Par Talk

[Steve West]

What's often missed is that a player can exceed expectations and (I think) you need to consider that in your calculations. Hence a scramble for par, hitting a C2 putt, or a par in spite of an OB is exceeding expectations.

That could very well be the next refinement.

The existence of scores that exceed expectations is one reason I approach par from the low scores up. If there are enough 2s, it's a par 2. If not, then if there are enough 3s plus 2s, it's a par 3, etc. So, all the throws that exceeded expectations enough to result in a lower score are automatically folded into the calculation of par.

That still leaves the possibility that a few players got par as a result of an error plus an exceptional throw. Since this only results from two rare events during the same hole, and since most holes fall squarely within an obvious par value, it will only very rarely push a hole over into a lower par. That would only happen when a hole was on the bubble between two pars. Since the method is a tool for the TD to use, not a rule that will dictate all pars from now on, they should be looking closely at those holes anyway.

Also, I calibrated the cutoff values, so the cutoff is already at the level where those rare cases push the par over the tougher threshold just often enough. In other words, if I took them out (after examining the stats), then the threshold might need to be lower.

For example, the threshold now is that at least 47% of 1000-rated players need to get 3 or better for the method to recommend par of 3. If we find out that some of those threes were the result of recoveries after an error, perhaps the threshold would need to be that only 42% of players need to get errorless 3s or better for a hole to be par 3.

But, if we did that, then everyone would need Udisc-like stats to set par, and that seems impractical. For now.

I think your biggest problem with players exceeding expectations is that you've chosen the 1000(ish) rated player as an "expert" and there are quite a few players with ratings indicating a 4+ throw per round advantage over the 1000 rated player. And those are the players we usually see on video (lead & chase cards), they have developed fan bases, and many of the arguments I've read on this forum that par is too easy is based on the play of these elite players.

There are people arguing par is too easy? Not pars set by my method. Almost always the holes where I hear anyone say a par is too easy are the holes with pars that are higher than my method would generate.

The argument usually goes that my method would reduce the number of birdies and that's bad because Paul and Ricky (and now many others) should be getting a lot of birdies.

The truth is that using the 1000-rated player as the expert assures that there will still be plenty of birdies by those guys (and about as many bogeys as birdies by the 1000-rated players.) . The top guys on video will consistently get about 4 or 5 birdies per round, instead of a random number anywhere from 4 to 17 per round.

 

[beereakfast]

Par for every tournament in a year. Add up the money. That guy loses his card.

It matters in context of the discussion being had.

True. But USGA versus PGA is a huge difference in standards. Kinda like finishers scoring 35 under versus 5 under to win, for the same level of awesome play.

 

[PMantle]

True. But USGA versus PGA is a huge difference in standards. Kinda like finishers scoring 35 under versus 5 under to win, for the same level of awesome play.

Again, context.

 

[biscoe]

I'm having a hard time envisioning scenarios where "heroic" shots (or exceptional or whatever you choose to call them) should have an effect on par.

 

[Steve West]

I'm having a hard time envisioning scenarios where "heroic" shots (or exceptional or whatever you choose to call them) should have an effect on par.

That's because they shouldn't. However, there is no way to remove them from the dataset by just looking at scores. What teemkey is pointing out is that perhaps we can now remove them by looking at Udisc stats, which would produce a number for par even more closely related to the definition.

Here's an analogy: High heels shouldn't have an effect on categorizing people into not tall and tall. However, if the only information we have is how far the top of their head is from the floor, we can still do a pretty good job of determining who is tall. Maybe that cutoff point is 6'1". If we were able to subtract the height of each person's heels, we would have a more accurate way to tell who is tall, but the cutoff might change to 6'0".

 

[Doofenshmirtz]

However, there is no way to remove them from the dataset by just looking at scores.

Just like throws with errors. Until you also remove those from your analysis, inter alia, you aren't measuring par with SOCMOBR.

 

[Steve West]

Just like throws with errors. Until you also remove those from your analysis, inter alia, you aren't measuring par with SOCMOBR.

True, it's not a measurement of par; it is a way to come up with a number that is very likely to be the same number as if we could measure par.

 

[Doofenshmirtz]

. . . that is very likel . . .

Or so you hope, given your failure to analyze what you claim to. I fully expect that if you stopped including errors in SOCMOBR that it would spit out an even lower number for your definition of par.

 

[Steve West]

Or so you hope, given your failure to analyze what you claim to. I fully expect that if you stopped including errors in SOCMOBR that it would spit out an even lower number for your definition of par.

What your real agenda?

I'm not making a new definition, and I've come up with a pretty good way to look at scores and figure out what score would be expected of an expert with errorless play. I've acknowledged it has faults, and we're still working on improving it.

So, why do you keep making false and insulting statements?

What do you want to happen as a result?

 

[Steve West]

The germ of a method which might supplant my current method.

http://www.stevewestdiscgolf.com/TwoParameterModel.pdfhttp://www.stevewestdiscgolf.com/TwoParameterModel.pdf

 

[DavidSauls]

Can we start a campaign to get "errorless" out of the definition, and just go with expected score?

 

[lyleoross]

Can we start a campaign to get "errorless" out of the definition, and just go with expected score?

The problem with that is that someone has to admit they aren't playing perfectly when they score a birdie. Takes a ton of arrogance to call any thing done by a human being, errorless.

To err is human.

 

[Doofenshmirtz]

What your real agenda?

I have none.

I'm not making a new definition . . .

Ahh, so much to unpack. Of course it's not new, you've been using it for a long time.

But you did make up your own definition. I thought you had pretty much admitted to that. So, while I can't tell whether this is a semantic argument (e.g. "Only the PDGA can define par, thus I cannot make up a definition.") or a refusal that acknowledge that SOCMOBR, your statistical method, one that didn't change when the definition changed, isn't a statistical model for determining par, it is obviously a false.

. . . and I've come up with a pretty good way to look at scores and figure out what score would be expected of an expert with errorless play. I've acknowledged it has faults, and we're still working on improving it.

Except that you leave in the errors and then rely upon assumptions in ignoring that they are still there.

So, why do you keep making false and insulting statements?

What false statement? What insulting statements? Isn't "why do you keep making false and insulting statements?" intended to be a false and insulting statement?

What do you want to happen as a result?

As a result of what?

 

[biscoe]

As someone said way back Steve's method works great for evaluating par. IMO not so great for setting it however due to the chicken or egg thing with needing scores to be able to do it. This is fine though because setting par correctly is fairly easy in the vast majority of cases and in the cases where setting par is difficult it is fairly likely that neither answer is wrong.

 

[DavidSauls]

I think the help that Steve's method---or any other system for evaluating par---might bring to setting par is giving us a clearer idea of what we mean by "expected score".

 

[DG_player]

I think the help that Steve's method---or any other system for evaluating par---might bring to setting par is giving us a clearer idea of what we mean by "expected score".

I think that's key, Steve's system evaluates par.

It's what you do after you evaluate it that I think creates most of the disagreement.

 

[Steve West]

Speaking of evaluating, DGPT - Utah Open Presented by Prodigy Disc, 2018 Mulligan's Creekside did well. They could have put a par 2 label on hole 6, but hey.

The pars even worked pretty well for FPO, too.

Whether round 2 is included or not did not change any par by a whole parlecule.

 

[Steve West]

Sula Open 2018 presented by Discmania - EPT#2/Vasset Discgolfpark/Championship

 

[Doofenshmirtz]

I think that's key, Steve's system evaluates par.

By measuring it against what? "Evaluation" seems to be a specious talking point. Par doesn't need evaluation. Either the par that is set for a course matches the definition (i.e., is set by the TD) or it doesn't (made up by someone else, you know, like in this thread). The only evaluation necessary to determine if par is correct is to ask who set it. If the TD set it, then its correct. If not, then it's not par.

SOCMOBR is being used as a measuring stick, except that it isn't a calculation whose sum is par. Instead, it is an invention that ignores the definition of par, old and new and not just any one part of the par definition. It's just a way to use "mathiness" to make birdie reduction seem like it is based on the par definition when it isn't. You won't have to look too far back in this thread to see Steve admit that SOCMOBR has been tweaked to arrive at the number of birdies that Steve believes should occur in a tournament round. He also completely ignores the parts of the definition that he doesn't have the data to analyze - or just doesn't like.

Using SOCMOBR to "evaluate" par is like me making up a new definition of one foot, by eyeballing a stick for how close it matches the length of my forearm, and then going around "evaluating" the measurement that everyone else has made with a ruler to see if it is "correct."

 

[NegaSnapples]

By measuring it against what? "Evaluation" seems to be a specious talking point. Par doesn't need evaluation. Either the par that is set for a course matches the definition (i.e., is set by the TD) or it doesn't (made up by someone else, you know, like in this thread). The only evaluation necessary to determine if par is correct is to ask who set it. If the TD set it, then its correct. If not, then it's not par.

That's not the definition of par. That isn't even the definition of par that the PDGA uses. You're being willfully obtuse just to push your beliefs on what par is and isn't. At least include the entire definition and not just a snippet from it. It's been reference about 100 times in this thread. It shouldn't be hard to find.