Author Message
rob c
Joined: 08 Feb 2006
Posts: 14
PostPosted: Wed Feb 08, 2006 3:31 pm Post subject: pace factor Reply with quote
Something has been bugging me for a while and I wondered if you guys may have the answer:
Assuming both teams in a match will try to play their usual game etc. how would one go about predicting game pace?
e.g. If one team's pace factor for the season is 10% above average, and the opposition's is 10% below average, would it follow that, in the absence of any factors which may affect the pace (i.e. gameplan etc.) that the game between the two would be played at "average" pace. Or is it more than likely that the "superior" team would dicate the pace, but not by as much as their usual 10%?
Or, another example. If the two teams playing each other both average 10% over the League average pace, what would be the expected pace of a game between these two teams if the two teams are of equal ability? 21% above average?? (1.1 * 1.1) 20% above? 10% above?
Is there any obvious equation to predict game pace? Has anyone studied this?
Back to top
View user's profile Send private message
Mark
Joined: 20 Aug 2005
Posts: 807
PostPosted: Mon Apr 24, 2006 12:27 am Post subject: Reply with quote
It is an important topic. Are there meaningful general trends or is it very team specific and matchup specific? I dont know the pace actually played for the 1230 matchups. You'd have to compile and sort from the game by game records. That is too much labor for me right now.
I would guess that team win % differential may correlate with getting a pace outcome in accord with team preference at least moderately strongly. Perhaps a pace prediction instead of being the product of the two teams pace / league average might be more heavily weighted toward the stronger team, more dynamically the bigger the win % differential. Of course some teams care more about pace and dictate pace more than others of similar win % but it might be a slightly better way to go.
I looked at the averages for top, median and low 5 team blocks by offensive possession rank and win % rank and found modest impacts. The bottom 5 team block by offensive possession rank actual matchup win % varied only 2% points by opponent pace profile, but the median and high pace team blocks varied by 9 and 10 % points respectively. The top and median 5 team blocks by win % varied by 6 % points on actual matchup win % and the lowest team blocks by 9 % points.
Home court and recent winning/losing streaks might have some impact on dictating pace.
Back to top
View user's profile Send private message
edijorj
Joined: 21 Apr 2006
Posts: 30
PostPosted: Mon Apr 24, 2006 10:27 pm Post subject: Reply with quote
I think it's unlikely that either the faster or slower team would dominate the pace, since the distribution of team pace is not skewed, at least not very skewed. I have not looked at the game pace data, but my guess is
game.pace=team.pace+opponent.pace-league.pace
So, game with a 5% slower team playing a 5% faster team would run at average pace. And a game with a 5% faster team playing another 5% faster team would run about 10% above average pace.
Back to top
View user's profile Send private message
Neil Paine
Joined: 13 Oct 2005
Posts: 774
Location: Atlanta, GA
PostPosted: Mon Apr 24, 2006 11:06 pm Post subject: Reply with quote
Or what about...
Game Pace = (Team A Pace / Lg Pace) * Team B Pace ?
Ken Pomeroy suggested this for tempo predictions in college games:
Quote:
One question I will answer in this spot is how to make a tempo prediction: Multiply the adjusted tempos of the opposing teams and divide by 68.
...Which is the same as my above equation.
Back to top
View user's profile Send private message Visit poster's website
Mark
Joined: 20 Aug 2005
Posts: 807
PostPosted: Tue Apr 25, 2006 12:42 pm Post subject: Can pace matchup / outcome be understood better? Reply with quote
Those methods are out there and pretty simplistic and not that satisfactory and I think the goal would be to try to refine the method.Getting a single pace prediction is unrealistic, setting a probability distribution table is more doable. Perhaps the method could be something like this:
Look at teams game by game record and divided it into high, average and low pace opponent groups and compute the % of games in each group played at at least the 3 pace levels or perhaps 6- very high, high, higher than average, average, just below average and low pace.
Then for an upcoming matchup look at the correct pace classification group distribution for each team. Perhaps increase (by 25, 50% or double?) the weight of games against teams with a similar win % (over or under .500 would be a simple approach), assuming that the pattern against similar win level teams is more indicative of what will happen. Or maybe use that data exclusively and ignore the data from the dissimilar win % teams? If the teams have meet previously in same year I'd give a large extra weight to that (4 times? more?) as the best indicator of what might happen but you probably don't want to rely just on a few datapoints.
Weight the two distribution tables with these factors- team x win% / (team x win% + team y win %) and team y win% / (team x win% + team y win %) and sum into one distribution table. The weighted average blending of the two distribution tables would be your expected distribution of pace outcomes. You could collapse it into a single number average expectation (by assigning a pace number to each of the pace range descriptor labels) if you wanted to but I prefer the distribution table myself.
This Pace pattern overlap distributon table approach could offer a little more guidance on pace outcome but individual game pace strategy is still a significant unknown variable. Scratching below overall pace you can look at ability to fastbreak, tendency to shoot early or late in half court and defend against those. You can look at the teams shot clock distributions and even look at quality of defense on key scorers to get some additional clues about how fast/well the offense might run. But then you can't predict how coaches and point guards will act /respond to defensive challenge that particular game and on specific plays with consideration for junk defenses, referee impact, who is playing well/not, etc.
In the end you may not be able to get that much farther with an experience based distribution approach- there is still a pretty good chance that the game pace will break any of three ways and we are not really able to say strongly which, just (maybe) which is slightly to somewhat more likely. It isnt much more than a 10% variation in team average paces but pace pattern/preference, pace choice/action and pace outcome can make a difference in a number of close games so trying to get a better handle on it matters and could be worthwhile alongside all the other aspects you can try to influence.
And really pace as a whole game average measure may not be as important as managing pace game section by game section. When the offense is flowing, when it is not, in runs for you or against you, when you are ahead or behind on score, what part of the game you are in, if your team is tired (because of schedule) or if you have foul trouble, etc. All this matters in the moment and I dont know how much you react to it or keep an overall game pace strategy intact. That is coaching discipline and feel; teams that are patient or opportunistic. Hard to know which to be but the better teams guess right more often.
I might check out a few teams game by game patterns in detail later for more clues into understanding of the pace matchup / pace outcome relationship.
Back to top
View user's profile Send private message
Mark
Joined: 20 Aug 2005
Posts: 807
PostPosted: Wed Apr 26, 2006 11:29 pm Post subject: Study of Suns Reply with quote
I looked at the Phoenix Suns, the fastest paced team, and found that their opponents ended up playing above their average team pace about two-thirds of the time and below it one-third while Phoenix played under their average pace about 45% and above it about 55%, suggesting that Phoenix, a good team, had more than 50% influence on pace outcome compared to the I assume near average mix of opponents, as I thought possible.
Phoenix was 6-4 W-L when "forced" to play 5 or more posessions below their average pace, 16-2 when only 0-5 below their average pace. 10-5 in games played 0-5 possessions above their average, 10-6 in games played 5-10 posessions above, but only 12-11 in games played 10 possessions above their average.
Splitting the data into 3 average pace groups, Phoenix was 11-11 against teams with average pace below 89 possessions and the average offensive possessions actually played by Phoenix was around 93. They were 6-3 when the actual stayed below 89, 1-2 when it was between 89 and 91.5, and 4-6 when it was above 91.5.
They were 21-9 against teams with a pace of 89-91.5 and the average actual offense=ive possessions was around 90. They were 8-2 when the actual pace stayed below 89, 3-2 when it was between 89 and 91.5, and 10-5 when it was above 91.5.
They were 24-8 against teams with an average pace above 91.5 and the average actual was around 98. They were 7-1 when the actual pace stayed below 89, 2-0 when it was between 89 and 91.5, and 15-7 when it was above 91.5.
So against the slowest pace teams Phoenix playing above league average offensive posessions did worst of the three groups and about the same against average and high pace teams. Figures. They did very well in low posession games (surprising), better % wise than in high pace games overall. Average posession games was their weakest performance but rare.
Last edited by Mark on Fri Apr 28, 2006 6:50 pm; edited 2 times in total
Back to top
View user's profile Send private message
cherokee_ACB
Joined: 22 Mar 2006
Posts: 157
PostPosted: Thu Apr 27, 2006 2:32 am Post subject: Reply with quote
Do you think is it worth differentiating between offensive pace and defensive pace? Sure, you cannot entirely decouple them (a high OR% would tend to decrease both; while steals increases them), but it could be interesting nevertheless. And it shouldn't be that difficult to compute from play-by-play logs; I'm sure 82games has those numbers.
Back to top
View user's profile Send private message
Charles
Joined: 16 May 2005
Posts: 134
PostPosted: Thu Apr 27, 2006 8:53 am Post subject: Re: Pace study of Suns Reply with quote
Mark wrote:
Phoenix was 6-4 W-L when "forced" to play 5 or more posessions below their average pace, 16-2 when only 0-5 below their average pace. 10-5 in games played 0-5 possessions above their average, 10-6 in games played 5-10 posessions above, but only 12-11 in games played 10 possessions above their average.
I used the formula (FGA + .44 * FTA + TO - OREB) / MIN * 48, averaging offense and defense and am unable to replicate these results.
What formula do you use to determine pace? What should Phoenix' average pace for the season work out to? Thanks.
Back to top
View user's profile Send private message
Mark
Joined: 20 Aug 2005
Posts: 807
PostPosted: Thu Apr 27, 2006 10:48 am Post subject: Reply with quote
I used the average pace reported at basketball reference.
I simplified the handling of offensive rebound term in calculating actual offensive possessions for speed (different from b-r.com but same as your formula but since you are using your method for both average and actual this is a source of variance between what I reported on pace and what you are calculating) and I also assumed (a shortcut) actual offensive possessions was close enough to actual pace and a comparison to average pace. I relabelled it actual offensive possessions. I didnt think that it would affect the results that much over the season and I was just after a one team general impression but the possibility of some variance in offensive and defensive possessions and resulting pace is noted. A more careful method (with exactly the same formulas and metric comparison) would report somewhat different results but would likely show pretty much the same big trends. The opponent game data is available at hoopstats.com, maybe I'll redo it later, but if you want to do it you are welcome to supercede what I've reported. I was mainly sketching out an approach, making a start to look at basic trends but if I do more on pace analysis I'll make these adjustments.
Last edited by Mark on Fri Apr 28, 2006 6:53 pm; edited 1 time in total
Back to top
View user's profile Send private message
Kurt
Joined: 10 Jan 2005
Posts: 30
Location: Los Angeles
PostPosted: Thu Apr 27, 2006 7:13 pm Post subject: Re: Pace study of Suns Reply with quote
Mark wrote:
I looked at the Phoenix Suns, the fastest paced team, and found that their opponents ended up playing above their average team pace about two-thirds of the time and below it one-third. Phoenix played under their average pace about 45% and above it about 55% suggesting that at least they, a good team, had more than 50% influence on pace outcome as I thought possible.
Phoenix was 6-4 W-L when "forced" to play 5 or more posessions below their average pace, 16-2 when only 0-5 below their average pace. 10-5 in games played 0-5 possessions above their average, 10-6 in games played 5-10 posessions above, but only 12-11 in games played 10 possessions above their average.
Splitting the data into 3 average pace groups, Phoenix was 11-11 against teams with average pace below 89 possessions and the average pace actually played was around 93. They were 6-3 when the actual pace stayed below 89, 1-2 when it was between 89 and 91.5, and 4-6 when it was above 91.5.
They were 21-9 against teams with a pace of 89-91.5 and the average actual pace was around 90. They were 8-2 when the actual pace stayed below 89, 3-2 when it was between 89 and 91.5, and 10-5 when it was above 91.5.
They were 24-8 against teams with an average pace above 91.5 and the average actual pace was around 98. They were 7-1 when the actual pace stayed below 89, 2-0 when it was between 89 and 91.5, and 15-7 when it was above 91.5.
So against the slowest pace teams Phoenix playing above league average pace did worst of the three groups and about the same against average and high pace teams. Figures. They did very well in low paced games (surprising), better % wise than in high pace games overall. Average pace games was their weakest performance but rare.
I think Phil Jackson was using a similar line of thinking. I've been tracking this for the first round Laker/Suns series, and (using the Hollinger pace equation) the two games were played at 10 and 8 possessions slower than the Suns regular season pace. It clearly has had some success.
Which brings up the question of game planning, as Mark first mentioned. During the regular season there is little time to really alter your basic game plan from team to team, night to night. However, in the playoffs, with the more intent focus and a series over several weeks (how long does the first round go on, a month?) you can make more and more dramatic changes. So, maybe predicting pace will depend somewhat on circumstances.
Back to top
View user's profile Send private message Visit poster's website AIM Address
Mark
Joined: 20 Aug 2005
Posts: 807
PostPosted: Thu Apr 27, 2006 7:31 pm Post subject: Reply with quote
Pace studies using playoff series could be very informative and I've thought of looking at strong contrast series (like Kings/Spurs so far played at Kings preferred pace but won by Spurs. Kings worst win % is against low paced teams- 47%. Spurs worse against high pace teams but still win 71%). Quite different from regular season but perhaps a little easier to figure what is going on with up to 7 games.
Last year the Suns controlled the pace of the whole series with Memphis and won 4-0. This year, Memphis is down 2-0 and lost game 2 by more than game 1 but probably pulled the pace down. May need to succeed in lowering the pace even further to win. Bad losses in turnover and rebounds need to change too.
The Bulls made a fight of it last year in the playoffs but played nearly all the series with Washington at a high pace and that did not seem to their advantage to me- even though their w-l profile against different paces was even high to low and their pace was a little above average, the Wizards were moderately higher paced and seemed to have more offensive weapons. The Bulls are playing fast pace against Miami so far this year and it is not working again. Heat's much higher team efficiency probably carries the series, though the Bulls were quite efficient in game 2 also.
I assume that part of the answer why Phoenix seems to do well in slower paced games in regular season this year is the relative efficiency of their offense and the Suns style being somewhat unusual to teams. Fast pace games on average maximize that advantage but if they still have that efficiency edge more often than not in slower pace games they can do well. So in addition to looking at pace and game outcome it probably would make sense to look at TS% across opponent average pace groups and actual paces played. Will watch the playoff box scores.
Pace by quarter, pace with leads of different length or behind, pace with stars on court/off, etc. I previously wondered about pace impact on individual players and Dirk N. or Shawn Marion or D Wade or Gasol might be interesting cases.