Process and results « Queen City Hoops

Process and results

First time trying something like this – a stream of consciousness post, where a thought has come to me and I try to follow it for a little while. I do not know how far it will go, or how often it may occur. Feedback or inspiration welcome. Thanks for humoring me. -Brett

“I start with the platitude that one cannot judge a performance in any given field (war, politics, medicine, investments) by the results, but by the costs of the alternative (i.e., if history played out in a different way).” Fooled by Randomness, Nasim Taleb, p22.

The hero ball shot. Thinking specifically of Kemba Walker dribbling out the clock, crossing his defender for a step-back jumper that goes in at the buzzer.

Don’t just consider the make or the miss – what was given up to get to that position. Teams routinely receive the ball with 10 seconds on the clock, and the idea is to win or lose at the buzzer – don’t give your opponent a chance to re-take the lead. Would it not be better to take the shot with 5 seconds with the opportunity for an offensive rebound? Or 4 or 3, need numbers to decide – how long does a specific type of shot take: A couple of seconds for a 3 to complete its arc, but what about a layup? Which has the better chance for an offensive rebound?

That chance for an offensive rebound: It depends on the strength of your own offensive rebounders, the opposition’s rebounding, the type of shot that ultimately is taken, as well as the timeout situation of the opponent. If the opponent has no timeouts remaining, getting the ball back 94 feet from their basket and the clock ticking (in the case of a miss) has very little value. We see the heaves on Sportscenter when they go in, but not the numerous misses.

Could this be modeled as a decision tree? With expected values calculated for the various branches?
Shot with 5 seconds left – Two outcomes:

    Make -

  • Other team gets ball (~3 seconds remaining)
    • Two outcomes

    • Opponents makes – you lose
    • Opponent misses – you win

    Miss –

  • Offensive rebound
    • Two outcomes

    • Last second shot off scramble goes in – you win
    • Another miss and defeat is secured – you lose
  • Defensive rebound – victory is theirs, with less drama since the final attempt came before the buzzer sounded

There is more to it than this simple tree – as previously mentioned, timeouts must be considered, foul situations and the attractiveness of overtime, the likelihood of a whistle being blown, and the personnel in both uniforms and their relative merit in such a situation. And not whether or not they are “clutch”, but their ability to impact substantially what is generally about a 70% probability of a defensive rebound. Shooting percentages are higher immediately following an offensive rebound (this counts as stream of consciousness, as I was recalling something 82games.com has published previously – The value of offensive rebounds).

Playing with this simple probability tree in excel has me leaning against eschewing traditional behavior. Does preferring overtime to defending the opposition for 2 seconds while you hold a lead make rational sense? The longer game and more possessions favors the better team, as it reduces the role of randomness in the outcome. But even this assumes that there is a clear better team in overtime, or that it has not changed as the game progressed, due to injuries or players dismissed due to fouls.

Again, I go back to Kemba’s game winning shot against Minnesota last season. Perhaps the timing is not the issue I have with the process. It is the other part of the process, the execution – a midrange jumper off an isolation play, which is a low percentage, undesirable shot throughout the game, is celebrated as the source of victory. The Bobcats won in spite of this shot, not because of it?

4 comments to Process and results

  • charlottean

    the value of the offensive rebound margin, the turnover margin, and the odd possessions (jump balls, end of quarter 2 for ones/clock management, etc.) decides the games. I bet it’s something like 75% of the time the team that gets the most possessions wins. or the team that has at least 3 more possessions wins.

    that said……..you cannot calculate for an end of the game shot planning for the improbable offensive rebound. in the example you have kemba…..who isn’t the most efficient, but he is better in clutch than many……but GENERALLY an nba team is going to get you around a point per possession. using your possession too early devalues your possession entirely from a probability stand point. if he makes the shot, you just wasted a clutch basket by giving them the ball with 5 seconds left. and you are doing it so that you can get a hypothetical second shot at the basket? your numbers are better net by holding for the last shot.

    using your example of 70% (http://www.basketball-reference.com/leagues/NBA_2014.html less would be more accurate) defensive rebound rate and a rounded up 1 point per possession for nba teams…….

    if you hold for last shot you have +1. if you shoot at 5 seconds you are NET +.6 (kemba’s plus one + .3 (30% chance of offensive rebound) – the .7 (70% chance the opponent gets a shot at their +1))

    you want to create max possessions but you always want to simultaneously limit the opportunity for opponent possessions. shooting with 5 seconds left leaves the door open unnecessarily. the math completely ignored the fact that your % for offensive rebound is closer to 25% or lower if your opponent was given the opportunity to put its best defensive lineup on the floor. and ignoring the fact that your offensive rebounder is getting the ball in a clutch situation with limited amount of time left and there is a % of offensive rebounds that then lead to turnovers when the ball is brought down. and then there’s the variable of the ref swallowing his whistle when the offensive rebounder gets hacked with .6 seconds left on the put back.

    i’m absolutely dead set on take the last shot of each quarter. scheme your entire one minute offense around making sure you get that extra shot up. get good looks, but get to them with the appropriate time.

    • I agree with taking the last shot of the quarter – 2 for 1 to close out quarters is regarded as accepted wisdom at this point. But it struck me as something worth thinking through at the end of the game – what situation would create an opportunity to shift general practices on the timing of the last shot. If the opposition has no timeouts remaining, and a make gives them the ball at the far baseline and trailing with 2 seconds to play – that’s not the same possession that they’ve been dealing with the rest of the game. They’re looking at a shot from midcourt to close the game – is that not better than going to overtime? Even a defensive rebound with 2 seconds remaining and a tie-ball game – a 1 in 10 chance to win the game from halfcourt (again, no timeouts to advance the ball) – with the upside of a chance at an offensive rebound and potentially a tip-in, foul, or a last second shot. Additionally, offensive rebound rates do shift based on the type of shot (and even the player who is taking that shot), which could further influence the percentages used in such a calculation of benefit.

      I’m certainly not entrenched in this position or any of the points I have made – it is just something that was running through my mind and subject to further revision as additional considerations and data become available.

  • charlottean

    The no timeout caveat changes everything. If the opposition has no timeouts, I would definitely side with shooting earlier rather than later. Maybe even as early as 7 seconds on the clock.

    if the game is tied, you take the last shot. can’t lose on that possession. you shoot too early and leave the door cracked unnecessarily (even for that 1 in a million heave) you’re giving away chance.

    It’s like how we won earlier this year (can’t remember the opponent) and they chose not to foul and henderson shot left no time on the clock and the opposing coach was criticized for not fouling and clifford said he would have done the same thing……….even that 2 second heave is better than widening the gap which is what statistically happens when you foul to elongate an NBA game. high school and college kids freeze up way more than the pros do. a LOT of pros are actually better clutch free throw shooters than non-clutch. the video game mentality thinks you’re going to foul, they make 2 and you make a quick 3 but the stats say fts = 1.5 PPP and the 3 ball is 1 PPP in non-clutch and even less in clutch.

    your best option there is to hope for the 1 stop to get the 1 shot. makes it 1 complete tossup instead of a constant widening of the margins.

    these kind of end of game scenarios are about the cleanest analyzing you can get in basketball. there are just way too many variables in most other aspects of the game to gather true value in the stats. i despise individual +/- because of that.

  • Spencer/SDS

    @charlottean

    The opponent you were thinking of was the Raptors, plus nice posts ^_^