Now that Lebron is back to being “the guy I always kinda rooted for”, and Bulls fans can say “73 wins mean nothing without a ring” the ever-so-long NBA season has come to a close. And with the close, we all know what comes next, yup, you guessed it: NET CLOSE WIN/LOSS analysis. Yes, you saw my research on the NHL, and couldn’t get enough, I get it… So, by (popular?) demand, here is what I found using similar analysis with the NBA. Below you’ll find a brief excerpt from the previous post explaining the study, what it can tell us about the next season and why I was so accurate:

(Skip past the small italics sized print if you read the NHL article (unlikely), and/or you really don’t care, and just want an answer (likely))

*…All things in life have certain mathematical rules, guidelines and properties they must follow. Whether you know them, don’t know them, or like most, were more concerned about how to score beer/get the cute chick’s number across the room rather than paying attention in math class, they exist, all around us, 24/7/365. One such guideline, is the “tendency (or regression) towards the mean”. If you are unfamiliar, not to worry, it’s very easy to understand. Wiki describes the phenomenon: that if a variable is extreme on its first measurement, it will tend to be closer to the average on the second, and third and so on. Very simply put, things want to go back to normal. If you’re stopped at a red light for 10 seconds, over the next 10 seconds it’s much more likely that it will be green RATHER than stay red (ignoring orange/yellow light). If it’s an unseasonably warm day for 3 days straight, it’s much more likely (all other things being equal) to be cold the next. The most famous way this was described (fancy pantsers call it: experiments used to demonstrate the empirical validity of the fundamental law) was by John Edmund Kerrich. Johnny boy proved the “Law of Large Numbers”, which it’s also known, by tossing a fair ping pong ball 10,000 times and recorded what side it landed on while being “interned” in a Nazi camp before Tom Hanks came for Matt Damon. The theory states, the larger the sample size, the closer the outcomes will come to the predicted probability. Think of it this way; if you toss a quarter in the air, we all know there is a 50/50 chance you’ll get a tail (assuming it’s a perfectly fair coin). But, let’s look at this under a microscope. Let’s say you want the outcome to be “tails”. We know for certain, you have 50/50 shot of it coming up your way with one toss. BUT, if we were to toss it 5 times, although each toss gives us an independent 50% chance of hitting tails, it’s not absurd if it hit “heads” all 5 times is it? Of course not, even though the odds of that are 3.125% (probability ^ instances, or .5x.5x.5x.5x.5)…what if we did it 50 times, all heads: that would start to get crazy…how about 500 times, you’d have to be cubs-fan-thinking-it’s-this-year-crazy to imagine that would happen (still technically possible, because each time you toss the coin, it’s not like the coin knows about the previous toss(es), that’s called an event being mutually exclusive, but now we’re really getting boring). So, as you can see, whether you know about the regression toward the mean or not, you’ve probably seen it, maybe even understand it, even if you ended up figuring out how to get beer during that lecture years ago.*

*Now, the “tendency towards the mean” can easily be applied to sports (like literally everything else). In any sport, really in any application of the sport itself (scores made by one ball club, wins over a course of a year, interceptions Jay Cutler has thrown this week…or this quarter of the game, since it’s Jay) you can see this in motion, and more important, use it to your advantage, particularly when predicting outcomes/handicapping. Skill and collective talent of a team, its coaches and their front office clearly plays a huge role in club success. But, what about those smaller factors that are collectively usually referred to as*

__luck__? The way a football bounces, a momentary 10 mph shift in wind, etc. These factors are sporadic, intangible and unfortunately unable to apply directly to prediction. However, we can take a back door towards it. In my study below, the first of a 3 part series, I look at how closes wins or losses by a team, correlate to a decrease or increase respectively, to the team’s record next year. The basic idea is when you win by a large margin of victory, it’s clear you’re the better team (in most occasions, while close wins, could have been determined by a “margin of error” (“sporadicy”, luck, etc.), and I intend to turn that into an advantage for my readers. For the first segment, we’re going to take a look at my better half’s favorite sport: hockey.

*Hockey, relative to the other “big 4” sports, give us a unique look at “closeness” of wins and losses. Since, hockey games can end in both an overtime win/loss (when 3 periods pass and neither team has outscored the other, the game goes to a sudden-death additional period), and shootout win/loss (if both teams are still tied after the extra 20 minute OT period) beyond the typical win/loss that occurs during a standard 3 period match. So, instead of looking at games that ended in a 1-2 goal victory/defeat, which would make yours truly stay up another 24 hours digging for the much more unavailable data, the OT and SO data provided me an easy way to investigate these scenarios.*

*I looked at the records over the past 10 years of every team in the NHL, and derived a very simple formula awarding one arbitrary point for an overtime win, and 2 for a shoot-out win (and using the same concept, minus one and two for losses respectively). The basic theory is the “luck factor”, that is, the event moving a team away from the norm, is twice as significant if you win in a shoot-out victory, as it is if you simply win in an overtime game. Then I would take the sum of the two products and award each team a score, based on the “closeness” of their respective outcomes the past year. I looked at the aggregate of those numbers (by grouping the teams with 5-10, 11-15 and 16 or more net wins or losses in a year, and correlated that to the upcoming year’s results for those particular teams) to provide me with insight on how this helps provide info if a team is going to move up or down next season.*

*These wins/losses that could be caused by such sporadic instances (relative to a larger margin of victory/defeat) as a quarter of an inch projectile of a slap shot (as coach Bombay would tell Charlie), an invisible piece of ice shrapnel, etc. for example. All of these random items and an incalculable amount of others could be the cause of a close win or loss. And of course because of the tendency towards the mean, these items, that lead to “luck wins and losses” will always, eventually, move the other direction, back towards the mean…*

**New Stuff/NBA**

As I suspected, the negative correlation between Net Close Wins (Loss) to next season losses (wins) is even stronger for the NBA than hockey. Why?

There are more points scored in a basketball game than a hockey game. Hence, a “close game” in b-ball is even CLOSER than in hockey. The “closer” a game, the higher “luck/random factor” (see above). Its known as “Weber’s Law” (I think, pretty sure I read that somewhere, or I’m making it up) Weber’s Law (essentially/or in my reality) states the larger the quantity, the less impact an additional quantity will bestow on it…for all the meatheads out there, think of it like this: you bench press 225 (or 2 plates, in crossfit lingo), but today you’re going to stretch it out, and you add 10 lb.s. It will be harder no doubt, but you probably won’t *feel* much of a difference. But then, you decide to do a set of wrist curls. You typically grab the 20 lb. dumbbell and rep it out (likely looking in the mirror), but today you add 10 lb.s. Same additional amount as with the bench press attempt, but since this is a 50% increase, rather than a 4ish% increase, its a much tougher set–Weber’s Law. So, in other words, although the “closeness” of a game still stays constant, that is 1-3 points, the overall quantity of points scored in a game (NHL v. NBA) is greater…so close games in b-ball, are even *closer* than they are in hockey.

**Findings**

You can see the full data set, and how I came to this conclusion by clicking here. After looking at the past 10 years in the NBA, I found 38 examples of teams that had won or lost 6 or more net close wins or losses. Why 6 you ask…honestly, no particular reason, its just as far as I went, (I’m tired and the cubs are about to start, lay off me). I’m sure you can find correlation with 5, 4 etc. losses as well, but 6 or more was decided to be the start. And of the 38 total examples, I found 29, or 76% of the occurrences to support my prediction, of varying degrees, of course. Here is a the conclusion section of the research:

-8 or more | W | Close Record | Net | ||

Next Year Change (plus = additional wins) | |||||

07 Boston | 24 | 4-12 | -8 | 42 | |

08 Memphis | 22 | 2-11 | -9 | 2 | |

08 Miami | 15 | 3-11 | -8 | 38 | |

09 Sacramento | 17 | 2-14 | -12 | 8 | |

15 Indiana | 38 | 4-12 | -8 | 7 | |

15 Miami | 37 | 4-10 | -8 | 11 | |

15 Phoenix | 39 | 4-12 | -8 | -16 | |

92 | 6/7 | ||||

-6 and -7 | 13.14285714 | ||||

07 Memphis | 22 | 5-12 | -7 | 0 | |

10 Washington | 26 | 5-12 | -7 | -3 | |

10 New Jersey | 12 | 1-7 | -6 | 12 | |

10 Golden State | 26 | 2-8 | -6 | 13 | |

10 Minnesota | 15 | 3-9 | -6 | 2 | |

11 Minnesota | 17 | -7 | 9 | ||

11 y – Miami | 58 | -6 | -12 | ||

12 Portland | 28 | 3-11 | -7 | 5 | |

12 Golden State | 23 | 5-12 | -7 | 24 | |

12 Toronto | 23 | 1-7 | -6 | 11 | |

13 Orlando | 20 | 2-9 | -7 | 3 | |

13 Minnesota | 31 | 3-10 | -7 | 9 | |

14 Phoenix | 48 | 2-9 | -7 | -9 | |

64 | 9/12 | ||||

4.923076923 | |||||

8 or more | |||||

07 z – Dallas | 67 | 12-3 | 9 | -16 | |

08 Portland | 41 | 10-2 | 8 | 13 | |

09 x – Portland | 54 | 9-1 | 8 | -4 | |

15 x – Brooklyn | 38 | 10-2 | 8 | -17 | |

-24 | 3/4 | ||||

-6 | |||||

6 and 7 | |||||

07 y – Miami | 44 | 13-6 | 7 | -29 | |

08 x – Houston | 55 | 9-3 | 6 | -2 | |

08 Golden State | 48 | 9-2 | 7 | -19 | |

09 x – Dallas | 50 | 10-4 | 6 | 5 | |

09 y – Orlando | 59 | 9-2 | 7 | 0 | |

09 x – New Orleans | 49 | 8-1 | 7 | -22 | |

10 y – Dallas | 55 | 9-2 | 7 | 2 | |

11 x – Atlanta | 44 | 6 | -4 | ||

11 Washington | 23 | 6 | -3 | ||

12 y – LA Lakers | 41 | 10-4 | 6 | 4 | |

13 z – Miami | 66 | 9-3 | 6 | -12 | |

14 y – Oklahoma City | 59 | 10-4 | 6 | -14 | |

14 x – Memphis | 50 | 10-4 | 6 | 5 | |

14 z – San Antonio | 62 | 7-0 | 7 | -7 | |

15 x – Memphis | 55 | 9-3 | 6 | -13 | |

15 x – New Orleans | 45 | 10-4 | 6 | -15 | |

-124 | 11/15 | ||||

-7.75 |

So, as you can see, the data is pretty telling of our story:*(All within the past 10 NBA seasons)*

- Of the teams that lost 8 or more CLOSE games in the previous year, 6/7 of them improved the next year, and on average, by 13 additional Ws (wow).
- 6-7 NCLosses, 9/12 of them improved, and by an average of 5 games
- Of the teams that won 8 or more CLOSE games in the previous year, 3/4 of went “down” the next year, and on average, by 6 Ws
- 6-7 NCWins, 11/15 went down, by an average of 8 Ws

- The Chicago Bulls (6 NCWs) will win less than 42 Ws (16′ Ws), likely 8 less, or 34 total
- The Philadelphia 76ers (6 NCLs) will win more than 10 (shocker, right?…sorry not that sexy), likely 15 Ws.