APBRmetrics

Stating in terms of difference is probably better but the ratio of estimated positive impact in this case is as I stated. RAPM is already stated in terms of estimated positive and negative impact. I didn't convert it to that. I converted that into two ratios and left another as one large apparent difference. A minor choice for descriptive purposes.

Crow wrote:Stating in terms of difference is probably better but the ratio of estimated positive impact in this case is as I stated. RAPM is already stated in terms of estimated positive and negative impact. I didn't convert it to that. I converted that into two ratios and left another as one large apparent difference. A minor choice for descriptive purposes.

except it's not so clear. it's hard to say player X was some product better than player Y without throwing in total minutes played, which somewhat confounds the matter. I think we can simply say, however, that on a per-minute basis, Garnett was far superior to Rondo.

Before this is put to bed, I wanted to return to the original point of this post (What reasoned belief should one have about the practical value of analytics to an NBA franchise?) incorporating the new data Jeremias made available since it was made.

Jeremias' latest, average RAPM (incorporating more accurate data in the early years, as I understand) casts the performance of the Mavs over the ten year period in question (2002-11) in a slightly different light. Whereas before, the contributions of players not named Dirk Nowitzki and Josh Howard, had a net positive over the period in question (again, by my calculations, the equivalent of one player, playing about two thirds of a season's minutes, providing +1.3 points per 100 possessions) that is now gone.

With DN at +6.9 (playing 75% of minutes) and JH now at +3.5 (playing 35% of minutes), what is implied is that to have reached an average of 57+ wins per season, all that was required of a "supporting cast" was 0. This to me is rather awesome, considering the circumstances and in the implication for the future.

Again, here was a franchise, having informational/analytic advantages never before seen and never to be seen again, with top-down pressure to have these advantages realized within the organization and willingness to put money where the numbers were, and the net effect was, in a very real sense, zero: no net positive performance.

Unless this accounting framework (and/or the specific estimates of DN's and JH's value) is rejected, the only alternative conclusion is that +0 was actually a very real accomplishment, that the forces of retrogression for a successful team are so strong (and/or that the Mavs happened to be unexpectedly unlucky on the injury front), that apparently standing still involved a whole lot of running and was actually a significant feat. And maybe that's true, and actually, I would like to believe that it is true.

But where is the evidence for that?

Regarding "excess" injuries, I look at the ten years in question, and I simply don't see any. To the contrary, my impression is that the Mavs were relatively less injury prone than the typical NBA franchise. But maybe this is a misimpression.

And then aside from possible "excess" injuries, the only other retrogression factor to account for are the net negative effects of player aging. What is certainly true is that the "other Mavs" were an older group than average. Over the decade in question, the minute-weighted average age of the "other Mavs" ranged from 27.8 to 30.8, with a straight average across the years of 29.0. How to account for the impact of the supporting cast being "over the hill"?

Well, we are operating here in an APM world, and Daniel has provided us with an aging curve based upon Advanced SPM: https://picasaweb.google.com/1183961692 ... 7883209074. Taking data from this graph and applying it to the "other Mavs" over the years, what I get is that their being over the hill on average implied an expected yearly retrogression (assuming identical minutes for the year following) of 1.0 points per 100 possessions.

So, perhaps that is it. The net benefit of Mark Cuban's earliest adoption and full implementation of "analytics" was +1.0. Or maybe not. This result is completely dependent on average player performance peaking at age 26. If the "top of the hill" is 28, by contrast, the +1.0 essentially vanishes.

And this is another discussion, but I am not yet persuaded that the Adv. SPM shown best represents aging effects in APM world. My sense is that peak performance does not occur, on average, at age 25 to 26 but somewhat later.

Maybe this intuition is incorrect, but to the extent that box score stats aren't picking up "all the little things" and that these are learned with age and experience, the peak should shift rightward. And then there is (admittedly incomplete) information gleaned from Jeremias' recent estimates. Starting at the superstar end of the tables, the performance peaks I see (for players early in their careers when the series starts) average later than 25 to 26. Perhaps Jeremias can confirm this impression?

Can't. Stop.

OK, a clarification and a comparison. First, the clarification. Jeremias' initial 10 year estimate included years 2002 to 2011 and my calculations corresponded. Now the base for all multi-year estimates is 2001, what happily coincides with the first of the Mavs 50 win seasons. So, job one is to correct the calculations to reflect the 11 year run. Then the comparison: the tale of the Texas tape. Coincidentally, the Mavs' 11 year run occurs within the Spurs' ongoing 13 year run, and the coincidences don't stop there, so an interesting comparison can be made, with perhaps a minor implication concerning the realized value of analytics.

So, facing off in the lucky lottery club are the Mavs and the Spurs. Dirk Nowitzki and Josh Howard for the former, Tim Duncan and Manu Ginobili for the latter. It just so happens the Spurs got luckier. The relevant RAPM, playing time average within the 11 year run (prorated in the cases of JH and MG) for the individuals and the combo are as follows:

Dallas: Dirk Nowitzki +6.9/0.74, Josh Howard +3.5/.31, Combined +6.2/1.1
San Antonio: Tim Duncan +8.9/0.68, Manu Ginobili +5.8/.41, Combined +8.4/1.1

Then during this 11 year span, Dallas averaged 56.4 wins per year to San Antonio's 57.6. However, were each franchise to have fielded supporting casts (each of 3.9 full-time players) with a +0 rating, Dallas would have been expected to average in the low 57s, whereas San Antonio would have been in the 62s.

So, relative to the star endowment, Dallas did better, about +1.8 better (with Dallas' supporting cast averaging about -0.3 and that of San Antonio about -2.0).

So, this being a different perspective on the value of analytics issue, perhaps some of that +1.8 could be attributed to the superiority of Dallas' analytics. But then again, it also cost them dearly in terms of salary. Over the 11 years, Dallas' average salary bill was $22.2 million greater than that of San Antonio's; what comes to $12.6 million per point.

Overall, $244 million more dollars in the pockets of the Spurs' owners (and three more championships too).

It really, really pays to be lucky.

Just to add in some more anecdotes, what then just might sum up to data, if you take the top 30 players, as ranked in the last year of the latest and greatest (http://stats-for-the-nba.appspot.com/PBP/2012np.html), and identify the age of each player at his peak year (where age is defined as age at the start of the season, unless there was a birthday within season, in which case the age is identified as "age at start plus 0.5"), here's what you find for the averages.

Well, first you want to cull out the oldies, where it can be presumed that the actual career year is not identified for a lack of pertinent data (Shaq, Stockton, and Terry Porter) and also poor Yao Ming, where his actual rating has improved in the three years after he stopped playing. So this leaves 26 folks on the list, 15 of whom had career years prior to the last one, leaving 11 "youngsters" many of whom, it can be inferred, have not yet reached their peak.

And here's what you find:

(1) For the "past prime" group, those for whom the peak is identified, peak (multi-year/career) RAPM was reached at the age of 29.2. And this conclusion likely would be bolstered (and even increased) with the inclusion of the omitted three.

(2) Then for the "youngsters", which interestingly includes Steve Nash, the peak occurs at 27.4, and excluding the ever youthful SN it drops to 26.4.

Perhaps these data are unrepresentative of the whole, but if they are not, they suggest the RAPM peak is to be found three years later than what Daniel found with Adv. SAPM.

And, returning to the main story line, this in turn suggests that Mark Cuban's Mavs weren't combatting the vicissitudes of old age in maintaining the mini-dynasty, at least in terms of RAPM. The average age of players he acquired corresponded to the peak performance age, around which expected yearly improvements and retrogression would have been very small. And furthermore, my guess is that he knew quite well the approximate age when peak performance could be found. (How could he not, given the information he had on hand?)

I look forward to learning what the true populations results are.

Interesting analysis.

It is not directly on what you just discussed but I am caused to wonder how much of the Spurs longterm overperformance was directly related to their overperformance on the big 3 draft picks. Has anyone calculated the average and peak RAPM by draft pick number or range? I don't recall seeing it. I think Daniel did ASPM for draft picks, which is something useful, but that leaves out shot defense and other indirect impacts.

Crow wrote:It is not directly on what you just discussed but I am caused to wonder how much of the Spurs longterm overperformance was directly related to their overperformance on the big 3 draft picks.

Just to make a simple point: live by RAPM (or any metric), die by RAPM (or any metric). Assuming you mean Tony Parker and not Bruce Bowen, RAPM does not accord him "big" status (though he is above average for a point guard). Sticking with the 2001-11 framework, Tony Parker was in 10 of these years, averaging +0.7. Then once you take into account the actual playing time, not nothing, to be sure, but really an asterisk in explaining the Spurs over-performance.

But a couple other observations on Spurs drafting and lucky-duckiness. On the one hand, looking over their draft history, aside from Ginobili and Parker, there are no obvious coups or even RAPM positive players (though George Hill looks promising). This, of course, is in an absolute sense, not with respect to draft position norms. But, that does leave Ginobili and Parker to explain, and here, I think you see the genius of Popovich, call it Auerbachesque.

To my mind (and maybe I'm putting structure on things that isn't merited) these picks represented a willingness to go against the conventional wisdom and hazard something (though not much frankly) on the notion that non-native US players in skill/athletic positions could be productive NBA players. Of course, the fact that Ginobili was the next to last guy picked in the second round, after Leon Smith got their first round selection, indicates that precise foresight wasn't involved, but still, as I peruse the draft charts, after Ginobili and Parker (and the Argentine victory in Athens in 2004) such "free lunches" were off the table. I say Auerbachesque in the sense that Popovich profited from the chauvinism of others in the same way that Red did from racism. And then there was the apparent ability of Popovich to recognize the undervalued category of defense and three point shooting, but that is another matter.

The point is that once you take into account the Spurs' initial endowments (and let's not forget three years of David Robinson in this run) there is a whole lot less than nothing to explain. As noted, just the average contributions of the Big 2, implied 62 win seasons.

Crow wrote:Has anyone calculated the average and peak RAPM by draft pick number or range? I don't recall seeing it. I think Daniel did ASPM for draft picks, which is something useful, but that leaves out shot defense and other indirect impacts.

This would be an interesting study.

*********

P.S. Just to emphasize, if you add in three, prorated years of The Admiral (+.74) and ten, prorated years of TP (+0.43) into the 11 year run, you get an average Spurs lottery endowment of +9.4, what corresponds to 64 plus wins with a supporting cast of 0.

It really, really, really pays to be lucky (and being wise, of course, doesn't hurt).

A supporting cast of "zero" is very, very good; especially if the guys they are supporting are exceptional.
If you can lose your 2 best players and go 41-41, or maintain a lead, that's an elite team, once you've got your best guys back.

Nowitzki and Duncan may have been lucky picks; but their organizations opted to build their teams around them, do what it took to keep them, and acquire players who were complementary.

When superstars are commanding a large part of the available salary, getting "NBA average" supporting players for less-than-average salary is an excellent feat of general managing.

Mike G wrote:A supporting cast of "zero" is very, very good; especially if the guys they are supporting are exceptional. If you can lose your 2 best players and go 41-41, or maintain a lead, that's an elite team, once you've got your best guys back.

Mike, I don't want to get into what is ultimately a sterile, normative debate about the meaning of 0. Yes, if the contributions of all but your two or three best players sum to 0, you are, by definition, an above average team (assuming the "stars" really are). But I don't know how this is relevant, at least, to any points I am making.

Mike G wrote:Nowitzki and Duncan may have been lucky picks; but their organizations opted to build their teams around them, do what it took to keep them, and acquire players who were complementary.

Here I think I completely disagree. Duncan absolutely wasn't a lucky pick. (If he hadn't been drafted first, Spurs management would have been run out of town. I think what you meant to say is that the Spurs were lucky to have won the lottery.) And Nowitzki was a very known factor. I cannot recall from back in the day if he "fell" to his draft position, but I do know that Boston right behind them was disappointed that the Mavs got him (but in the instance, settling for the very nice consolation prize that is Paul Pierce.)

And then as for giving credit to the respective organizations for opting "to build their teams around them, do what it took to keep them, and acquire players who were complementary", this is setting the bar so low as to make the accomplishment meaningless. I think I can say with confidence that there isn't a franchise in NBA history that wouldn't have done the same: made these players the centerpiece, offered the max. contracts, and found other players. (Who exactly isn't complementary to Tim Duncan?)

Mike G wrote:When superstars are commanding a large part of the available salary, getting "NBA average" supporting players for less-than-average salary is an excellent feat of general managing.

It is true that getting average players for below average pay is a sign of above average general managing. And this may indeed be the case, in the case of the Mavs and Spurs. But I suspect, in the case of the Mavs at least (I haven't looked at the Spurs) that this is not true. As noted above, the Mavs supporting cast (defined as not-Dirk and JH) cost a lot of money. This alleged "net 0" group was paid over the initially-specified decade in question $8+ million more per year than the entire salary of the respective median teams (which, competitively, was a slightly above average team: 42+ wins) and to acquire less than 80% of the minutes. And as for the Spurs, their supporting cast, as the numbers above suggest, hasn't been anywhere close to 0. What's consensus "replacement level", again?

To remind, the only reason 0 was introduced into the discussion is that this is the number is what happened to drop out in the initial Mavs calculation. (The second time around it was -0.3, but no matter.)

And the point about the Mavs wasn't the 0, per se, but that having been in such a commanding position analytically (in fact, unique: one never to be seen again, by anyone, ever) and having been willing to spend to win, 0 (or -0.3) is all that was able to be gained. This, to me, is a very striking fact and should frame people's opinions on these matters. (If true, that is.)

schtevie, have you ever heard of a thing called confirmation bias? Just asking, because all your posts have exactly that bias. You are trying to "prove" your idea and with that you are looking only for evidence which is supporting your idea. And whenever something is not exactly fitting in, you are interpreting it in a way that it either is supposed to be not meaningful or that it even can be twisted to fit into your idea.

The average supporting cast (team without the respective best player) is cleary below 0, I calculated it to be around -2.5. And that includes the 2nd best player. If you would remove the two best players and you come up with a supporting cast of 0, the job done by the management can be considered really great.

And then again, the salary comparison is not useful, because the CBA is designed to make it more expensive for better teams. That should make it possible that worse teams can get better more quickly in order to have some sort of competitive balance. You are still ignoring that fact.

...all that was required of a "supporting cast" was 0. This to me is rather awesome, considering the circumstances and in the implication for the future.

Again, here was a franchise, having informational/analytic advantages never before seen and never to be seen again, with top-down pressure to have these advantages realized within the organization and willingness to put money where the numbers were, and the net effect was, in a very real sense, zero: no net positive performance.

Yes, if the contributions of all but your two or three best players sum to 0, you are, by definition, an above average team (assuming the "stars" really are). But I don't know how this is relevant, at least, to any points I am making.

Is the definition of zero and the awesome implication a big deal, or is it not?

I was first mystified why you'd remove Nowitzki from your analysis and then proceed to see if the Mavs have done well aside of him.
Then you included Josh Howard, and said, well, other than those 2, they haven't done all that much.
Then you said the rest of the cast had been a merely NBA-average bunch, and this was equivalent to "no net positive".

So I'm sorry, but I can't tell what it is you're saying.

mystic wrote:schtevie, have you ever heard of a thing called confirmation bias? Just asking, because all your posts have exactly that bias. You are trying to "prove" your idea and with that you are looking only for evidence which is supporting your idea. And whenever something is not exactly fitting in, you are interpreting it in a way that it either is supposed to be not meaningful or that it even can be twisted to fit into your idea.

If you wish to do more than parrot "confirmation bias", it would be helpful to provide particular evidence about its magnitude (never mind the sign).

The entire point of this post was to determine the extent to which "analytics" has had practical effect in the NBA. Not whether it ever provided interesting and true information (it surely has, and we can talk about that); but whether such information has ever brought "significant" competitive advantage, on net, over time. Toward that end, the historical analysis necessarily must begin with the Dallas Mavericks, for all the reasons that have been repeated. Perhaps there is another club to look at, as well, but none that so fully meets the criteria as the Mavs.

So, in determining the possible effect of "analytics", the first step is deciding upon an analytical framework for appraising competitive success. I chose (R)APM, which is at least internally consistent in that something along these lines is the same basic framework the franchise in question uses.

The next step then is to subtract out the contribution of all "non-analytic" inputs. It seems to me that initial talent endowments are an obvious and uncontroversial debit. Mark Cuban inherited Dirk Nowitzki. End of that story, really, unless one wishes to argue that some significant fraction of Dirk's dirkness is owed to Mark Cuban's special sauce. (What is an interesting, separate, potential line of inquiry: whether players when on the Mavs were better, in RAPM terms, taking into account aging effects, than when not.)

And as for "confirmation bias" on this account, to the extent that it exists, it almost surely cuts the other way. As we are told, RAPM biases estimates to the mean, and I suspect that an unbiased, APM estimate, over the same eleven years, would show DN's contribution to have been slightly more than the +6.9 over the years in question.

The next "non-analytic" input to subtract is positive draft surprises. This is a fair debit, conceptually, but one can disagree about its measure. As for whether the dread "confirmation bias" horribly distorted the picture I painted is for the reader to decide. My intent was to fairly represent and account for this factor. If the gentle reader wishes to attribute the success of Josh Howard, in full, to the foresight of the Mavs front office, as opposed, in the other extreme, entirely to luck, fine. But that wouldn't be right either. If so, Marc Gasol and not Nick Fazekas would have been a Mav (and with the Spurs, Manu Ginobili would have been taken in the first not second round, and Popovich would have moved heaven and earth to move up in the draft order).

So, the specific issue at hand is what portion of JH's +1.1 contribution shouldn't have been subtracted. Can it really be that large? If so, this implicit assumption is that drafting is a precise science, which we all know not to be remotely true, right? And then, for consistency, what about adding back a bit of JH's RAPM undervaluation with respect to APM? Do we have much in the way of confirmation bias to this point? And which sign is it, again?

But I digress, if it makes one happier, consider the argument as being about the practical effect of analytics, net of initial endowments and lottery success.

So, where does that leave things? Well, once you have the long-term team performance net of the debited inputs, account needs to be taken of all possible other explanatory factors. And I think I was being all-inclusive in summarizing these as aging and injury effects. Is there another category? If so, please advise. And if there is confirmation bias in the way I handled these issues, feel free to identify it. Specifically:

* Please identify the excess injuries that the Mavs faced over the ten/eleven years in question, and it is the excess that matters. I am quite confident that you cannot, but you can try. And as noted, again, I think my "confirmation bias" cuts the other way. My reading of the tables suggests to me that the Mavs were likely less injured than the average NBA team.

* Please identify how I might have mishandled the discussion on aging effects. I provided an estimate based on Daniel's Adv. SPM approach which suggests an upward hill to climb, then contrasted this with my calculation based on some of Jeremias' recent RAPM data, which suggests that the hill could well have been flat. I chose the latter (tentatively) for the reasons stated, and let me just say that I look forward to any future estimates that will identify the sign and magnitude of this confirmation bias.

Taking all these explicitly noted and discussed factors into account (again, lower-bound initial endowments, lottery "luck", the dearth of excess injuries, and a possible peak age of 28 or above) and you get the net contribution (in an RAPM framework) of the Mavs "supporting cast" to be approximately 0 (-0.3).

So, the question then turns to whether 0 (conditional on all that was explicitly stated) is a big number (or an awesome one...)

mystic wrote:The average supporting cast (team without the respective best player) is cleary below 0, I calculated it to be around -2.5. And that includes the 2nd best player. If you would remove the two best players and you come up with a supporting cast of 0, the job done by the management can be considered really great.

And then again, the salary comparison is not useful, because the CBA is designed to make it more expensive for better teams. That should make it possible that worse teams can get better more quickly in order to have some sort of competitive balance. You are still ignoring that fact.

Again, I am very interested in seeing persuasive evidence that "analytics" had significant, positive impact on the Mavs' competitiveness over the eleven year stretch in question. Let's say, in that regard, that I share your confirmation bias.

Toward that end, identifying the correct "supporting cast" baseline is the first step. But ultimately, salary considerations must be taken into account: it is an input into the competitiveness production function, and cannot be ignored. But from these and previous remarks, I get the impression that you are unwilling to admit that Mark Cuban's money had anything to do with the Mavericks competitive fortunes. Very strange. Confirmation bias, perhaps?

And finally, about the awesomeness of 0 (whether it be a big or small 0), maybe there is some value in elaboration. Again, historical context is everything. This was a time (the only time, one never to be repeated) when such an informational advantage was available to a single franchise. With every realized transaction, the Mavericks had a huge advantage over (almost) every other team in the league. On each, there must have been a reasoned belief that they would get the better of it, in terms of +/-. And if each transaction (and there were many of them over 11 years) yielded, in expectation, a non-negative return, it is in this context that I find a final tally of (approximately) 0 kind of awesome.

schtevie, I neither have the time nor the interest in disproving your idea, also because that is not how science works anyway.

The issue here is that you don't have sufficent informations. You neither know which teams are using analytics for their decision making process nor do you know to which degree the decisions are influenced by analytics. Even the Mavericks have not made their decision based alone on analytics. Take their interest in Deron Williams as an example, in which non-analytical stuff seemed to have played a more important role. Various statements by Cuban can be interpreted as evidence that the analytics would have suggested not to pursuit him at all. Thus, even in your prime example you may have noticable influence of non-analytical approaches. On the other hand the Mavericks were clearly better than league average during the last 12 years, mainly based on the fact that Nowitzki yielded such high impact. But a non-analytical approach might have lead to the conclusion that Nowitzki is actually not a good piece to build around, an idea very often presented in the media due to Nowitzki atypical game for a power forward. How much do you think was the Mavericks approach (building around Nowitzki specifically from 2004 onwards) actually based upon the +/- based analysis by Winston&Sagarin? A team without such analysis would probably decided to either trade Nowitzki or try to change his game in order to accommodate a rather traditional style of play with likely less success.

And obviously, the way you want to "prove" your idea is biased. You don't have an established method in order to distingush the decisions made by the organizations into their analytical and non-analytical part. You just look specifically at things in order to find prove that analytics do not have an positive impact.

mystic wrote:schtevie, I neither have the time nor the interest in disproving your idea, also because that is not how science works anyway.

If you have neither the time nor interest in engaging the empirics of the argument, fine. But honestly, the appeal to authoriteh on the workings of science, that is a bit embarrassing.

mystic wrote:The issue here is that you don't have sufficent informations. You neither know which teams are using analytics for their decision making process nor do you know to which degree the decisions are influenced by analytics. Even the Mavericks have not made their decision based alone on analytics. Take their interest in Deron Williams as an example, in which non-analytical stuff seemed to have played a more important role. Various statements by Cuban can be interpreted as evidence that the analytics would have suggested not to pursuit him at all. Thus, even in your prime example you may have noticable influence of non-analytical approaches. On the other hand the Mavericks were clearly better than league average during the last 12 years, mainly based on the fact that Nowitzki yielded such high impact. But a non-analytical approach might have lead to the conclusion that Nowitzki is actually not a good piece to build around, an idea very often presented in the media due to Nowitzki atypical game for a power forward. How much do you think was the Mavericks approach (building around Nowitzki specifically from 2004 onwards) actually based upon the +/- based analysis by Winston&Sagarin? A team without such analysis would probably decided to either trade Nowitzki or try to change his game in order to accommodate a rather traditional style of play with likely less success.

It would appear that among other things, there has been a failure to communicate, and I must take my share of responsibility for this.

First, It is apparent that a fundamental aspect of the argument I was making is completely misunderstood. I was/am specifically not trying to disentangle the analytic from the non-analytic. Indeed, that is the underlying point. I was trying to account for the net effect of analytics, necessarily having had to pass through the gauntlet of "tradition", because that always has been the case and will be for the foreseeable future.

The importance of the Mavs example in this regard is that this was the most fertile ground for the potential benefits of analytics to bear fruit (as was stated and restated). The guy cutting the checks was hiring folks who were supposed to be sympathetic with these perspectives and telling people to listen up, hence traditional pushback can reasonably be assumed to have been lower than elsewhere. If the resulting underlying score was analytics +5, and the dead hand of tradition -5, netting to 0, to a first approximation, I don't care; the point is made.

But wait, it gets worse! Nowadays the informational near-monopoly that Mark Cuban enjoyed is now gone, at least as regards the performance of established players (perhaps one franchise has concocted some new special sauce regarding draft valuation that will obtain some significant, but inevitably ephemeral advantage, perhaps it's even the Mavs, but I don't think we've seen such results, as of yet, have we?)

And second, it is of course important to fully tell the tale, to account for the experiences of other teams with an analytics bent within the time period in question. No argument there. More is generally better. So, go ahead and make an effort in that regard! But the overarching point is that the Mavs story over 11 years should yield results that are clearer than the truth, given its special circumstances (If not them, who? If not then, when?) and they don't appear to.

And then as far the last half of the paragraph is concerned, frankly, I simply cannot follow the train of thought, at least as to how it pertains to the discussion at hand.

mystic wrote:And obviously, the way you want to "prove" your idea is biased. You don't have an established method in order to distingush the decisions made by the organizations into their analytical and non-analytical part. You just look specifically at things in order to find prove that analytics do not have an positive impact.

Sigh. See above, and above, and above, and above.

schtevie, you still seem to have the impression that I want to argue against your idea. But that is simple not the case, because there is really nothing to discuss here at all. If someone comes up with a conclusion first and then is trying to justify that conclusion by using just a really small amount of data while twisting the data in a fashion that it is in agreement with his previously made conclusion, he is really doing it wrong. If you can't even grasp that, it makes no sense to discuss anything at all. My advice: Go back to the drawing board, come up with a method which can determine the influence of analytics, then collect the necessary data for all teams and then draw a conclusion based on hard evidence. Anything else is just a fruitless game of rhetorics, a game someone is playing, if he is really not interested in the truth, but rather trying to win a debate.