Playoff predictions?

[Mike G]

...using all playoff games ends up inflating the true value of HCA...

I just applied the weight of (for example) SAS' 232 playoff games and their Home-Away differential. For each team.
If you win a series with MOV of +7 in (4) home games and -2 in (3) away games, that's still a 9-point H-A differential over 7 games.
If a batter hits .325 against lefties and .250 against righties, it doesn't matter how many of each he faced; he's still .075 better vs lefties.

It seems odd to exclude game-7's, since any series that has gone 3-3 thru 6 should be more evenly matched than a random series.
Yet since 2000, the home team has won 38 of 50 G7's, and 76% is several wins better than the overall 65%.

Skimming off all G7 -- 50 of 1414 playoff games -- leaves Gm 1-6 home Win% at .645, rather than .649

[shadow]

You need to use an even number of games to account for differences in team strength. Otherwise the value you calculate for HCA will be inflated. This is pure fact. I don't know why this is so hard for you to understand. Here's Ed's quote:

"If all playoff games are included, the home team wins by 5.05 points on average. With an uncertainty of 0.46 points, it is crucial to restrict the set of games to account for the differing strengths of teams."

Your baseball example is also laughably bad. You aren't accounting for the strength of the pitchers he faced or the sample size. He could have faced a handful of stud right handed pitchers and a substantially higher number of scrub left handed pitchers.

[Crow]

Useful exchange on the numbers.

Perspectives vary and words can strike different ways. Usually good to try to not take things personal but it sometimes happens.

[AcrossTheCourt]

Ignoring games is silly. You can just weigh those odd games more so it's even:
https://ascreamingcomesacrossthecourt.b ... yoffs.html

[shadow]

And it's still not close to 5.5 points even with that approach, which was the whole point of the argument. If it really was 5.5 points we'd all be rich because we could just blindly bet home teams to cover the spread in the playoffs. Except home teams in the playoffs barely cover more than 52.4% of the time (52.6% according to query linked below), which is the break-even percentage at -110 odds.

http://killersports.com/nba/query?sdql= ... +D+Q+L+%21++

[Mike G]

So this post is not anything more than a table of facts, plus a question about what "home court advantage" means. It seems to be the point differential benefit for the team playing at home.
In the playoffs, if you aren't at home, your opponent is at home; you not only lose the HCA, the opponent gains it. So your Home-minus-away MOV is twice what we normally see as HCA?

Ranked by playoff MOV, here's the rundown of Home-minus-Away for playoff teams, 2000-2016:

Code: Select all

tm     H-A    G   home   away
Cle    7.8   114   7.7    -.1
SAS   10.1   232   8.7   -1.5
GSW   10.8    77   9.0   -1.9
Mia   10.4   181   7.5   -2.9
LAL   10.0   197   7.0   -3.0

Det   10.9   137   7.3   -3.6
Phx    7.9    86   5.0   -2.8
OKC   11.6   114   6.5   -5.1
Bos    7.8   148   4.2   -3.5
Brk    7.7   103   4.1   -3.6

Orl    7.4    78   3.9   -3.5
LAC   10.9    64   5.5   -5.3
Ind    7.1   136   3.5   -3.6
Was    4.1    49   1.7   -2.4
Dal    8.3   148   3.7   -4.6

Sac    9.1    64   3.7   -5.4
Chi    3.0    87    .6   -2.4
Den    9.1    64   3.2   -5.9
Cha    4.3    38    .1   -4.2
Phl    7.2    85    .9   -6.3

Uta   12.0    74   2.9   -9.1
Por    9.3    76   1.5   -7.8
Min    9.0    35   1.4   -7.6
Hou   12.3    74   2.9   -9.3
NYK    5.2    46   -.8   -6.0

Tor   10.6    64   1.7   -8.9
Mil    4.2    58  -2.1   -6.3
Atl   12.3    88   1.9  -10.4
Mem    7.8    71   -.6   -8.4
NOP   10.9    40    .9  -10.0

avg    8.6    94   3.5   -5.2

The average is unweighted; note teams near the top tend to have been in more playoff games; so the majority of teams are 'below avg' in playoffs.

Is it just misleading to say teams are 4 (or 3, or 2) points better when playing at home? Better than what? Almost never are games held on neutral courts. Most teams (25 of 30) are 7 or more points better at home vs away.

[Crow]

Of the 5 higher seeds I suggested might have a tougher time than commonly expected based on performance against top 10 regular season, currently 3 are tied (worse than expected) or behind. Still too early to know how many upsets actually happen. But what other method has identified the targets for upsets better?

[shadow]

My projections indicated the Grizzlies, Bulls, Bucks, and Pacers were potentially undervalued underdogs. Pacers were swept, but were outscored by a total of 16 points. That has to rank as one of the closest 4 game sweeps in NBA history. The other 3 are tied 2-2 so we'll see how those play out.

[AcrossTheCourt]

So this post is not anything more than a table of facts, plus a question about what "home court advantage" means. It seems to be the point differential benefit for the team playing at home.
In the playoffs, if you aren't at home, your opponent is at home; you not only lose the HCA, the opponent gains it. So your Home-minus-away MOV is twice what we normally see as HCA?

Ranked by playoff MOV, here's the rundown of Home-minus-Away for playoff teams, 2000-2016:

Code: Select all

tm     H-A    G   home   away
Cle    7.8   114   7.7    -.1
SAS   10.1   232   8.7   -1.5
GSW   10.8    77   9.0   -1.9
Mia   10.4   181   7.5   -2.9
LAL   10.0   197   7.0   -3.0

Det   10.9   137   7.3   -3.6
Phx    7.9    86   5.0   -2.8
OKC   11.6   114   6.5   -5.1
Bos    7.8   148   4.2   -3.5
Brk    7.7   103   4.1   -3.6

Orl    7.4    78   3.9   -3.5
LAC   10.9    64   5.5   -5.3
Ind    7.1   136   3.5   -3.6
Was    4.1    49   1.7   -2.4
Dal    8.3   148   3.7   -4.6

Sac    9.1    64   3.7   -5.4
Chi    3.0    87    .6   -2.4
Den    9.1    64   3.2   -5.9
Cha    4.3    38    .1   -4.2
Phl    7.2    85    .9   -6.3

Uta   12.0    74   2.9   -9.1
Por    9.3    76   1.5   -7.8
Min    9.0    35   1.4   -7.6
Hou   12.3    74   2.9   -9.3
NYK    5.2    46   -.8   -6.0

Tor   10.6    64   1.7   -8.9
Mil    4.2    58  -2.1   -6.3
Atl   12.3    88   1.9  -10.4
Mem    7.8    71   -.6   -8.4
NOP   10.9    40    .9  -10.0

avg    8.6    94   3.5   -5.2

The average is unweighted; note teams near the top tend to have been in more playoff games; so the majority of teams are 'below avg' in playoffs.

Is it just misleading to say teams are 4 (or 3, or 2) points better when playing at home? Better than what? Almost never are games held on neutral courts. Most teams (25 of 30) are 7 or more points better at home vs away.

Similar methods here:
http://www.espn.com/nba/story/_/id/1920 ... etokounmpo

[Mike G]

... looked at how teams performed at home versus on the road relative to their point differential and the opponent's during the regular season ...

This (from link above, Kevin Pelton) is not what I did, just going for quick home-minus-away differential. In playoffs, we face the same teams at home and at their place; and the home MOV minus the road MOV is all that was needed. It doesn't matter the opponent or the number of home and road games.

Breaking it down further, one could look at each game, the opponent, their normal road or home performance, etc. That's what I call 10 times as much work for 10% more info -- not worthwhile. And it's probably more like 100 times for 1% better.

It does seem curious that while playoff HCA seems to be greater than in RS, the win% difference is even more extreme than the MOV difference.