Par Talk

[Steve West]

Okay, so what definition were you using when you came up with that? Because, it is apparently not a definition that includes two close range throws or fairly analyzes the current definition.

The connection is there. It may not be as obvious as using the mode or average, but I think it better represents the definition than either of those - or anything else I've heard of.

I started with "As determined by the Director, the score an expert disc golfer would be expected to make on a given hole with errorless play under ordinary weather conditions, allowing two throws from close range to hole out."

First, I ignored the "As determined by the Director" part, because I wanted to find a way to look at hole scores and find out what score would be expected.

I used a composite 1000-rated player (NOT all players rated 1000 and above) as the expert, because that way I could have a standard reference across all courses even if the mix of players was different.

As has been pointed out earlier, I did not make any special provision for "allowing two throws from close range to hole out". This phrase does not imply adding two to anything. If close range is where you expect to throw two more times, then this phrase just "allowed" the last two throws - they overlap two throws that are already there. As has been pointed out, the number of throws that actually are made within close range is not always exactly two. So, to exactly match the definition, I would have needed to use the actual score, minus the throws that were made from close range, plus two. However, there is no way to figure out how many throws were actually made from within close range. Two seems like a good estimate. My method needs to be practical, and using two requires no adjustments to scoring data.

Because I don't have weather data, I also relied on the expectation that most rounds are held under ordinary weather conditions, and if not, over several rounds of data, the good scores would be those that happened in ordinary weather, and it's only the good scores that figure into par. In some cases, like last year's USDGC, where I knew the weather was not ordinary I didn't use that data.

I am quite comfortable with using "top X% of all throws" as a way to get a number to attach to errorless play. Errorless play is certainly some subset of all throws. It's not much of a stretch to say the errorless throws resulted in the low scores. Given a percentile that represent errorless play (including better than errorless play), the score an expert disc golfer would be expected to make on a given hole just pops out.

What I like most about it is that it generates pars that take an equally high level of play to achieve across all pars. As equal as possible.

 

[Steve West]

I feel very confident that more than 56.25% of players rated over 1000, would 2 a 365 foot wide open hole with no OB.

I don't use over 1000. I use exactly 1000.

 

[_MTL_]

I don't use over 1000. I use exactly 1000.

Interesting. Why?

 

[lyleoross]

Interesting. Why?

I don't know Steve's reasons, but in research, when I wanted to get defined data, I stuck strictly to the defined pool. Makes analysis easier, and arguing the rational easier.

 

[lyleoross]

Steve, are you going to do DeLa? I'm curious to see the data on hole 17. I'd really love to have you compare it to last year's numbers. They changed the hole from a fast sloping green with easy roll-aways to a terraced green where you could land on three terraces, thus avoiding roll aways. I'd like to know whether the hole was a good distributor in past years, or was random (I suspect random) and how much the terraces lowered scores (I'm hypothesizing they did) and if they made better score distribution.

 

[Steve West]

Interesting. Why?

I use a single rating to minimize any inconsistencies from different mixes of players at different events. I want par to be comparable from course to course and event to event. Looking at all players above 1000 could give you anything from a few players rated barely above 1000, to Nate & Paul showing up to support a charity event with no other 1000 rated players.

I picked 1000 because it's a round number, easy to remember, was originally set to be the last cash line, is probably a popular answer to the question "what rating do you need to get to be an expert", I've heard it is a cutoff for some bonuses for sponsored players, and the pars it generates are generally in the money at large events.

So, if par is set by looking at the 1000-rated player, and you stay at even par you'll likely stay in the cash.

Any higher rating would have more trouble getting enough data. For example, here is the amount of data I have for the Masters Cup by rating. Very few events have this many high-level players, yet the amount of data still falls off quickly for ratings higher than 1000.

 

[Steve West]

Steve, are you going to do DeLa? I'm curious to see the data on hole 17. I'd really love to have you compare it to last year's numbers. They changed the hole from a fast sloping green with easy roll-aways to a terraced green where you could land on three terraces, thus avoiding roll aways. I'd like to know whether the hole was a good distributor in past years, or was random (I suspect random) and how much the terraces lowered scores (I'm hypothesizing they did) and if they made better score distribution.

I might, in the Masters Cup thread. Is the numbering the same both years?

 

[lyleoross]

I might, in the Masters Cup thread. Is the numbering the same both years?

I believe so. Thanks!

 

[Doofenshmirtz]

As has been pointed out earlier, I did not make any special provision for "allowing two throws from close range to hole out". This phrase does not imply adding two to anything.

Well, this depends on how you determine close range. On an easy par 4, you would expect a lot of threes, no? Should you not exclude the throws that took 2 throws at close range?

I guess that what I am saying is that if you discard the part of the definition that allows for two throws at close range, then you cannot say that any of your data shows that par was set incorrectly.

If close range is where you expect to throw two more times . . .

Nope. It is not so defined. If you used this definition, you are now testing a different measure. I agree that it should be defined as the distance from which an expert is expected to hole out in two throws. That that distance needs to be determined first, and then the hole analyzed. This is a bad definition that you address below.

then this phrase just "allowed" the last two throws - they overlap two throws that are already there.

Only if it took two shots from close range. If only one shot was taken from close range, then including that score would be using invalid data.

As has been pointed out, the number of throws that actually are made within close range is not always exactly two. So, to exactly match the definition, I would have needed to use the actual score, minus the throws that were made from close range, plus two.

See, you're getting there . . .

However, there is no way to figure out how many throws were actually made from within close range. Two seems like a good estimate. My method needs to be practical, and using two requires no adjustments to scoring data.

:wall: You are measuring current par against your new definition.

It seems to me that it is not unreasonable to surmise that the holes with the lowest scores relative to par are also the holes where a single close range throw was most likely. Aside from this inherent bias, I would argue that you are not trying to match the minimum standard idea (which I argue is implied by the allowance of two close range tries) but you are instead criticizing the application of the existing rule by using the data of a proposed rule.

Because I don't have weather data, . . .

No problem real with this, but perhaps the existing round rating data could be factored in?

Given a percentile that represent errorless play (including better than errorless play), the score an expert disc golfer would be expected to make on a given hole just pops out.

Which is, apparently, the definition you want instead of the definition that we have. You also don't want it to be a minimum standard, but instead a standard designed to match the scoring data. Since you know the target you want to hit with your data, why not use a round rating type analysis of each tournament. Wouldn't that even be more accurate? Let par be decided after the tournament. Indeed, screw the holes, let there just be course par. Wouldn't that be more accurate still, taking out the independent hole data and just use a number that matched course difficulty that day?

What I like most about it is that it generates pars that take an equally high level of play to achieve across all pars. As equal as possible.

I guess that I will just never understand the level of embarrassment felt about how much under par the pros reach in some tournaments. The level of play will not change. You will just assign different numbers to it. They will still be integers and they will still be just as inaccurate as they are now. There will just be a different definition. I'm guessing it will be one that matches your current data, the same data you are claiming shows the inaccuracy of the current definition, even though you didn't use the current definition to calculate it.

 

[DavidSauls]

I guess that I will just never understand the level of embarrassment felt about how much under par the pros reach in some tournaments. .

Don't bother trying---I haven't noticed anyone here making that complaint.

The argument for making par closer to what is expected of an expert isn't about how it looks for them to shoot -50 for a weekend. It's about par being meaningful, and useful.

 

[Doofenshmirtz]

Don't bother trying---I haven't noticed anyone here making that complaint.

The argument for making par closer to what is expected of an expert isn't about how it looks for them to shoot -50 for a weekend. It's about par being meaningful, and useful.

Meaningful = fewer pros shooting too far down in too many tournaments? The end result of changing the definition equals a different stamp on the same scores. It's obvious that there has been no analysis as to whether any pars are set inaccurately according to the existing definition. Instead, "inaccuracy" is being touted as an excuse for changing the definition with the primary argument (not the one people say they are making, but, instead the one that the methodology points to), not that pars are inaccurately set, but that pars are just set too high. Because the proffered excuse doesn't really jibe with the analysi offered to support it, I see the obvious end result as the sole reason for the change. I.e., because your claimed reason doesn't really make any sense, I think you may have a different reason.

Useful? How so. If a par 3 changes to a par 2 or a 4 to a 3, how is that any more useful? Will you play it differently? Will it be an easy par 2 or a difficult par 2? As long as fractional pars are not used, hole difficulty hasn't been described. The change will have just made a hole either difficult or easy to par, instead of difficult or easy to birdie. That doesn't seem useful to me unless we are just trying to get rid of birdies.

Why not just change the definition of par to "the average scores of the 1000 rated players in the tournament?" That would take not only course difficulty into consideration, but also weather and all other conditions that may impact play. You would also have the benefit of not having to use the real golf terminology at all - you could just call it "Expert Average Result" or "EAR." Wouldn't that even be more useful? If usefulness is about accuracy, then why not a more accurate measure than "Par."

Additionally, I wonder what unintended (or perhaps secretly intended?) results might result from changing thousands of holes to par 2 and par 3 (instead of 3 and 4). Will a course be viewed as an expert level course if it has too many Par 2s? Will designers avoid shorter holes if they are worried that, after the course has been finished, that the PDGA will stamp a Par 2 on it if the course plays easier than expected, especially if they only have a more open area to work with? Will some designers pre-emptively assign par 2 to open holes that no-one but a 1000 rated player can par? Can guidelines be written that say something other than giving a methodology based on scoring history in tournaments? How much data collection are you going to require of a designer before he can assign par and be in-line with the definition? If the guidelines only change for tournaments, then what is the supposed benefit to anyone else? Why would any course designer use the new definition when the old definition is expected by more than 99 percent of the people who will play the course? Heck, most of those people have never read the rules, much less the par definition.

 

[Steve West]

...even though you didn't use the current definition to calculate it.

If you want to use the entire definition, we must say that every par that was ever set by every TD ever was exactly right. Which is true.

That's why I've never (I think never, at least I've tried not to, and if I did it was a lapse.) said "par was set wrong".

As I've already acknowledged, I used just "the score an expert disc golfer would be expected to make on a given hole with errorless play under ordinary weather conditions" part of the definition. My method is consistent with what I said it was.

Leaving it out produces pars that are more useful than leaving it in, so I wouldn't have included it even if I had thought about the difference it could make. I wanted to see how far pars are from the most useful possible.

What exactly would you like to happen now that you've pointed out a slight technical difference between my method and the non-TD part of the definition?

If all you want is an acknowledgement that what I thought didn't make a difference, actually could, you've had it a few posts up. Thank you for the analysis, you made a good point. What now?

 

[DG_player]

I think DG-Player's issue here---and it's a real one---is holes where results are more or less evenly split between two scores.

For simplicity, think of a hole where 50% of experts get a 3, and 50% get a 4. What's the expected score?

It may or may not be a bad hole. It's a tweener par, and not great at scoring separation, but it's still separating players based on performance.

My own thought is that these holes are fairly rare, and we can live with the par being slightly off, either way, when the holes on the course are set accurately. Steve suggested we could decide between 3 or 4, to bring the total course par in line.

No matter how par is determined---even with "number of throws to the basket, plus two"---there are going to be holes that are borderline between two numbers.

It's not that I'm concerned that there will be borderline holes, it's more that it feels like an admission of failure by disc golf. Steve can call this hole a par 3, even though he knows it's not truly a par 3. I can call it a par 4, even though I know it's not truly a par 4. I realize often times the resources aren't available to fix every hole and make them all play true to par, even at the pro level. But it just feels defeatist to me to give up, slap the closest number on a hole, and act as if it's OK.

 

[Disczealot]

It's not that I'm concerned that there will be borderline holes, it's more that it feels like an admission of failure by disc golf. Steve can call this hole a par 3, even though he knows it's not truly a par 3. I can call it a par 4, even though I know it's not truly a par 4. I realize often times the resources aren't available to fix every hole and make them all play true to par, even at the pro level. But it just feels defeatist to me to give up, slap the closest number on a hole, and act as if it's OK.

You make a good point here.

What is the ideal scoring spread for a par 4? 6% 2s, 24% 3s, 40% 4s, 24% 5s, 6% 6s?

My concern, if this was possible to achieve, is that it would limit design freedom.

For every par 4 be a "true" par 4, they would need to all have the exact same difficulty.

Some holes that aren't the "truest" par 4s could still have great scoring spread, and be some of the most exciting holes out there.

The closer to "true" you stay, the less freedom you have over design options. Where do you draw the line?

 

[DavidSauls]

To me, "design to suit par" is the tail wagging the dog.

Design holes for good play, and good scoring spread---including a variety of scoring spreads. Even 50/50 holes.

Design par for the best approximation of what players should get (high level players on high-level courses), so it's most useful.

Don't sweat if par isn't a precise and exact measurement of anything.

 

[Steve West]

To me, "design to suit par" is the tail wagging the dog.

Design holes for good play, and good scoring spread---including a variety of scoring spreads. Even 50/50 holes.

Design par for the best approximation of what players should get (high level players on high-level courses), so it's most useful.

Don't sweat if par isn't a precise and exact measurement of anything.

While I agree, remember that it is baked into traditional designers DNA that the first thing they decide about a hole is that it will be a par X. They'll say things like: "I need another par 4", before deciding anything else about that hole.

I start with how cool the throws are, what emotions the hole will induce in the player, and how pretty and pleasant the scene is. Then if it's for organized competition, adjust it to produce scoring spread. Then figure out what par to assign to it.

 

[NegaSnapples]

I just want to put my .02 cents in on a couple of things in this thread.

In disc golf, the results aren't all that much crazier than some PGA tour events. Look at the Sony Open results. They went pretty far under par for the event. https://www.google.com/#q=pga+tour+sony+open+results&spf=1495547737716

I think the biggest problem with disc golf, at the top level, is just the courses they are playing on. With golf, I used to shoot scratch at my local 9 hole public course. I wouldn't be able to shoot anywhere near par at Augusta playing from the tips.

The top level tournaments just need to have more challenging layouts and distances. Don't have a 225 ft par three that everyone on the card birdies. I think changing par doesn't address the larger issue of poor hole design, and that is where the real issue is IMO.

 

[Steve West]

...What is the ideal scoring spread for a par 4?...

We can compute the maximum possible scoring spread width under the constraints of being a par 4 under my method, and the mix of scores that are involved.

2	0%	0%	21%	0%	0%	21%	0%	20%
3	42%	0%	21%	33%	0%	21%	25%	20%
4	58%	50%	58%	33%	33%	29%	25%	20%
5	0%	50%	0%	33%	33%	29%	25%	20%
6	0%	0%	0%	0%	33%	0%	25%	20%
Ave	3.58	4.50	3.37	4.00	5.00	3.66	4.50	4.00
SSW	1.98	2.00	2.65	3.00	3.00	3.95	4.00	5.00

 

[DavidSauls]

Meaningful = fewer pros shooting too far down in too many tournaments? .

That's a red herring. I said it before. I don't think anybody here's making the argument that pros' shooting far under par is a problem in itself. I certainly know that I haven't.

By meaningful, I mean that "par" should have a meaning. Not just a formula. Not just a methodology Not just the result of a methodology.

What does it stand for? What does it represent? Not, How is it calculated?

 

[DavidSauls]

While I agree, remember that it is baked into traditional designers DNA that the first thing they decide about a hole is that it will be a par X. They'll say things like: "I need another par 4", before deciding anything else about that hole.

I start with how cool the throws are, what emotions the hole will induce in the player, and how pretty and pleasant the scene is. Then if it's for organized competition, adjust it to produce scoring spread. Then figure out what par to assign to it.

Do they? Sorry for my ignorance, I'm only responsible for one course, and it never dawned upon us to look for a certain par. I can't even imagine designing by quota.