Fixing Basketball's Flawed Endgame

Two weeks ago, the NCAA Rules Committee announced a large number of proposals to speed up the game. In particular, a lot of them - reduced timeouts, shorter delay for substituting a player who fouls out, coaches can't call timeout with the ball live, timeouts don't reset the 10-second count - seem aimed at speeding up the final minute or two, which (in close games) is always an incredible drag. Tim Burke at Deadspin tallied up the total real-time span of the final minute of 2014 NCAA tournament first-weekend games, and the results weren't pretty: nearly six minutes of real time for the final minute on average, and the shortest final minute in a game that was single-digits at that point was over five minutes. The numbers are similar for NBA games.

Removing a timeout and reducing incentives for using them will help, but the problem is more fundamental than that. The last minute takes so long because teams have an incentive to foul instead of playing basketball the way they did for the first 39 minutes:

• If you're trailing, you foul to stop the opponent from running out the clock.
• Teams up 3 will sometimes foul to prevent the opponent from tying with a single shot (although, as far as winning the game goes, every study I've seen that actually looks at game results instead of a statistical model says it doesn't help at all).
• In the NBA, a poor free-throw shooter is often fouled away from the play in order to force them to shoot free throws instead of a normal possession. The NBA changed its rules to prevent this in the final 2 minutes, but the strategy still happens prior to then (sometimes even in the third quarter). This year's Rockets-Clippers series was particularly notorious for it.
• There has also been some analysis around the idea of fouling in a tie game when the shot clock is turned off instead of letting the opponent hold for the last shot.
• A paper at the Sloan Sports Analytics Conference this year suggested that the leading team actually benefits significantly by fouling and should start earlier even than the trailing team should. I find the paper completely unconvincing (like most studies suggesting that fouling with the lead works, it assumes that shooting percentages are unchanged by situational factors like "we can't spend 30 seconds trying to find a good shot, we have to settle for whatever we can get quickly" or "the other team needs a 3, let's just guard the arc and ignore the paint"), but the fact that such a strategy was even considered worth studying is troubling.

The rules designating certain acts as "fouls" are intended to discourage them, but they are instead having the opposite effect. No other major sport has this problem:

• Hockey penalties don't stop play until the penalized team has control of the puck, and if they prevent a breakaway attempt or a dangerous scoring opportunity (certain infractions in the crease), the chance is restored via a penalty shot.
• Soccer doesn't stop play for a foul if the fouled team is likely to have a better outcome than a free kick by continuing. Fouls that prevent a clear goal-scoring opportunity are severely punished with a red card.
• Fielder's interference in baseball means all runners advance an extra base, while interference by a runner means he's out and no other runner may advance. The penalty is harsh enough that both are extremely rare, because it's not worth any sort of advantage you might have otherwise gained.
• Penalties in football (other than pre-snap penalties for which there is no play) can always be declined in favor of the result of the play, and football even has a special rule covering "palpably unfair acts" which allows the referee to impose any penalty they deem necessary when someone is abusing the rules (a sideline player interfering with play or repeated penalties to kill the clock).

With those examples available, why does basketball have such a serious strategic foul problem? Until the 1952-53 season, NCAA rules permitted the fouled team to decline free throws and just take the ball out of bounds instead. The problem was that, with no shot clock, teams with the lead could, and often did, stall out several minutes with the ball and just play keep-away. In desperation, the trailing team would try to steal the ball, usually commit a foul in the process, and the leading team would just decline the free throws. Allowing teams to force their opponent to give up the ball by shooting free throws was far from an ideal solution, but it was better than letting the team ahead effectively declare the game over. Trailing teams couldn't abuse this too effectively to come back because there was no 3-point shot at the time. Since then, two major rule changes have made the mandatory bonus less effective at its intent:

• The introduction of the shot clock in 1985 (and its reduction from 45 seconds to 35 in 1993 and pending reduction to 30 now) was a better solution to removing the stall tactics that prompted making the bonus mandatory in the first place.
• Adding the 3-point line in 1986 made the bonus no longer able to award as many points as a "normal" possession, barring a miss and offensive rebound.

As a result of those changes, the bonus is completely inadequate compensation for the possession the opponent would otherwise have. A team that is ahead late in the game would rather run out extra clock, even at the expense of a lesser scoring opportunity; conversely, a team that trails by enough would often prefer a lesser chance of scoring 3 points to a good chance of scoring 2, assuming they take equal time. It's a simple axiom of game design that a penalty which doesn't compensate for the opportunity taken away by the foul will be abused, and you need only watch any close basketball game to see that in action.

How can this be fixed? A couple of out-there suggestions from The Cauldron:

• Either make fouls by teams that trail by 10+ in the final two minutes automatic intentional fouls, or keep a running clock during their fouls. This doesn't really work because, as Burke's article noted, it's not the blowouts you have to worry about for interminable final minutes. If it's close, this doesn't do anything at all, and if it isn't, the end is usually pretty quick anyway. Besides that, it feels gimmicky to have a rule that kicks in specifically based on scoring margin other than "mercy rule" immediate end-of-game or running clock. This also doesn't help with the NBA's Hack-a-Shaq issues.
  
• Cut four minutes off the second half clock, but continue play when it expires and end when someone scores 7 more points than the leading team currently has. This one would probably work (mostly), but it's strange to have a game that's mostly timed followed by an untimed "play to X" segment. It wouldn't completely eliminate strategic fouling discussions either - if you're ahead but the opponent only needs 3 to win, do you risk them ending it on one shot, or do you foul to make it more likely you'll get a shot to win (at the risk of making it easier for the opponent to win if you don't take advantage of that shot)? - but they would be far rarer than they are now.

The Sloan paper mentioned above also suggests a fix, changing the double-bonus to add a third free throw if the first two are made (which eliminates the advantage they claimed for a leading team to foul in most circumstances and delays the point at which it becomes advantageous for the trailing team). While that would help, it still doesn't solve the problem entirely.

My solution is simpler and based on the idea of making sure that the penalty is adequate compensation, whether you value time or points more. We don't need to mess with shooting fouls, as the free throws already provide adequate compensation for the value of the shot attempt. But on a non-shooting foul that would result in free throws, I would fix things by allowing a team with at least one free throw remaining to decline the remaining free throws and take the ball, but without resetting the shot clock. To be more specific, a team's options in the single bonus would be:

• One-and-one as it is now
• One shot, live ball if missed, keep the ball (without a shot clock reset) if made
• Keep the ball (without a shot clock reset)

In the double bonus, the choices would be either two shots or one shot and the ball (with no shot clock reset). The reason for not resetting the shot clock is to avoid a team intentionally trying to draw a blocking foul with five seconds left just to reset the clock.

This allows a leading team to run off just as much clock as they would have without the foul while giving the trailing team a chance to go for a shot worth more points than the free throws would be. Strategic fouling would now be completely pointless for both sides. A team ahead would just take the ball (or one shot and the ball) and continue running out as much clock as the shot clock (and their ability to prevent a steal or turnover) allows them to. A team down 3 would happily attempt one free throw and then inbound needing only a two-pointer to tie. A team tied would be far better off letting the opponent hold for last shot than fouling to try to steal last shot. Hack-a-Shaq would be useless since it would no longer force them to waste the possession on a poor free throw shooter.

How would this affect the rest of the game? The impact in terms of average points on a possession (in normal situations where a foul happens but isn't intentional or strategic in nature) would not be too severe: assuming a typical offensive efficiency of 1.02 points per possession (this year's average in the NCAA) and offensive rebounds on missed free throws are at about 15%, here's how the various penalties look with different quality free throw shooters:

For an average free throw shooter, this change (using the optimum strategy for each in terms of average number of points, which should be the main consideration until near the end of the game) would increase the penalty for a foul in the bonus by 0.19 points, and a foul in the double-bonus would be penalized by an extra 0.27 points. Even with very lopsided foul counts, this would rarely have an effect of more than three points or so - quite possibly less since the majority of the double-bonus fouls in the game are of the last-minute desperation sort that this change would be intended to eliminate; additionally, the change might not be as extreme because situations late in the shot clock would not be as efficient as a full additional possession (potentially dropping the efficiency of the newer options). With poor free throw shooters, the change is more significant; it's still better to foul a bad free throw shooter than a good one, but it is no longer an advantage to do so compared to not fouling at all.

The one potential negative impact I see would be that a late comeback would be harder, as teams with the lead could run out 30 seconds a high percentage of the time and kill the game earlier. However, I don't view this as a serious problem. If it isn't possible for you to make a comeback by playing basketball, I don't see why the rules should allow you to make that comeback possible by deliberately breaking those rules and taking advantage of an inadequate penalty for doing so.

I'd like to hear other ideas for how to get rid of foul-fests or whether this one has flaws that I haven't spotted yet. But the changes that the NCAA proposed two weeks ago, while a good start when it comes to speeding up the interminable final minute of basketball games, don't go far enough to really solve the problem. Let's find a way to make the game 40 minutes of basketball instead of 39 minutes of basketball and 1 minute of a free throw contest.