2023-24 Title contenders

[Crow]

In GM Survey, about 90% think the title winner will be Celtics, Bucks or Nuggets.

Chances for Celtics and Bucks appear overstated to me considering conference logistics. Nuggets are probably about right.

[Crow]

GSW, PHX and MIL were all in highest 6 for inconsistency in regular season according to a measure at nbastuffer.
CIe, Mia and Lal were among the most consistent. Playoff data not available there. Would be helpful to add that. Playoffs may or may not follow the regular season trend. Each will be some mix of noise and signal. Not easy to interpret but probably better to look & try than not.

Consistency and inconsistency can both help or hurt. The impact derives from details deeper than the average.



Match-ups are about the efficiencies, pace, control of pace, consistency, factor matchups at positions, in lineups, coaching....

[Crow]

If you believe that a title winner needs a top 5 player and are strict about it, then it matters what listor lists you use.

Using Darko from last season for now, the 5 are Jokic, Embid, Giannis, Kawhi and KD. Holiday 6, Curry 7, George 8, Tatum 9, Lillard 10.

Different lists, different timing, different strictness and / or consideration of #2s will lead to different interpretations.

Irving 11, Butler 12, LeBron 13, Doncic 14, Harden 15, A Davis 16, Mitchell 17, Porzingis 18.

Who to consider qualified and not?

I am definitely not considering Cavs qualified. Very probably not Lakers or Mavs or Heat.

7 teams are qualified enough for me.

[Crow]

Clippers rising on betting tables but I have some questions / doubts about that. Will the defense get worse? Probably. Will the offense get better than currently? I doubt it.

[Crow]

By SRS, the leader group includes about who was expected, except the Bucks are way out and the Magic are in at the moment.

[Crow]

Somewhere I said rebounding would be important for Warriors.

Currently 5th best on offensive rebounding rate.

[Crow]

K. Leonard playing pretty well but BPM is lowest since sophomore season and FT rate is a small fraction from his lowest.Defensive rebound rate is a career low, as is block rate.

[Crow]

By SRS and a +5 cutoff, there are 10 "possible contenders". By their offensive and defensive efficiencies, it shakes as 4 two-way strong teams, 3 one-way strong and 3 no way strong by these raw efficiencies.

Switch to net rating and a +5 cutoff and it looks like 6 two-ways and 3 one-way teams.

Some differences in the qualifying lists. 6 on both. Funkiest of SRS variance from net ratings should diminish with more games.

Edges to two way teams.

[v-zero]

At this point in the year SRS is total shit unless it has been regularized. I do wish BBRef would regulirize it. That being the case I would look at net rating.

Well, no, I personally wouldn't look at net rating, but if you had to look at one or the other than that's the one.

[Crow]

The various measures will get better by 20 games in. But that is a month from now.

[Crow]

BRef only gives 4 teams more than 5% chance at title right now. Celtics, Sixers, Nuggets... and Timberwolves.

Warriors at 2.2%, Bucks at 0.5%.

The +5 to 6 SRS teams are not holding up in the projection to the +10 to 17s either.

[Crow]

BRef title projections are not that serious / believable for some teams. Thunder at 15%? I'd guess more like 2-5%.

[Crow]

BRef rigidly uses a formula... and gives Nuggets less than a 4% chance of repeating.

The prediction is more important than the formula and is this case both are absurd.

[Mike G]

So the Pels suffered a 44-point beatdown by the Lakers, and it wasn't due to star players being out of the lineup. Mostly just a 38-pt advantage-LA at the arc and FT line. That and 59-42 rebounding mismatch.

In any case, the Lakes gained about 4 projected wins on the year, while NOP lost that much. The teams basically switched positions between #6 and #11 in the West; and it's only one game.

An improved SRS might take some exponential (<1.0) value of each game's pt-diff, rather than the full 44.

Code: Select all

exp   dif
1.00   44
0.90   30
0.80   21
0.70   14
0.60   10
0.50    7

So 44^(0.8) converts this NOP nightmare to only 7 times as bad as a 3 pt loss, rather than 15 times as bad.

Another factor is recency, in which early season games don't count as much. Some team rating systems seem to do this; but it would greatly complicate or contradict things like player BPM, WS etc.

Denver might rate better with blowout deflation but would suffer from recency: They started 8-1 and have since gone 6-7.
OKC looks like they are 4-1 in decisions of 20+ pts.

[Crow]

Not that you have to answer, but do / did you use exponents a lot in your work?