Hole and Course Performance Statistics

[ToddL]

A new paper focusing on just the ability to assign the scores to players, not the ability to generate a lot of scores.

http://www.stevewestdiscgolf.com/Calculating_Sorting_Ability.pdf

paper written by an LLC? I guess Citizens United is correct in the end...

 

[Steve West]

Some messing around with a course performance stat and the DGPT Dream 18 votes.

http://www.stevewestdiscgolf.com/Are_the_Most_Popular_Holes_the_Best_at_Testing_Skill.pdf

 

[Steve West]

For 2020 DGPT - Dynamic Discs Open/Emporia Country Club/ECC Longs; 18 holes; Par 65; 10,980 ft.

Given a group of six players rated 970, 980, 990, 1000, 1010, and 1020 this course will assign a lower score to the higher rated player an average of 21.14 times out of the 30 possible pair-wise combinations of players. This places it at the 39th percentile (i.e. in the best half) of the courses I've studied.


Here is how each hole contributed.

 

[Cgkdisc]

It would appear the chart confirms that a player's golf distance is the most important factor for sorting skill levels?

 

[Steve West]

It would appear the chart confirms that a player's golf distance is the most important factor for sorting skill levels?

That would be an unjustified conclusion, though not necessarily wrong.

What can be said is that on this particular course there is a tendency for longer holes to contribute more to the ability to sort players by rating. Because longer holes generally have higher scores, that could also be stated as a tendency for holes with higher scoring averages to be better at sorting.

On this course, the relationship between length and sorting ability is actually stronger than the relationship between average and sorting ability. Perhaps pumping up the scores of shorter holes was done in a way that was not as effective as lengthening those holes would have been. Maybe.

 

[Steve West]

Since there were 45 players who played all three rounds at both DDO and TPC, I did a little comparison.

For Scoring Spread of total scores, DDO did a better job, with a scoring spread of 24.35 vs. TPC's 22.59. To put it another way, the typical player at DDO was tied with 0.85 other players, while at TPC they were tied with 0.99 other players.

Because single-point scoring spreads can be quirky, I also looked at how the courses would do based on 10,000 simulated rounds, with scores for each hole picked randomly from all three rounds.

DDO's averaged a score of 65.87, while TPC's average was 56.00. In 100% of the simulated rounds, DDO had a higher average score.

DDO generated an average scoring spread of 15.47, beating TPC's average of 12.71. In 94% of the simulated rounds, DDO had a better scoring spread.

 

[Steve West]

I tested whether we could find types of courses that favor specific types of players.

http://www.stevewestdiscgolf.com/To_test_whether_certain_players_are_better_on_certain_courses.pdf

Spoiler

No.

 

[Casey 1988]

I tested whether we could find types of courses that favor specific types of players.

http://www.stevewestdiscgolf.com/To_test_whether_certain_players_are_better_on_certain_courses.pdf

Spoiler

No.

This is due to now even in top Am 1, on both Male and female side the players can play a variety of courses due to several factors, 1 there is more courses and styles to play now, 2 YouTube of how to play courses, and 3 Courses if enough space are more varied in shot design.

 

[Steve West]

A few stats from Idlewild.

 

[Steve West]

2020 DGPT - Idlewild Open presented by Dynamic Discs & The Nati/Idlewild/Idlewild Open 2020 MPO; 18 holes; Par 68; 9,194 ft. as a course came in at 70%%, or 44th percentile among all courses. Right between Toboggan and Brewster Ridge.

Hole 17 is the safest one to mess with.

This chart focuses on players rated 970-1020, the charts in the post above are based on all MPO players.

 

[ballgolfconvert]

How does weather conditions play into this or does it reduce to not being very meaningful? A headwind vs a tailwind can change the characteristics of how a hole plays. So how many of these holes would have different rankings if let's just say the wind was blowing in the opposite direction?

 

[Steve West]

How does weather conditions play into this or does it reduce to not being very meaningful? A headwind vs a tailwind can change the characteristics of how a hole plays. So how many of these holes would have different rankings if let's just say the wind was blowing in the opposite direction?

These show the performance of the hole. They are not a direct measure of any characteristic of a hole. Just like a player's score shows how they performed and is not a direct measure of how good a player they are.

How much difference it makes is unknowable.

 

[nothinbuttree]

2020 DGPT - Idlewild Open presented by Dynamic Discs & The Nati/Idlewild/Idlewild Open 2020 MPO; 18 holes; Par 68; 9,194 ft. as a course came in at 70%%, or 44th percentile among all courses. Right between Toboggan and Brewster Ridge.

Hole 17 is the safest one to mess with.

This chart focuses on players rated 970-1020, the charts in the post above are based on all MPO players.

How are those hole lengths being measured?

 

[Steve West]

How are those hole lengths being measured?

Those in particular were from Udisc, but for another course. Just ignore them.

 

[nothinbuttree]

Those in particular were from Udisc, but for another course. Just ignore them.

OK. I think it notable that the two par 5's are the highest--again as others have mentioned, possibly due simply to more spread. Even on a tough hole, a par 3, will not have as large a dispersion of scores as a par 5 will.

Also, I am not surprised 17 is not very predictive. Lots of things happen to otherwise very good drives there. I am not alone in that sentiment, nearly every commentator I have heard over the years mentions something along those lines...

 

[Billipo]

Translating performance to par

Steve,

I found another document that you created which shows several ways to determine par based on player performance. Referencing 1000 rated players.

I strongly agree actual scores are the best way to determine par. Based on averages, median, spreads, etc. I also feel the posted should coincide with intended skill level when multiple layouts exists. IE: Blue use blue rating averages, white use white rating average, etc.

I know the original post is much more scholarly than my simple inquiry, but I'm interested a way to put numbers into practice.

I do have a question on average score numbers in your document. Par 3s up to 3.75 avg., Par 4s up to 4.95, and par 5s up to 6.05.

These seem high. I would think would be more like Par 3s up to 3.6avg., Par 4s up to 4.6, and par 5s up to 5.6.

Interested in thoughts.

 

[Steve West]

Steve,

I found another document that you created which shows several ways to determine par based on player performance. Referencing 1000 rated players.

I strongly agree actual scores are the best way to determine par. Based on averages, median, spreads, etc. I also feel the posted should coincide with intended skill level when multiple layouts exists. IE: Blue use blue rating averages, white use white rating average, etc.

I know the original post is much more scholarly than my simple inquiry, but I'm interested a way to put numbers into practice.

I do have a question on average score numbers in your document. Par 3s up to 3.75 avg., Par 4s up to 4.95, and par 5s up to 6.05.

These seem high. I would think would be more like Par 3s up to 3.6avg., Par 4s up to 4.6, and par 5s up to 5.6.

Interested in thoughts.

The difference between par and average is the extra throws caused by errors. (Technically, unmitigated errors: the net of the extra throws caused by errors minus the throws shaved by good luck.)

The higher the par, the more opportunities for errors. So, as par goes up, the difference between the average score and par increases. These particular values were the best fit of average score to the value generated by my preferred method: picking the lowest score for which at least 76.7% of throws were good enough to get par. See page 3 of http://www.stevewestdiscgolf.com/ScoringDistributionGraphs.pdf.

I don't much like using these average ranges. If you have the data to compute averages for 1000-rated players, you can use the better method. This method is just in there for the sake of the old stick-in-the-muds who cannot grasp that par is not average. These ranges are better than just rounding the average score to the nearest integer, but there will be a couple of holes per course where these ranges would not give the "correct" par.

 

[Billipo]

Thanks for response.

Besides I depth analysis, I do believe there should be a simpler, easily applied, universally recognized way to determine par based on performance versus the pdga distance/ foliage colorful chart or reach green plus two.

A simple system may not be perfect, but better is always better. Especially if using real performance numbers. I forsee it more of a litmus test using distances, averages (skill appropriate players), score distribution, special circumstances (water,dogleg, judgement) as factors towards determining par. Ideally plug in scores (maybe current player rating), spits out proposed par.

Something universally understood for practical application by average Joe. Just my thought.

 

[Ryan P.]

It'd help a lot if designers made and tested holes a bit before setting par. It's not easy, but creating a course and having people at the appropriate ratings come test it would really help.

 

[Cgkdisc]

Course design validation process. I would probably tweak this article based on some of the work Steve West has done but the basic idea of testing layouts for the skill level intended is still important, at least for courses used for tournament play.