Disc Golf Data

Which discs and companies fly like we expect?

2021-07-19T00:00:00+00:00

Recently I looked at how discs have changed through time - and how manufacturers have updated their disc technology. It was suggested that I should look at how companies compare in the consistency of their discs.

Looking at disc consistency itself is a bit difficult to do - I don’t have, for example, access to have individual reviews of thousands of individual Innova Destroyers. Instead what we can do is compare manufacturer flight numbers to the number reported by Infinite Discs. On their website, users can give their own flight numbers to a disc and then Infinite posts those aggregated flight numbers to each manufacturer’s page (e.g. See all of Discraft’s discs and see how the flight numbers are not a 1:1 match to Discraft’s own numbers).

Of course, there’s still differences like different plastics, runs, and other sources of variation - but by comparing Infinite’s numbers to manufacturer numbers we can at least ask questions like:

Which companies have the most predictable flight numbers?
Which individual discs are most true to their numbers? Which are most divergent?

So, let’s get into it.

Which companies have the most predictable flight numbers?

In the above plot, Innova comes in as having the most accurate flight numbers and Gateway has the least. It’s interesting to note that Gateway and Discmania both have really large error bars - this means that some of their discs are exactly what people expect and some are nothing like what people expected.

Shown above is the “total difference” with no sense of directionality (i.e. we don’t know if people thought these discs were “too fast” for their numbers or “too slow”…we just know if they were as expected or not). For every disc I calculated the difference between each manufacturer flight number and Infinite Disc’s flight numbers and then added up the absolute value of all those differences. Then, the mean value is presented for each manufacturer. For example, an Axiom Envy has a manufacturer’s speed of 3 but Infinite users scored it as 2.7. We repeat that calculation for each flight number and add up all the differences to get a total divergence score.

tl;dr: Discs with high numbers are bad. Discs with low (or zero) are good - exactly as expected by the thrower.

Next, let’s consider some flight numbers with directionality preserved. To save you from chart overload, we will only look at Speed. The other values all sort of correlated as expected from that anyway (i.e. if users thought a disc was faster than its numbers, they also thought it had less turn and more fade).

How do companies compare across their expected speed?

Lastly, here’s a couple tables of discs that are most true (divergence = 0) and some that are least true (divergence score > 2) from their manufacturer flight numbers.

What discs are most true to their numbers?

	Speed	Glide	Turn	Fade
Discmania Dd3	12	5	-1.0	3.0
Discmania Md1	4	4	0.0	2.0
Discmania Link	2	3	0.0	1.0
Discraft Wildcat	11	5	-2.0	3.0
Discraft Talon	8	3	1.0	3.0
Dynamic Discs Witness	8	6	-3.0	1.0
Dynamic Discs Guard	2	5	0.0	0.5
Gateway Rage	10	5	-0.5	3.5
Gateway Hybrid	7	4	-0.5	3.0
Innova Destroyer	12	5	-1.0	3.0
Innova Krait	11	5	-1.0	2.0
Innova Max	11	3	0.0	5.0
Innova Cheetah	6	4	-2.0	2.0
Innova Gargoyle	6	3	-1.0	2.0
Innova Python	7	3	-1.0	2.0
Innova Raven	6	4	-2.0	2.0
Innova Teebird	7	5	0.0	2.0
Innova Viking	9	4	-1.0	2.0
Innova Barracuda	5	4	-2.0	3.0
Innova Bulldog	4	3	0.0	2.0
Innova Griffin	5	3	1.0	3.0
Innova Lycan	4	5	0.0	1.0
Innova Scorpion	5	5	-1.0	2.0
Innova Aviar	2	3	0.0	1.0
Innova Birdie	1	2	0.0	0.0
Innova Classic Roc	3	3	0.0	2.0
Innova Coupe	2	3	-1.0	1.0
Innova Nova	2	3	0.0	0.0
Innova Pig	3	1	0.0	3.0
Innova Stud	3	3	0.0	2.0
Latitude 64 Primitus	9	5	0.0	2.0
Latitude 64 Medius	5	5	-1.0	2.0
Latitude 64 Macana	2	5	0.0	1.0
Prodigy F3	7	5	-1.0	2.0
Prodigy Pa1	3	3	0.0	2.0

What individual discs are the most unlike their numbers?

	Speed	Glide	Turn	Fade
Discmania Enigma	13	5	-1.0	3
Discmania Md5	5	4	0.0	2
Discmania P3	3	2	0.0	3
Discraft Flick	9	4	1.0	4
Dynamic Discs Escape	9	5	1.0	2
Gateway Apache	9	6	-1.0	2
Gateway Diablo	9	5	0.0	2
Gateway Illusion	12	5	-1.0	2
Gateway Speed Demon	12	4	0.0	3
Gateway Blaze	6	4	0.0	3
Gateway Voodoo	2	3	0.0	2
Latitude 64 Saint Pro	8	5	-0.5	2
Latitude 64 Trident	6	4	-0.5	3
MVP Tesla	9	5	-1.0	2
MVP Entropy	3	3	0.0	4

There ya have it. Does this match what you expect? Are the numbers simply just made up? I think this analysis really just scratches the surface and shows what could be possible with better data. We could compare plastics, runs, retooling of molds, different user ratings, etc. I’d reckon a big part of why Innova scores so well is because it’s sort of the “Standard”. All 12, 5, -1, 3 drivers are judged against the Innova Destroyers as the standard. So anything that diverges from your expectation as a “Destroyer Replacement” gets funky numbers (even though Destroyers are known for their big differences between runs…).

Anyway - I hope y’all found this enjoyable. :)

If you like these analyses please consider supporting my project by purchasing discs through the Amazon or Infinite Discs links provided above. Head over to my disc database to explore all of Infinite Disc’s stock in a simple table.

Who will win and who will outperform at the Las Vegas Challenge?

2021-02-24T00:00:00+00:00

With the first big tournament of the season this weekend, I thought it would be fun to take a look at the current rankings and past results to see if we could make some guesses about who will win the Las Vegas Challenge.

Admittedly, I’m not hugely familiar with how these types of analyses are done for something like ball golf and I don’t have access to a big sophisticated dataset. If possible, it would be nice to include things like who putts better in the wind, distance v. technical courses, injuries, etc etc. But that’s beyond me right now.

Additionally, my day job is to do analysis with ecological data and not really make “predictions”. BUT! All those excuses aside, I figured we could do a little poking around nonetheless.

I operated under a few plain assumptions to determine who I think will win:

Top players (highest PDGA ratings) will perform well
Players who have played well at the LVC before will do well
A player’s performance at the LVC in a recent year is more valuable than an earlier year

Here’s who I think will win ranked in order for both the MPO and the FPO. First is a scrollable table of the rankings I generated and second is a plot that shows you who has over- or under-perform based on their player rating compared to previous LVC finishes. That is, players who “overperform” are those who play better at the LVC than they “should” compared to their PDGA rating. (Granted, there’s only 4 tournaments in the dataset)

Below the tables and figures you can see a description of how I generated my results.

MPO

PDGA #	Name	2020 Ratings Rank	Mean LVC Place	Weighted LVC Place	Times Played	LVC Performance Weight	Weighted Score	Predicted Finish
37817	Mcmahon	3	2.0	1.8	3	-1.2	9.7	1
38008	Wysocki	1	3.8	4.3	4	3.3	10.6	2
45971	Heimburg	2	5.5	6.1	2	4.1	15.3	3
27523	Mcbeth	1	6.3	7.7	3	6.7	16.3	4
18824	Sexton	4	6.2	5.7	4	1.7	16.4	5
24341	Leiviska	8	9.2	9.4	4	1.4	26.8	6
41760	Jones	8	6.7	6.3	3	-1.7	27.7	7
26416	Brathwaite	10	16.8	14.9	4	4.9	46.8	8
38464	Withers	5	22.0	19.5	4	14.5	49.0	9
48346	Gibson	8	20.3	17.7	3	9.7	55.3	10
63765	Presnell	11	14.0	13.2	2	2.2	57.4	11
13864	Gurthie	7	21.3	17.7	3	10.7	60.3	12
44382	Barela	18	21.0	19.2	3	1.2	64.3	13
50401	Clemons	12	22.3	23.1	3	11.1	65.2	14
57493	Rothlisberger	24	23.0	23.0	1	-1.0	67.0	15
17295	Conrad	11	25.3	29.0	3	18.0	70.0	16
68835	Hannum	17	27.0	22.1	3	5.1	70.2	17
53332	Barham	21	20.0	21.2	3	0.2	71.3	18
47472	Keith	16	18.3	18.9	3	2.9	71.8	19
66362	Russell	19	27.8	24.4	4	5.4	76.8	20
99053	Tanner	38	4.0	4.0	1	-34.0	77.0	21
53565	Oakley	17	20.8	16.4	4	-0.6	79.8	22
27171	Ulibarri	15	28.7	35.8	3	20.8	86.7	23
49431	Castro	25	28.0	28.7	2	3.7	87.4	24
121715	Aderhold	24	13.0	13.0	1	-11.0	91.0	25
54049	Turner	23	24.5	30.0	2	7.0	91.0	25
26269	Skellenger	29	29.0	29.0	1	0.0	98.0	26
33705	Koling	13	37.0	43.3	3	30.3	102.7	27
59419	Samson	41	14.0	14.0	1	-27.0	114.0	28
56555	Meintsma	22	38.0	36.4	3	14.4	116.9	29
58814	Messerschmidt	24	30.3	34.6	3	10.6	118.1	30
35876	Montgomery	21	39.3	35.2	3	14.2	118.4	31
57425	Elmore	27	40.3	38.6	3	11.6	119.1	32
65737	Perkins	30	35.5	42.5	4	12.5	121.0	33
85850	Gilbert	21	29.7	28.8	3	7.8	124.6	34
36777	iv	41	33.0	33.0	1	-8.0	126.0	35
68286	Queen	15	39.0	39.0	1	24.0	129.0	36
45879	Earhart	26	45.0	45.0	1	19.0	130.0	37
15857	Barsby	18	49.0	57.0	2	39.0	132.0	38
41918	Shotwell	34	50.3	42.1	3	8.1	136.2	39
50312	Johnson	35	46.0	47.5	4	12.5	141.0	40
57365	Hammes	7	57.0	57.0	1	50.0	144.0	41
51740	Spradlin	41	23.0	23.0	1	-18.0	149.0	42
81039	Polidori	36	48.0	48.0	1	12.0	151.0	43
91498	Keseloff	41	45.5	48.0	2	7.0	161.0	44
85457	Gilbert	35	39.0	39.0	1	4.0	165.0	45
48950	Bell	22	75.0	63.3	3	41.3	168.7	46
80636	Lyon	37	37.3	33.1	3	-3.9	170.2	47
74719	Pinegar	31	59.3	52.6	3	21.6	171.1	48
72423	Hoop	17	58.0	55.3	2	38.3	172.6	49
81739	White	26	65.0	62.1	2	36.1	182.3	50
49519	Beckner	41	55.5	53.4	4	12.4	198.8	51
56485	Herr	41	52.0	52.0	1	11.0	203.0	52
41220	Metzler	39	61.0	61.0	1	22.0	203.0	52
43762	Hebenheimer	41	56.0	56.0	1	15.0	205.0	53
96832	Guthrie	41	56.0	56.0	1	15.0	207.0	54
56486	Herr	34	82.5	80.4	2	46.4	213.9	55
49865	Meyer	41	48.0	48.0	1	7.0	216.0	56
28091	Berger	41	74.5	78.0	2	37.0	218.0	57
88497	Martin	28	70.5	71.0	2	43.0	226.0	58
43369	Bates	41	72.0	72.0	1	31.0	228.0	59
19453	Kapalko	41	84.0	70.6	3	29.6	229.1	60
46190	Sullivan	41	68.0	70.8	3	29.8	237.7	61
49649	Smith	41	76.0	76.0	1	35.0	239.0	62
89816	Cookson	41	64.0	64.0	1	23.0	246.0	63
45478	Bilodeau	19	90.5	87.8	2	68.8	250.5	64
36806	Callahan	41	90.0	90.0	1	49.0	252.0	65
68103	Stoll	41	68.0	68.6	2	27.6	254.1	66
37909	Russell	41	79.7	82.2	3	41.2	254.3	67
29812	Dryden	41	83.0	83.0	1	42.0	261.0	68
52776	Laputka	41	80.5	88.8	2	47.8	263.5	69
62910	Miller	41	78.2	73.1	4	32.1	264.2	70
34964	Whalen	41	68.0	68.0	1	27.0	265.0	71
49514	Castillo	41	81.0	81.0	1	40.0	265.0	71
109666	Caplin	41	81.0	81.0	1	40.0	270.0	72
78421	Saggboy	41	77.0	77.0	1	36.0	271.0	73
79978	McDaniel	41	90.0	90.0	1	49.0	273.0	74
77246	Lucerne	41	79.5	80.2	2	39.2	274.4	75
96436	Johnson	41	85.0	85.0	1	44.0	281.0	76
127314	Schram	41	81.0	81.0	1	40.0	307.0	77
61067	Sutherland	41	82.5	87.3	2	46.3	313.7	78
69796	Vassari	41	111.0	111.0	1	70.0	366.0	79
73363	Derochie	41	132.0	132.0	1	91.0	389.0	80
75420	Malone	41	149.0	149.0	1	108.0	446.0	81

Notes: Darkness of point represents how many times the player has played the LVC (lightest is 2 times, darkest is 4 times). Only the top 20 shown.

In the plot, players to the left of the dashed line have performed better in past LVCs than expected based on their player rating. Those to the right, have performed worse. Keep in mind the higher ranked a player is, the harder it will be for them to “outperform” their rating - for example since Ricky Wysocki is ranked #1, he would have to have won every LVC to break even.

So it looks like according to the weighted score Eagle McMahon is favored to win. Kevin Jones typically performs a little better at the LVC than other tournaments and comes off his big win at the end of last season.

However, I think given the amount of data here, it’s pretty reasonable to expect anyone in the top ~10 to win and even then someone lower ranked could have a really breakout performance. For example, if the over/under chart had included 30 players, Tristan Tanner would be a huge outlier with a LVC Weighted Score of -34 due to his big performance and 4th place finish last year.

FPO

PDGA #	Name	2020 Ratings Rank	Mean LVC Place	Weighted LVC Place	Times Played	LVC Performance Weight	Weighted Score	Predicted Finish
44184	Allen	4	1.8	1.6	4	-2.4	6.2	1
29190	Pierce	1	3.0	4.0	3	3.0	10.0	2
34563	Hokom	2	4.3	4.8	3	2.8	13.7	3
15354	Allen	6	3.7	3.8	3	-2.2	17.7	4
50656	Weese	7	6.0	5.7	4	-1.3	19.4	5
81351	King	4	6.0	6.0	1	2.0	21.0	6
32654	Fajkus	13	9.5	10.4	2	-2.6	29.8	7
48976	Scoggins	5	13.0	13.0	1	8.0	31.0	8
59431	Walker	13	9.2	9.3	4	-3.7	32.6	9
64751	McMorran	16	7.5	7.1	2	-8.9	33.3	10
51229	Bradley	10	12.0	12.0	1	2.0	36.0	11
58303	Ananda	17	10.5	11.0	2	-6.0	40.0	12
39504	Andyke	11	14.5	14.8	2	3.8	44.5	13
32917	Cox	17	14.5	13.9	4	-3.1	47.8	14
34751	Bailey	14	13.5	14.8	2	0.8	51.5	15
27832	Panis	17	15.2	14.4	4	-2.6	58.8	16
60798	Tomaino	20	17.0	17.0	1	-3.0	62.0	17
71262	Stinchcomb	18	16.3	15.1	3	-2.9	63.3	18
104030	Allocco	26	18.0	18.0	1	-8.0	73.0	19
84007	Maes	26	19.0	19.0	1	-7.0	94.0	20
63005	Bilodeau	26	27.0	27.0	1	1.0	96.0	21
56987	Richardson	26	30.0	30.0	1	4.0	108.0	22

Catrina Allen is favored in the FPO with Paige Pierce right behind her. There are several overperformers in the FPO compared to the MPO, which I found interesting. Probably there is more variability in who succeeds at the LVC in the FPO. Interestingly an email just came from the folks at SkipAce saying their expected overperformer will be Ohn Scoggins due to her spin putting in the wind. My analysis doesn’t capture that, so it will be fun to watch for her play.

Alright! That’s the basic analysis I ran. Hopefully you find it interesting. Because I’ve got this set up now, I’m aiming to in the future add more variables and sophistication into the analysis, maybe have an actual model going. Right now this is pretty simple, but with my limitations in time and current knowledge, we’ll see where it goes.

If you like this, consider leaving a tip or using some of the links on the disc database to purchase discs and help me support my hosting costs.

Peace,

John

To generate the rankings, I did the following things.

First, I gathered the PDGA ratings for all players from 2020. Next, I gathered the results of the past four Las Vegas Challenges (technically called the “Gentlemen’s Club Challenge” in older years). I used the past four simply because I figured only recent events would be reflective and five years ago just seemed off when I looked at the results. Subjective, but it’s what I went with.

Taking those basic assumptions I calculated a few things for every player. First, I calculated their “mean place” in previous LVCs (so if they came in 10th once and 6th another time, their mean place = 8). Next, I calculated their weighted mean place, basically giving greater importance to more recent LVCs. Each player was then ranked according to their rating in 2020 as well as their ratings at all LVCs. For all of these measurements, a lower score is better (just like disc golf).

Lastly, to give a little more weight to the LVC over just general ratings, I calculated the difference between a players weighted mean LVC placement and their rank position in the 2020 ratings. Players with negative values are therefore expected to typically overperform at the LVC. With all that, we can then calculate a total score (our expectation of who will win given their PDGA rating and past performance at LVC).

Oh - and then I filtered out players who aren’t registered for the tournament (or at least on the waitlist at DGScene).

Whew.