3-to-make-2, 2-to-make-1 (and other ignored FT changes)

[wigglestrue]

In which periods and under what circumstances were players given 3 attempts to make 2 free throws or 2 attempts to make 1? Were the extra free throws counted in the contemporaneous box scores toward a player's overall free throw totals and percentages? What impact (because there is surely some) would a proper reckoning of 3-for-2 and 2-for-1 have on statistical comparisons between players who played in various eras under various rules?

Not even apbr has good info on it.

edit: Deleted off-topic crap.
edit2: Took out "statheads", since it might come off as hostile.
edit3: Changed title to reflect the pre-penalty 1 FT rule on non-shooting defensive fouls.

[wigglestrue]

Bump, after removing the irrelevant ramble

[Paydro70]

...when would this occur? Are you just talking lane violations? Because in that case the first shot basically didn't happen (in the case of a miss). This seems like a tiny effect, unless I'm missing something...

[Tim Lehrbach]

3-to-make-2 was for backcourt fouls, or something weird like that.

[giberish]

...when would this occur? Are you just talking lane violations? Because in that case the first shot basically didn't happen (in the case of a miss). This seems like a tiny effect, unless I'm missing something...

It used to be a somewhat regular occurrence, I'm not sure the years (80's or earlier) or the situations (shooting fouls when in the bonus perhaps).

It would mess up the 0.44 FTA/FGA ratio used in TS%, and perhaps some possession counting. Not by a huge amount, enough that a serious study looking at the years that it was used (not recently) would have to take it into account.

[mysticbb]

Until 1981, if a backcourt foul occured while the fouling team was over the limited, the fouled player was allowed to take 3 free throws in order to make 2. It didn't happen that often to really mess up a thing between 1954 and 1981 when that rule was active.

[Manuel Calavera]

Wouldn't this make a big difference in tight end of game situations?

[wigglestrue]

It didn't happen that often to really mess up a thing between 1954 and 1981 when that rule was active.

It may have happened often enough to certain players to have a significant effect, i.e., Wilt.

The fact that there is so little information about the rule and its effects, I find disturbing.

Another free throw rule from the past, one that may have had more of an effect, is that before a team reached the penalty a fouled player only shot one free throw, not two. Any information from the metrics world on that?

[wigglestrue]

Some of the scant information online about all this:

Not sure how accurate this one is, but it does seem like the rules history on nba.com is incomplete...

http://actionpcsports.yuku.com/topic/96 ... 55-Present

NBA/ABA Free Throw Rules 1955-Present

NBA 1955-1966: 2 FT on FG miss and all backcourt fouls and 1 FT on FG make.
1 FT on non-shooting defensive fouls with possession going to the defensive (fouling) team after the free-throw attempt.
After 6th team defensive foul, 3 to make 2 and 2 to make 1 on all shooting fouls, 3 to make 2 on backcourt fouls, and 2 FT on non-shooting defensive fouls.

NBA 1967-1969: 2 FT on FG miss and all backcourt fouls and 1 FT on FG make.
1 FT on non-shooting defensive fouls with possession going to the defensive (fouling) team after the free-throw attempt.
After 5th team defensive foul, 3 to make 2 and 2 to make 1 on all shooting fouls, 3 to make 2 on backcourt fouls, and 2 FT on non-shooting defensive fouls.

NBA 1970-1972: 2 FT on FG miss and all backcourt fouls and 1 FT on FG make.
A free-throw attempt is no longer awarded on loose-ball fouls if the fouling team is not in the penalty.
Play is stopped and the ball is inbounded near the spot of the loose-ball foul.
If the fouling team is in the penalty, then 2 FT would be awarded on a loose-ball foul.
1 FT on non-shooting defensive fouls (except loose-ballfouls) with possession going to the defensive (fouling) team after the free-throw attempt.
After 5th team defensive foul, 3 to make 2 and 2 to make 1 on all shooting fouls, 3 to make 2 on backcourt fouls, and 2 FT on non-shooting defensive fouls.

NBA 1973-1979: 2 FT on FG miss and all backcourt fouls and 1 FT on FG make.
After 5th team defensive foul, 3 to make 2 and 2 to make 1 on all shooting fouls, 3 to make 2 on backourt fouls, and 2 FT on non-shooting defensive fouls.

NBA 1980-1981: 2 FT on 2pt FG miss and all backcourt fouls and 3 FT on 3pt FG miss and 1 FT on all FG makes.
After 5th team defensive foul, 4 to make 3 (on missed 3PT attempt), 3 to make 2, and 2 to make 1 on all shooting fouls, 3 to make 2 on backourt fouls, and 2 FT on non-shooting defensive fouls.

NBA 1982-Present: 2 FT on 2pt FG miss and 3 FT on 3pt FG miss and 1 FT on all FG makes.
After 5th team defensive foul, 2 FT on non-shooting defensive fouls.

ABA 1968-1976: 2 FT on 2pt FG miss and 3 FT on 3pt FG miss and 1 FT on all FG makes.
After 6th team defensive foul, 4 to make 3 (on missed 3PT attempt), 3 to make 2, and 2 to make 1 on all shooting fouls and 2 FT on non-shooting defensive fouls.

If all that is true, then it was probably more frequent than people having been saying, and consequently would have had more of an impact on stats (more than negligible, that is, more than just on certain players, maybe even significant across the board). Also, the whole 1 FT on non-shooting defensive fouls, possession to the defense thing...why kind of effect might that have? Would it skew possession totals much in a cross-era comparison

The best (not saying much) thread to date about it on RealGM...

viewtopic.php?f=6&t=1124518

According to a couple dudes, all 3 shots on a full 3-for-2 and all 2 shots on a full 2-for-1 would have counted in the box score. But who can say for sure? Anyone?

The sadly underwhelming apbr thread on it...

http://www.apbr.org/forum/viewtopic.php?f=27&t=4035

[wigglestrue]

I wonder if the 1 FT on all non-shooting defensive fouls before the penalty was a 1-and-1 or just plain 1.

[Dipper 13]

After 6th team defensive foul, 3 to make 2 and 2 to make 1 on all shooting fouls, 3 to make 2 on backcourt fouls, and 2 FT on non-shooting defensive fouls.

In the '67 Celtics Sixers Gm. 4 footage, I have noticed the Celtics late in the 3rd quarter fouling Greer just as soon as he crossed midcourt, this way the most points they can score on that possession is 1 (barring a missed FT, offensive board, and field goal sequence). The commentators even referred to them as "strategy fouls". Restore these rules in the modern NBA as well as remove the 3 pt line and you would see team's offensive efficiency plummet.

[wigglestrue]

The answer to the question "Ever Accounted For?" is clearly no...so I edited the thread title, also to include the other free throw changes that have gone unexamined. I think it's kind of embarrassing for basketball stat geekdom to be inventing elaborate metrics without first addressing this most basic kind of counting problem. For all the obsessing over efficiency and possessions, no one knows (or has even seriously estimated) the statistical significance of either the 3-to-make-2/2-to-make-1 rule or the pre-penalty non-shooting-foul single free throw. Geeks, you're better than this.

[penbeast0]

Off topic point . . . I'm not fond of the word geeks, traditionally it meant the gross out artists of the carney who bit the heads off of live animals, put nails in their foreheads and similar things; when applied to social misfits, it means the weirdos who like to do and say gross out things.

I prefer the word nerds, traditionally applied to social misfits who were into school and things like reading and math, to say nothing of Dungeons and Dragons. Seems the more accurate description of statistic obsessed poindexters like us.

Though I admit, I just like reading and talking about them; too lazy to do the work myself.

[Nivek]

The answer to the question "Ever Accounted For?" is clearly no...so I edited the thread title, also to include the other free throw changes that have gone unexamined. I think it's kind of embarrassing for basketball stat geekdom to be inventing elaborate metrics without first addressing this most basic kind of counting problem. For all the obsessing over efficiency and possessions, no one knows (or has even seriously estimated) the statistical significance of either the 3-to-make-2/2-to-make-1 rule or the pre-penalty non-shooting-foul single free throw. Geeks, you're better than this.

Dean Oliver addressed the 3 to make 2 in his book. I don't remember the details, but I know he had an adjustment in his formulas to account for it. Don't know whether that adjustment is reflected in the numbers at b-r. You could contact Justin (the guy who runs the site) and ask him.

[wigglestrue]

Good to know, Nivek! Okay, so Oliver has addressed it. I found some mention of how Oliver accounted for it elsewhere on the net.

http://www.insidethebook.com/ee/index.p ... r_the_nba/

#6 terpsfan101 (see all posts) 2009/03/13 (Fri) @ 21:55

Did the old ABA have the 3 to make 2 and 2 to make 1 rule on foul-shots? The NBA discarded this rule before the 1981-82 season. According to Dean Oliver’s book, the weight for free-throw attempts changes from 0.4 to 0.36 in the possessions formula when the NBA had the 3 to make 2 and 2 to make 1 rule on foul shots.

What I'd like to know is if Oliver was working with the incomplete rules history where 3-to-make-2 only happened on backcourt fouls, or the complete rules history where it was also in play on every shooting foul post-penalty. Since it's Oliver, I'm assuming the latter. I would love to know everything he said about the rule. It can't take up much of the book, right? Did the rule have a greater impact on certain players like Wilt?

Perhaps more impactful on stats than 3-to-make-2 is the single-FT on pre-penalty non-shooting-fouls. That has the potential in cross-era comparisons to seriously skew possession totals (and free throw stats) if not accounted for. Once again, I'm assuming Oliver has done so?

[wigglestrue]

Off topic point . . . I'm not fond of the word geeks, traditionally it meant the gross out artists of the carney who bit the heads off of live animals, put nails in their foreheads and similar things; when applied to social misfits, it means the weirdos who like to do and say gross out things.

I prefer the word nerds, traditionally applied to social misfits who were into school and things like reading and math, to say nothing of Dungeons and Dragons. Seems the more accurate description of statistic obsessed poindexters like us.

Though I admit, I just like reading and talking about them; too lazy to do the work myself.

Just know, I personally use the terms "geek", "nerd", and "stathead" with affection and respect. Usually.

[wigglestrue]

Thanks to Dipper 13, we have more information about how 3-to-make-2 free throws were recorded, and the effect it might have on cross-era comparisons using efficiency metrics.

viewtopic.php?f=64&t=1207876&p=33359175#p33356999

Based on my viewing of Game 6 of the '63 Finals, I can see how FT's were recorded in the boxscore back then. The boxscore had Heinsohn making 6/9 free throws. But in watching the game, he actually made 4/6. This not including the missed shot before the penalty, since it would have given him no potential extra points had he made it to begin with. According to the boxscore Heinsohn missed 3 FT's, when in reality he only missed 2 FT's that had any kind of value to the team. With his three shots occuring at the very end of the game, we have to wonder how often this happened to Wilt's teams, and how it allowed their opponents to close the gap, preventing them from padding their margin of victory (and thus SRS) for future internet observers to gush over.

...I won't even dwell on the blatant inaccuracies below:

Wiley: 0/4 FT in boscore, 0/1 in game footage
Havlicek: 2/5 FT in boxscore, 2/3 in game footage
Russell: 2/5 FT in boxscore, 3/5 in game footage
K.C. Jones: 3/5 FT in boxscore, 3/4 in game footage
Selvy: 1/5 in boxscore, 1/1 in game footage

Of the players who missed FT's, only Sam Jones (1/3) & Jerry West (6/10) were properly recorded in the boxscore.

In those days, every single missed FT was recorded, so if a player was to hit the first one and miss the last two (one of which was a penalty) shot, boxscore would have him at a 33% accuracy during this sequence, instead of 50% as it would be in this era. You have to realize the highest number of possible points a player could score on any given trip to the line was 2, despite potentially receiving 3 attempts in the penalty. What you need to understand is that missing one FT in the penalty does not penalize the team at all, but rather the individual. And since all the possession estimates used for ORtg & DRtg here are 100% dependent on statistics that are already recorded in the box score, this heavily distorts not only the individual's FT% but also the team possession efficiency estimates.

How can something so obvious be so neglected?

Amen.

[Doctor MJ]

The answer to the question "Ever Accounted For?" is clearly no...so I edited the thread title, also to include the other free throw changes that have gone unexamined. I think it's kind of embarrassing for basketball stat geekdom to be inventing elaborate metrics without first addressing this most basic kind of counting problem. For all the obsessing over efficiency and possessions, no one knows (or has even seriously estimated) the statistical significance of either the 3-to-make-2/2-to-make-1 rule or the pre-penalty non-shooting-foul single free throw. Geeks, you're better than this.

I think you clearly mean your tone to be tongue-in-cheek, but it still bothers me.

First and foremost it bothers me because you're right. This is clearly something basic that really should be accounted for, and it's risky for us to be doing too much historical analysis without making such adjustments. So it bothers me because I may have come to atraditional conclusions that will actually end up being beaten by the traditional ones. This is embarrassing.

Second, though: How is it a sign of geek incompetence that the NBA has been prone to rule changes they don't document well, while recording basic data that they can't even get that right? What exactly are the geeks supposed to do have done differently? (Granting that Oliver did something about this, but I think you're right that not everyone doing estimates did.) "Get all your ducks in a row before you give conclusions" sounds like nice advice, but when you don't really know where to look to be sure you've factored everything in, at a certain point you've simply got to make due with what you have.

Third, and maybe most importantly: My concern when I read something like this thread is that people will have a reaction along the lines of "Oh, well I guess we shouldn't listen to the geeks after all", which to my mind should not be the takeaway. To my mind this is like a data analyst being hired as a consultant and then getting reprimanded by the client because the client forgot to tell the analyst something important.

Still, what it does mean is that caution is a virtue. If you see dramatically unexpected results without any causal basis given, you want to be careful running too fast with the results.

Now last, let me spend a little time analyzing the guy who clearly would seem to be the center of this: Wilt. Exaggerating the free throws he takes could inflate pace, which would then decrease estimates of actual offensive effectiveness. If the effect is large enough, then what if the entire trend of shockingly disappointing Wilt offenses is just a quirk of bad data?

Okay, well, the theory here would presumably have to entail that the exaggeration would be larger as Wilt's free throws increased, since that would increase the free throws of the team, which would put his team into the bonus which causes the 3-to-2 change.

Alright, by current estimates, Wilt's team had its huge breakthrough in '67 when he stopped volume scoring. Perhaps much of what happened is simply that the change caused less free throws which then caused less pace inflation. So if we see that the 76ers in '66 had way more free throws than in '67, this becomes a very strong possibility. To the data!

Free throw attempts:
'66 76ers: 3141
'67 76ers: 3411

So...no, the obvious test actually gives us results in the opposite direction: It would seem entirely possible then that we have UNDERestimated how big the improvement to team offense was when Wilt stopped shooting so much.

For the heck of a lit, let's go to Wilt's volume scoring peak in '62, where he shot 500 more free throws than he did in '67 while clearly playing at a much faster pace. Surely at that point the team free throws would dwarf the '67 76ers free throws:

'62 Warriors: 3207

Nope, still no trends along these lines.

Alright so where does that leave us? Same rule: Be cautious. Be cautious of relying too much on data, but also be cautious of ignoring data based on presumption that using incomplete data is worse than using none. No matter the choice, there is risk. The only way to go, is to understand everything and make your expert judgment from there.

[Dipper 13]

Okay, well, the theory here would presumably have to entail that the exaggeration would be larger as Wilt's free throws increased

While Wilt had various roles throughout his career, the two things he was sure to bring to every single team he played on was his rebounding and his foul shooting. Lets not forget how the NBA banned intentional fouls off the ball after 1966. You may be right here seeing as Wilt's FT rate per game increased as his FGA decreased post-Warriors. This does seem to indicate a better offensive system and improved shot selection. I don't see how he could shoot 68% FG on fall away shots.


1960: 0.43
1961: 0.43
1962: 0.43
1963: 0.40
1964: 0.44
1965: 0.42
SFW: 0.39
PHI: 0.47
1966: 0.49
1967: 0.76
1968: 0.68

Be cautious of relying too much on data, but also be cautious of ignoring data based on presumption that using incomplete data is worse than using none. No matter the choice, there is risk.

I assume the formulas used for the estimates have a fixed percentage for offensive rebounding (as well as turnovers). The available footage from Wilt's prime shows he was excellent at taking care of the ball, and since he did not get whistled for many offensive fouls in that time, can we fairly assume that he would have a negative impact on a team as far as cutting down turnovers? Will we dare to call him a "turnover machine" as numerous others on this board have? But that is minor compared to the rebounding impact. Apparently the idea with this fixed % is that Wilt (arguably the top offensive rebounder ever)would have the same impact on the boards vs. his peers year in, year out. His impact on Luke Jackson and his shift to F is a major part of this as well. As much of a reach as it may seem to be, there is no question if anyone was an outlier of this magnitude, it would be Wilt and those Sixers, who had the top rebounding frontline in league history.

[Dipper 13]

This is not as much about the team's total FTA as it is about how often they were being attempted by the career 51% FT shooter (Wilt). And of course, it has to be converted to an estimate (with fixed percentages) per 100 possessions. Given the FTA boxscore inflation, the way FT's were awarded period, as well as the banning of the "Hack-A-Wilt" off ball intentional fouls after 1966, does this not leave fair possible margin for error? Teams would keep on fouling him, dragging the game on forever until the league cracked down. Isn't the difference between a good offense and a poor offense merely one missed shot out of 20 possessions, or 5 out of 100? Another potential error with the fixed percentage estimates (in this case TOV%) may be underrating Big O's impact on the '71 Bucks (if you can believe it). As if he would not have had a major impact in the Bucks taking care of the ball. But then again we have absolutely no idea how well the '69-'70 Bucks took care of the ball, so we cannot statistically estimate the effect of Big O's disciplined ball control with any kind of accuracy.