Paradox Alley

by Jeff Fogle 22. January 2011 00:49

You'd think that when everyone was rested and focused, the best NBA teams would crush the worst teams given the talent disparities from top to bottom. And, you'd think that when everyone playing tonight was playing again tomorrow night, the tendency to save energy would lead to tighter scoring margins. Why go all out to win big when you have to come right back the next night and play again? Do what it takes to grind out a victory and take care of business.

Let me take you on a brief stroll down Paradox Alley.

Several days ago I talked about the importance of emphasizing "double hermit" games on the schedule. Those are matchups where BOTH teams didn't play the night before, and don't play the night after either. The game sits like a hermit on the calendar for BOTH teams. It's as clean a test as you're going to get because there are no fatigue issues, and no distractions.

Monday's holiday schedule saw nine double hermit games. The final victory margins in those matchups were (from low to high): 3-4-7-8-8-9-9-11-14.

Wednesdays schedule saw nine more double hermit games. The final regulation victory margins in those matchups were (from low to high): 0-2-4-5-8-8-9-13-15

Friday, none of the first eight games on the schedule were double hermits. In fact, EVERY single team in action was also playing Saturday night. That's 16 teams playing on night one of a back to back...all matched up against each other. The final victory margins in those matchups were (from low to high): 5-9-14-15-18-24-40-41.

EXPLOSION!

Let me stack those so it's easier to compare:
3-4-7-8-8-9-9-11-14 in Monday's double hermits
0-2-4-5-8-8-9-13-15 in Wednesday's double hermits
5-9-14-15-18-24-40-41 with no double hermits

When everyone was fresh and focused, we generally saw single digit victory margins (even in best vs. worst games like Toronto/San Antonio or Detroit/Boston). The midpoints were in the 8-9 range, about what you'd expect given that it's not a league of parity right now. But, when teams had other business on their mind the next night, scoring margins exploded.

If you fall behind in a double hermit, you keep your starters in and try to come back. If you fall behind on night one of a b2b, it's a bad gamble to try and come back because you could screw up two games instead of one. Many teams choose to take the loss and hope for a split rather than risk falling short and having nothing left in the tank in 24 hours.

So, suddenly, a good team like Atlanta can lose by a ridiculous score of 100-59 at home to New Orleans. Toronto, who hung very tough at New Orleans and San Antonio earlier in the week loses by 40 in Orlando. Utah falls by 24 in Boston, to the same Celtics team who just beat Detroit by four a couple of days ago.

You can see what's happening by looking at the minutes played by starters on those teams who took their medicine...

ATLANTA'S STARTING FIVE MINUTES PLAYED

15-19-30-31-34 Friday
32-33-37-39-45 in last hermit

TORONTO'S STARTING FIVE MINUTES PLAYED

15-23-27-28-31 Friday
29-33-37-38-42 in last hermit

UTAH'S STARTING FIVE MINUTES PLAYED

22-23-23-25-32 Friday
27-31-31-33-41 in last hermit

Big differences obviously.

Let's think about it this way. Who is Atlanta? Are the the Hawks a composite average of all their performances? Should ANY performance count in the evaluation? Or, just the games where starters are generally in the 32-33-37-39-45 range in minutes? Do you even want a laydown in the sample? All that does is tell you what can happen when they call it a night early in a special situation.

A 100-59 loss is going to muck up some averages! Is it a result that really tells you anything meaningful about the team? The team, as its constructed when it's actually going all out to win?  

This isn't that big a deal once 82 games are in the books to give you a full season sample size. It's a bigger deal at the halfway mark. It's an albatross if you're looking over a monthly or two week sample. Why try to analyze a team with an albatross blocking your view? 

If I asked you what to expect from Atlanta Saturday Night in Charlotte, or next Wednesday in Milwaukee, are you going to say "Well, it's going to be a disaster, those guys are 41 points worse than New Orleans"?

No, you'd pretty much throw the game out of the mix. You KNOW Atlanta blew off the second half.

Who is Atlanta? How do you capture that numerically?

Emphasizing double hermits, the cleanest games you're going to get (though only perfectly clean if all the key players are healthy, or not distracted by trade talks), allows you to discard or at least de-emphasize the outliers. Yes, the sample size is smaller. At least it's a relatively clean sampling that approximates the right minute averages for the starters.

As the season progresses on the fly, would you rather trust a dozen double hermit games for a team, or 30-40 games that include a 100-59 tank job? And, honestly, you'd want to give yourself some leeway to exclude outliers in the double hermits if they happen. Basketball can be a messy sport.

Transition Points

*The only double hermit game on the Friday card, by the way, was the Lakers/Nuggets game on ESPN. Will finish this post before that game ends.

*We had three Superleague games on Friday's card. New Orleans, Boston, and San Antonio get wins over Atlanta, Utah, and New York respectively. Let's run the standings through Friday's action:

EAST
Boston 12-5
Miami 7-9
Chicago 5-8
Orlando 7-10
New York 5-9  
Atlanta 5-10

WEST
San Antonio 12-5
Dallas 11-6
Utah 8-7
LA Lakers 5-5
Oklahoma City 8-9
New Orleans 8-10

Pretty good evidence that the West is stronger than the East in the upper echelons. The West is 10 games over .500 in these, the East 10 games under .500 (as all games within a conference would cancel out).

Saturday's Superleague games:
San Antonio at New Orleans
New York at Oklahoma City

(There are no double hermits Saturday). The following teams will be rested vs. opponents playing night two of a b2b:

Dallas at New Jersey
Charlotte vs. Atlanta
Miami vs. Toronto
Philadelphia vs. Utah
Chicago vs. Cleveland
Oklahoma City vs. New York
LA Clippers vs. Golden State

There's only one game Sunday because the NBA doesn't want to compete with Championship Sunday in the NFL. Rested Denver will host Indiana who will be on night two of a b2b.

See you again Monday night...

Bookmark and Share DotnetKicks dotnetshoutout

Double Hermits

by Jeff Fogle 10. January 2011 17:34

One of the most frustrating difficulties of NBA stat analysis is that it's hard to get all the teams on a level playing field so you can be truly confident of your on-the-fly assessments.

*Injuries are rampant in the sport because of the intense physicality of the modern game and the grueling 82-game schedules. You feel fortunate if you see a marquee matchup featuring two high caliber teams squaring off with everyone at 100%.

*Strengths of schedule are varied (particularly in-season), just because of the nature of scheduling...and the fact that one conference is often much better than the other. With unbalanced schedules that have you playing 63% of your games in one conference, this can be a monkey wrench that's very difficult to dislodge from the machinery even after a season is completed.

*Day-to-day and week-to-week schedule sequences are so varied that another serious wrinkle comes in to play. Is San Antonio "fresh" after a couple days off the same as San Antonio "tired" when playing their fifth game in seven nights?  Is Boston on night one of a back-to-back the same as Boston on night two? Does a team have one personality, or a few?

Throw all those together, and it's a jumbled mess. How good is Dallas against a quality opponent on night two of a back-to-back without Dirk Nowitzki in uniform? If the Mavs just go in the tank because they have no chance, should you even count the stats against them down the road? Nobody wants polluted stats.

Are there any "surging" teams who are creating illusions of quality lately because they've run into tired, injuried, or disinterested opponents? How can you trust the numbers? How can you trust your reads?

I don't have a solution, but I have a starting point that could help us focus on certain games, while downplaying others. Maybe that will start the ball rolling toward answering some of these questions.

NBA teams basically play three types of games. Night one of a back-to-back. Night two of a back-to-back. Or, a game that sits by itself on the schedule, with a day off before and after. It seems like there should be more than that. But, there are only three situations in terms of the calendar. In fact, it helps to visualize this by looking at those pocket calendars NBA teams post for their fans. Games are either connected within a back-to-back, or they sit by themselves like a "hermit" on the shedule.

What type of game is most likely to give you a good read on a team?

*Night One of a B2B: At least you know the team is relatively fresh because they didn't play the day before. But, the coach will certainly be looking for ways to save everyone's legs for the two-game tandem. Maybe starters will get fewer minutes than normal. Maybe the whole first team will be pulled in the second half if they fall way behind. Or, maybe the game coming up on Night Two is so important (a divisional rival, a big TV game vs. the Lakers or Heat, a revenge game from a recent loss) that teams kind of phone this one in so they're ready for that one. Trusting the stats from Night One might give you a false read or pollute season-to-date stats.

*Night Two of a B2B: You assume fatigue, but that's not always the case. Maybe the coach did save everyone's legs on the first night. Maybe this is a young team playing relatively early in the season and everyone's raring to go. Of course, on the other end of the spectrum, we've all seen some relative tank jobs over the years. Players can't be asked to play at 100% energy over 82 games. Smart coaches allow the occasional "night off" with an eye on the big picture. You're going to lose some games from late October through April. Accept your loss and move on.

*Hermit: This is as good as you can hope for as a stat analyst. Nobody's tired from the night before. There's no reason to cut anyone's minutes for the next night. What you see is what you get. Coaches can go all out for a win. San Antonio should be close to "San Antonio" as we know them, and Boston is "Boston" to the best degree possible in the hermit games (barring injuries).

So, it seems logical that these hermit games should deserve peak priority in your analytical process. You can use them to determine who a team really is. Then, you can use the back-to-backers to evaluate how that coach or that team handles fatigue, preparation, etc...

Or, in other words, you try to find out:

Who they are (as a baseline)
Who they are when tired
Who they are when holding back because a priority game is NEXT

Full season stat totals are going to represent a mishmosh of those (with further pollution from injuries and strength of schedule issues). See if you can break it down into separate categories that paint a more accurate picture.

Of course, we're only looking at the first team in a game at this point. Their OPPONENTS will also either be in a hermit spot, or on night one or night two of a back to back. They can pollute your read in the other direction!

The "cleanest" games for trusting what the numbers are tellling you are the "double hermits" I used to title this artice. Neither team played last night. Neither team plays tomorrow night. As long as there aren't major injuries keeping stars out of the lineup, these are the closest to "pure" games you can hope for.

I'm posting this today because every game played on Sunday was a double-hermit, and every game set for Monday night is also a double-hermit.

This past Saturday, none of the eight games on the schedule were clean. Tuesday, none of the seven games scheduled will qualify. Wednesday, only three of 12 games will be double hermits.

The results, and stats, from Sunday and Monday will give you as good a status report as you can hope for when trying to determine where teams stand right now. I won't suggest they're perfect. San Antonio played lethargically vs. Minnesota Sunday evening...and we all know the Spurs are more than three points better than the Timberwolves (a 94-91 win). Statheads setting up a separate database that ONLY includes double hermit games for the season may find that subset of games at least reduces the number of outliers, and gives a more accurate read over a certain volume of games.

You'll still have to deal with injuries (Chicago without Joakim Noah is different than Chicago with him). You'll still have to make mental adjustments for strength of schedule (a team playing only double hermits vs. lower rung opponents won't be as good as their numbers suggest). At least you're starting from a place you can trust a bit more.

Double hermits. Wanted to introduce the concept today so we can touch on it periodically the rest of the way. It should be of particular help when evaluating playoff teams. Barring occasional schedule quirks, most every playoff game is a double hermit.

Bookmark and Share DotnetKicks dotnetshoutout
Archive | FeedSubscribe | Log in