|
|
Post by The Bofa on the Sofa on Sept 23, 2014 16:33:12 GMT -5
Purdue has been playing in Mackey Arena occasionally, lately, and I have been wondering, do they still have a home court advantage there? Or is it more like a neutral court? There are probably lots of ways to think about it, but the approach I choose was to actually look at some nitty-gritty. In some of my analysis, I have discovered an interesting phenomenon in terms of home/road splits. One of the biggest effects that is found in terms of home/road splits occurs in serving. I discovered this a few years ago, and it's an amazing thing. The stats bear this out. Here's what you find:
There is, in fact, very little difference in the number of serving errors that a team makes at home vs on the road overall. It might go up a tad, but it's pretty constant. The difference you find in serving is that aces go way down on the road compared to home, and consequently, the error/ace ratio goes way up. This is easily explained in terms of serving aggressiveness. At home, teams serve more effectively, and therefore more aggressively. On the road, they are less effective at serving, so serve less aggressively. Now, I found this by looking at home/road splits for matched pairs in conference play (a few conferences, a few years), so I am not able to say anything about neutral matches, and whether they are like road matches or something different.
Which brings us to Purdue. To see if Purdue is playing Mackey like a home court, what I have done is to look at their serving performance in their Mackey matches in the last 3 years. They've played 7 matches there, including this year, so we are getting some numbers.
What I did was to calculate the expected number of aces and serving errors in their Mackey matches, given the number of sets played and their ace and error rates for home, road and neutral for the last three years, and compared them to the overall results. Here's what I get
Overall results (Aces, errors, error/ace)
28, 34, 1.21
Now, if they had performed at the rate they did the rest of the season (normalized by each season; based on 2014 so far, which is only 2 Away matches), they would expect the following
Home: 31, 35, 1.13
Away: 23, 43, 1.92
Neutral: 32, 54, 1.67
In comparison, the overall serving results are pretty much in line with what they do in Holloway Gymnasium. That suggests that, indeed, they are serving like Mackey is a home court.
A couple of comments:
1) You note that the number of expected serving errors is not the same home and road. That is because the Mackey data is weighted heavily toward 2012, where Purdue played 5 matches in Mackey, but their road/home error split was very high that season. This is likely a data limitation. If you look at Purdue's serving over three years (2012-2014), they they averaging 1.66 errors/set at home and 1.67 errors/set on the road, so 2012 is anomalous. However, mikegarrison will point out that, if that is the case, then the expected aces and errors for that year will also be anomalous. Therefore, I have also calculated the expected aces/errors based on the three year average. In that case, the expected results are
Home: 35, 41, 1.17
Road: 21, 42, 1.96
Neutral: 34, 52, 1.52
In this case, there is kind of an in-betweeny. In the Mackey matches, they don't have near enough aces, but they also don't have nearly as many serving errors. Overall, the ratio is still right in line with it being a home court. The way to interpret it is that they aren't quite as aggressive in their serving, and they serve more conservatively, but since it is at home, the error rate also comes down. Teams on the road tend to serve more conservatively, but that's because they make more errors on the road. That's a difference.
2) Purdue's neutral stats are very interesting. On neutral courts, the ace rate is almost as high as their home serving (1.41 aces/set at home vs 1.36 aces/set neutral vs 0.9 aces/set on the road). However, in neutral matches, their error rate goes way up (1.66 at home, 1.67 on the road, 2.06 neutral). How do we interpret that? It's like they are serving aggressively like they do at home, but they are making more mistakes. Consequently, the error/ace ratio comes out to 1.52 - interestingly, it is right in-between their home and road values. That tells me it makes sense, their neutral court results, but we can learn a lot from the rate values.
3) It's something I've hinted at before, but the difference in aces home and road is amazing to me. 1.4 aces/set at home vs 0.9 on the road. I mean, that alone is the difference of 2 points in a match, which is by far the biggest chunk of the home court advantage. THat's about a third of the total home court advantage right there. My position is that coaches don't spend NEAR enough time in serving practice in road gyms. You can really help your team by getting players more comfortable serving in the strange environment.
Conclusion:
Purdue's serving suggests that Mackey arena is really a home court for Purdue. They might be a little cautious serving there, but at least from a familiarity standpoint, they are able to serve there as well as they do in the other gym. When they face teams in neutral venues, they seem to be very aggressive, but make more errors than they would at home. On the road, they do like most teams do, and back off, and it costs them points.
|
|
|
|
Post by bkedane on Sept 23, 2014 17:04:25 GMT -5
Purdue has been playing in Mackey Arena occasionally, lately, and I have been wondering, do they still have a home court advantage there? Or is it more like a neutral court? There are probably lots of ways to think about it, but the approach I choose was to actually look at some nitty-gritty. In some of my analysis, I have discovered an interesting phenomenon in terms of home/road splits. One of the biggest effects that is found in terms of home/road splits occurs in serving. I discovered this a few years ago, and it's an amazing thing. The stats bear this out. Here's what you find: There is, in fact, very little difference in the number of serving errors that a team makes at home vs on the road overall. It might go up a tad, but it's pretty constant. The difference you find in serving is that aces go way down on the road compared to home, and consequently, the error/ace ratio goes way up. This is easily explained in terms of serving aggressiveness. At home, teams serve more effectively, and therefore more aggressively. On the road, they are less effective at serving, so serve less aggressively. Now, I found this by looking at home/road splits for matched pairs in conference play (a few conferences, a few years), so I am not able to say anything about neutral matches, and whether they are like road matches or something different. Which brings us to Purdue. To see if Purdue is playing Mackey like a home court, what I have done is to look at their serving performance in their Mackey matches in the last 3 years. They've played 7 matches there, including this year, so we are getting some numbers. What I did was to calculate the expected number of aces and serving errors in their Mackey matches, given the number of sets played and their ace and error rates for home, road and neutral for the last three years, and compared them to the overall results. Here's what I get Overall results (Aces, errors, error/ace) 28, 34, 1.21 Now, if they had performed at the rate they did the rest of the season (normalized by each season; based on 2014 so far, which is only 2 Away matches), they would expect the following Home: 31, 35, 1.13 Away: 23, 43, 1.92 Neutral: 32, 54, 1.67 In comparison, the overall serving results are pretty much in line with what they do in Holloway Gymnasium. That suggests that, indeed, they are serving like Mackey is a home court. A couple of comments: 1) You note that the number of expected serving errors is not the same home and road. That is because the Mackey data is weighted heavily toward 2012, where Purdue played 5 matches in Mackey, but their road/home error split was very high that season. This is likely a data limitation. If you look at Purdue's serving over three years (2012-2014), they they averaging 1.66 errors/set at home and 1.67 errors/set on the road, so 2012 is anomalous. However, mikegarrison will point out that, if that is the case, then the expected aces and errors for that year will also be anomalous. Therefore, I have also calculated the expected aces/errors based on the three year average. In that case, the expected results are Home: 35, 41, 1.17 Road: 21, 42, 1.96 Neutral: 34, 52, 1.52 In this case, there is kind of an in-betweeny. In the Mackey matches, they don't have near enough aces, but they also don't have nearly as many serving errors. Overall, the ratio is still right in line with it being a home court. The way to interpret it is that they aren't quite as aggressive in their serving, and they serve more conservatively, but since it is at home, the error rate also comes down. Teams on the road tend to serve more conservatively, but that's because they make more errors on the road. That's a difference. 2) Purdue's neutral stats are very interesting. On neutral courts, the ace rate is almost as high as their home serving (1.41 aces/set at home vs 1.36 aces/set neutral vs 0.9 aces/set on the road). However, in neutral matches, their error rate goes way up (1.66 at home, 1.67 on the road, 2.06 neutral). How do we interpret that? It's like they are serving aggressively like they do at home, but they are making more mistakes. Consequently, the error/ace ratio comes out to 1.52 - interestingly, it is right in-between their home and road values. That tells me it makes sense, their neutral court results, but we can learn a lot from the rate values. 3) It's something I've hinted at before, but the difference in aces home and road is amazing to me. 1.4 aces/set at home vs 0.9 on the road. I mean, that alone is the difference of 2 points in a match, which is by far the biggest chunk of the home court advantage. THat's about a third of the total home court advantage right there. My position is that coaches don't spend NEAR enough time in serving practice in road gyms. You can really help your team by getting players more comfortable serving in the strange environment. Conclusion: Purdue's serving suggests that Mackey arena is really a home court for Purdue. They might be a little cautious serving there, but at least from a familiarity standpoint, they are able to serve there as well as they do in the other gym. When they face teams in neutral venues, they seem to be very aggressive, but make more errors than they would at home. On the road, they do like most teams do, and back off, and it costs them points. An interesting issue and initial analysis. A few thoughts prompted by a quick read of the post: Are the teams Purdue plays in Mackey a representative sample of the opponents played in the normal gym? Or do they choose to schedule tougher (or weaker) opponents in Mackey? Similarly, are the neutral site opponents coming dominantly from a weaker pool (pre-conference opponents, for example) or are neutral court opponents most commonly tougher than average opponents (neutral site ncaa tourney opponents perhaps?). |
|
|
|
Post by The Bofa on the Sofa on Sept 23, 2014 17:21:31 GMT -5
Good questions
1) the Mackey opponents have ranged from Penn St to Jackson St, with teams in between. Overall, their record is 5-2, so it's fairly representative. Maybe a touch weaker overall, but not drastically.
2) The neutral matches are probably a little weaker overall. That could account for more aces, but why so many errors? That part surprises me
|
|
|
|
Post by bkedane on Sept 23, 2014 17:29:32 GMT -5
Good questions 1) the Mackey opponents have ranged from Penn St to Jackson St, with teams in between. Overall, their record is 5-2, so it's fairly representative. Maybe a touch weaker overall, but not drastically. 2) The neutral matches are probably a little weaker overall. That could account for more aces, but why so many errors? That part surprises me I agree that some of the numbers are surprising. So far i'm just thinking about loud about what else could be relevant. And i'm wondering if it could be just noise from the small sample size. I'll keep thinking and look forward to any updates and thoughts from others. |
|
|
|
Post by The Bofa on the Sofa on Sept 23, 2014 17:45:38 GMT -5
Could be small numbers, but it's something like 15 matches or so. Note that it includes a mix of preseason tournaments and the NCAAs
Thinking about that, it might actually be a little better than I thought.
|
|
|
|
Post by The Bofa on the Sofa on Sept 23, 2014 17:59:28 GMT -5
Just to repeat, the part about errors being the same home vs road is not Purdue specific or small sample. It's very large sample, in fact, and very well controlled.
The home/road split for aces is one of the biggest home court effects. Interestingly, before the ball handling rule change, the biggest home/road difference in the B1G was BHE. It was only in the B1G, not PAC 10 or MVC.
Personally, I attributed it to subconcious home team bias by the officials. Then again, I had just read Scorecasting, where they demonstrated a lot of official bias.
With the relaxed ball handling rules, the effect has disappeared
|
|
|
|
Post by n00b on Sept 23, 2014 19:44:17 GMT -5
Regarding the neutral-site missed serves... My hypothesis is that it's due to the fact that the VAST majority of those matches come in weeks 1-4. Serving improves as the season progresses, so week 1 would be the most errors followed by week 2, etc. Otherwise, it makes no sense to me to miss more on neutral courts than true road matches.
Also, I think the reduced aces on the road has less to do with serving less aggressively and more to do with passers being accustomed to the backdrop and sight-lines when in serve receive. Obviously this isn't really something testable, but that was my first instinct when reading through your findings.
|
|
|
|
Post by mikegarrison on Sept 23, 2014 20:01:56 GMT -5
Regarding the neutral-site missed serves... My hypothesis is that it's due to the fact that the VAST majority of those matches come in weeks 1-4. Serving improves as the season progresses, so week 1 would be the most errors followed by week 2, etc. Otherwise, it makes no sense to me to miss more on neutral courts than true road matches. Also, I think the reduced aces on the road has less to do with serving less aggressively and more to do with passers being accustomed to the backdrop and sight-lines when in serve receive. Obviously this isn't really something testable, but that was my first instinct when reading through your findings. These are interesting ideas. They show the limitations of trying to draw out causation from correlation. It would be possible to test these hypotheses if we had control over match scheduling, but of course we don't. I would put much more faith in the conference season records (because they usually control for opponents and home/away) than I would in the pre-conference or tournament results, which means the most reliable data doesn't include neutral sites. |
|
|
|
Post by The Bofa on the Sofa on Sept 23, 2014 20:22:36 GMT -5
In terms of neutral court, note this is one team with 3 seasons of neutral court results (note noob that for Purdue, preseason has been about 80% of their neutral court matches; however note also that the end of the season neutral matches are harder opponents-I think it's a wash at best). We'd have to see other teams
As I said, the home/road splits are consistent with what you find in conference only home/home pairs
|
|
|
|
Post by n00b on Sept 23, 2014 20:26:49 GMT -5
In terms of neutral court, note this is one team with 3 seasons of neutral court results (note noob that for Purdue, preseason has been about 80% of their neutral court matches; however note also that the end of the season neutral matches are harder opponents-I think it's a wash at best). We'd have to see other teams As I said, the home/road splits are consistent with what you find in conference only home/home pairs Ah, I think I misunderstood your original post then. What are the NCAA-wide Home/Away/Neutral serving splits? |
|
|
|
Post by mikegarrison on Sept 23, 2014 20:41:30 GMT -5
What's the difference between "away" and "neutral"? The other team!
So let's think about causality hypotheses. We could dream some up and then look at the data to see if they are falsified. For example, the idea that sightlines and the gym environment are key to serving:
1) Errors are dependent mainly on the serving team. So if sightlines and familiar environments are the causal factor, then we should expect to see errors/serve be similar neutral and away, but lower at home.
2) Aces depend (at least in part) on the other team. So if sightlines and familiar environments are important, we should see the ace rate on serves in play be similar for home and neutral matches, but be lower away (where the receivers are more comfortable).
We would ideally also block for other variables we want to exclude, like altitude effects, serving team quality, receiving team quality, time into the season, etc.
|
|
|
|
Post by The Bofa on the Sofa on Sept 23, 2014 21:01:50 GMT -5
In terms of neutral court, note this is one team with 3 seasons of neutral court results (note noob that for Purdue, preseason has been about 80% of their neutral court matches; however note also that the end of the season neutral matches are harder opponents-I think it's a wash at best). We'd have to see other teams As I said, the home/road splits are consistent with what you find in conference only home/home pairs Ah, I think I misunderstood your original post then. What are the NCAA-wide Home/Away/Neutral serving splits? Don't know. Let me explain again: I have in the past looked at home/road splits for serving by comparing the home and away matches played in the conference season. Multiple conferences, multiple years, hundreds of match pairs. In doing that, I found that there is little difference between home and road in terms of errors. Aces are much lower rates on the road than at home. In examining Purdue matches, that is exactly that I find in their home/road splits. The errors (over the long time period) are pretty similar, but there is a huge difference in aces. Thus, they seem normal in that respect. I have nothing to compare for neutral sites. I just find it interesting that the Error/Ace ratio for neutral sites is right in-between what they do for home and road. |
|
|
|
Post by The Bofa on the Sofa on Sept 23, 2014 21:06:48 GMT -5
What's the difference between "away" and "neutral"? The other team! So let's think about causality hypotheses. We could dream some up and then look at the data to see if they are falsified. For example, the idea that sightlines and the gym environment are key to serving: 1) Errors are dependent mainly on the serving team. So if sightlines and familiar environments are the causal factor, then we should expect to see errors/serve be similar neutral and away, but lower at home. 2) Aces depend (at least in part) on the other team. So if sightlines and familiar environments are important, we should see the ace rate on serves in play be similar for home and neutral matches, but be lower away (where the receivers are more comfortable). And this is why I don't know about he sight-lines explanation. #2 holds up. #1, however, doesn't. That's why I went with an aggressiveness component. Actually, although I mentioned about officials bias in terms of the BHE splits, it's worth noting that it is also explained by an aggressiveness explanation. If the home team is serving more aggressively, it leads to more aces but it also causes the opponents to have more BHEs. Hence, you see more BHEs for road teams. Now that there are so few BHEs, that doesn't show up. |
|
|
|
Post by mikegarrison on Sept 23, 2014 21:23:31 GMT -5
What's the difference between "away" and "neutral"? The other team! So let's think about causality hypotheses. We could dream some up and then look at the data to see if they are falsified. For example, the idea that sightlines and the gym environment are key to serving: 1) Errors are dependent mainly on the serving team. So if sightlines and familiar environments are the causal factor, then we should expect to see errors/serve be similar neutral and away, but lower at home. 2) Aces depend (at least in part) on the other team. So if sightlines and familiar environments are important, we should see the ace rate on serves in play be similar for home and neutral matches, but be lower away (where the receivers are more comfortable). And this is why I don't know about he sight-lines explanation. #2 holds up. #1, however, doesn't. That's why I went with an aggressiveness component. Actually, although I mentioned about officials bias in terms of the BHE splits, it's worth noting that it is also explained by an aggressiveness explanation. If the home team is serving more aggressively, it leads to more aces but it also causes the opponents to have more BHEs. Hence, you see more BHEs for road teams. Now that there are so few BHEs, that doesn't show up. It's true, if you imagine "aggressiveness" as a tunable dial, you could argue that maybe teams adjust that dial to keep a constant tolerable level of errors. If so, then perhaps the constant level of errors is not incompatible with the sightlines explanation. It would be interesting to break this down by player, rather than aggregated into team statistics. |
|
|
|
Post by c4ndlelight on Sept 23, 2014 21:29:21 GMT -5
What's the difference between "away" and "neutral"? The other team! So let's think about causality hypotheses. We could dream some up and then look at the data to see if they are falsified. For example, the idea that sightlines and the gym environment are key to serving: 1) Errors are dependent mainly on the serving team. So if sightlines and familiar environments are the causal factor, then we should expect to see errors/serve be similar neutral and away, but lower at home. 2) Aces depend (at least in part) on the other team. So if sightlines and familiar environments are important, we should see the ace rate on serves in play be similar for home and neutral matches, but be lower away (where the receivers are more comfortable). And this is why I don't know about he sight-lines explanation. #2 holds up. #1, however, doesn't. That's why I went with an aggressiveness component. Actually, although I mentioned about officials bias in terms of the BHE splits, it's worth noting that it is also explained by an aggressiveness explanation. If the home team is serving more aggressively, it leads to more aces but it also causes the opponents to have more BHEs. Hence, you see more BHEs for road teams. Now that there are so few BHEs, that doesn't show up. Increased aces for the home team is also increased reception errors for the road team- are you sure it's not the reception that's the main variable? (Especially if the error column on serving isn't effected too much) In the PAC, for example, each team has traditionally had its own ball contract, so it would be a very different experience passing in each gym. And sight lines/tracking would have as much, if not more, to do with passing than serving. Edit: Just read the whole thread and seems my point was made above. |
|
