint.

When you combine enough data from multiple models with the "eye test", you can then make Hypotheses and conclusions with a higher level of confidence, though obviously without absolute certainty as the nature of any statistical analysis is there is no such thing as absolute certainty.

In statistical analysis, all you're shooting for is a high or acceptable confidence level. In other words, you've accumulated enough corresponding data that you can comfortably say, "That's good enough to satisfy me" or "The results seem to imply that..."

I'll see about addressing your sample sizing at a later time. My apologies, but this is bordering on a rather lengthy novel as it is.
Click to expand...

Click to expand...

#819 torocan, Dec 12, 2013
pnr Member

Joined:

Aug 1, 2012

Messages:

2,387

Likes Received:

58

Torocan is Morey..

#820 pnr, Dec 12, 2013

(You must log in or sign up to post here.)

Page 41 of 45

Share This Page