Getting Mileage out of Milwaukee Part 1

Hey there Internet! It's Guido here with another Scoring Indy update. This week, I'll be doing a pair of updates. In both, we'll be looking at how to score an oval race. Today, I'll be laying some of the groundwork for what I hope to be a grand unveil of the final formula on Friday. So, let's get to it.


Something Old

This part is probably old news, but my original formula for comparing drivers using Race Scores doesn't work well in long races with lots of passing (ie ovals). You need only look back at the scores that this formula generated for the MAVTV 500 in Fontana back in June to see that. Those scores are here. Now there's only one glaring problem with this, and it's back in P7: the curious case of Buckshot Jakes. Now, I'm all for giving credit to drivers who move from the back to the top 10. Jakes deserves that credit. But, as Andy the Speedgeek (one of my tens of readers and top commenter) so aptly noted, he was in the "camera shot" for less than a minute over the course of a 500 mile race. This doesn't mean he was bad. It does mean he wasn't second-best.

So, I'm going to present a couple of case studies here today. First, we'll look at Milwaukee and fiddle with the numbers a little bit to try and get a sense of how important the "on track" term of our formula is. That term combines positions gained and laps led. And, after isolating that, we will come back on Friday to try and complete the picture.


Case Studies

Here is a mega-sheet of various scorecards for Milwaukee, which assign various weights to the two terms (finishing position and performance)

Finish	Driver	Grid	Led	Completed	50/50	67/33	75/25
1	Bourdais	11	118	250	95.34	96.89	97.67
2	Castroneves	24	0	250	95.65	95.65	95.65
3	Rahal	6	5	250	53.17	65.88	72.24
4	Montoya	8	0	250	52.17	63.77	69.57
5	Newgarden	1	109	250	54.41	63.81	68.51
6	Kanaan	4	3	250	35.38	49.68	56.82
7	Dixon	10	14	250	46.28	55.49	60.10
8	Andretti	9	0	250	36.96	47.83	53.26
9	Pagenaud	17	0	250	50.00	55.07	57.61
10	Carpenter	22	0	250	56.52	57.97	58.70
11	Chaves	12	0	250	30.43	39.13	43.48
12	Kimball	7	0	250	15.22	27.54	33.70
13	Hunter-Reay	16	0	250	30.43	36.23	39.13
14	Sato	13	0	250	19.57	27.54	31.52
15	Munoz	18	0	250	26.09	30.43	32.61
16	Vautier	20	0	248	26.02	28.94	30.40
17	Hawksworth	12	0	221	5.61	13.88	18.02
18	Wilson	15	1	219	7.51	13.70	16.80
19	Karam	3	0	183	-14.59	-2.48	3.57
20	Coletti	19	0	156	7.34	10.69	12.37
21	Briscoe	2	0	130	-14.96	-5.62	-0.96
22	Power	14	0	130	-4.70	-0.23	2.00
23	Jakes	5	0	113	-15.51	-8.89	-5.58
24	Mann	23	0	27	-0.23	-0.16	-0.12

Diving In

So, there you see three lovely color-coded columns. Each of the columns represents the mathematical "weight" given to finishing position and the "on track" term. Finishing position is first, and it gets more important as one moves from left to right. And, as you see, as the emphasis on the "on track" term decreases EVERYONE'S (well, everyone except Helio's) Race Scores increase.

This is due to the fact that the "on track" term is a) usually a smaller number than the Finishing Position term, and b) the only one of the two terms with the potential to be a negative number. So, let's now look at a break down of how those two terms interact. Here's the how those terms look for the 50/50 split:

Finish	Driver	Grid	Led	Completed	FP	OT	Race Score
1	Bourdais	11	118	250	50.00	45.34	95.34
2	Castroneves	24	0	250	47.83	47.83	95.65
3	Rahal	6	5	250	45.65	7.52	53.17
4	Montoya	8	0	250	43.48	8.70	52.17
5	Newgarden	1	109	250	41.30	13.10	54.41
6	Kanaan	4	3	250	39.13	-3.75	35.38
7	Dixon	10	14	250	36.96	9.32	46.28
8	Andretti	9	0	250	34.78	2.17	36.96
9	Pagenaud	17	0	250	32.61	17.39	50.00
10	Carpenter	22	0	250	30.43	26.09	56.52
11	Chaves	12	0	250	28.26	2.17	30.43
12	Kimball	7	0	250	26.09	-10.87	15.22
13	Hunter-Reay	16	0	250	23.91	6.52	30.43
14	Sato	13	0	250	21.74	-2.17	19.57
15	Munoz	18	0	250	19.57	6.52	26.09
16	Vautier	20	0	248	17.39	8.63	26.02
17	Hawksworth	12	0	221	15.22	-9.61	5.61
18	Wilson	15	1	219	13.04	-5.54	7.51
19	Karam	3	0	183	10.87	-25.46	-14.59
20	Coletti	19	0	156	8.70	-1.36	7.34
21	Briscoe	2	0	130	6.52	-21.48	-14.96
22	Power	14	0	130	4.35	-9.04	-4.70
23	Jakes	5	0	113	2.17	-17.69	-15.51
24	Mann	23	0	27	0.00	-0.23	-0.23

FP -- The Finishing Position term

OT -- The "On Track" term

So, in this break-down, you can see that Bourdais and Castroneves are the big beneficiaries of the "on track" term. Bourdais advanced his position by 10 over the course of the race and led a TON of laps, while Castroneves moved up a whopping 22 grid positions.

The next two biggest beneficiaries are Carpenter and Newgarden. Carpenter gets a sizeable bump from advancing 12 grid positions, whereas Newgarden only gains 13.10 in that term. He's penalized for dropping grid positions (the curse of the pole) and the bonus that he gets for leading laps only brings him up to that 13.10 level.

I'm not opposed to advancing position being more important than leading. This is because leading can often (and I'm mostly talking about road/street races here, but for ovals it's a little true) be a function of strategy, whereas where one ends the race is, in fact, what counts. I want to take multiple factors into account, that's why I have this blog, BUT where you finish matters (especially where you finish relative to your start).

The problem that I have with the 50/50 weighting system is that it gets a little TOO "swingy." I don't think Helio should get 47.83 of his score for advancing position, especially because most of his gains came late in the race. Likewise, I don't think drivers who were very much in the "camera shot" like Karam and Briscoe should be penalized that severely. They should definitely be penalized, mind you, but not that severely.

So, here's the break-down for 75/25:

Finish	Driver	Grid	Led	Completed	FP	OT	Race Score
1	Bourdais	11	118	250	75.00	22.67	97.67
2	Castroneves	24	0	250	71.74	23.91	95.65
3	Rahal	6	5	250	68.48	3.76	72.24
4	Montoya	8	0	250	65.22	4.35	69.57
5	Newgarden	1	109	250	61.96	6.55	68.51
6	Kanaan	4	3	250	58.70	-1.87	56.82
7	Dixon	10	14	250	55.43	4.66	60.10
8	Andretti	9	0	250	52.17	1.09	53.26
9	Pagenaud	17	0	250	48.91	8.70	57.61
10	Carpenter	22	0	250	45.65	13.04	58.70
11	Chaves	12	0	250	42.39	1.09	43.48
12	Kimball	7	0	250	39.13	-5.43	33.70
13	Hunter-Reay	16	0	250	35.87	3.26	39.13
14	Sato	13	0	250	32.61	-1.09	31.52
15	Munoz	18	0	250	29.35	3.26	32.61
16	Vautier	20	0	248	26.09	4.31	30.40
17	Hawksworth	12	0	221	22.83	-4.80	18.02
18	Wilson	15	1	219	19.57	-2.77	16.80
19	Karam	3	0	183	16.30	-12.73	3.57
20	Coletti	19	0	156	13.04	-0.68	12.37
21	Briscoe	2	0	130	9.78	-10.74	-0.96
22	Power	14	0	130	6.52	-4.52	2.00
23	Jakes	5	0	113	3.26	-8.84	-5.58
24	Mann	23	0	27	0.00	-0.12	-0.12

So, just by looking at this this break-down you can see that the "on track" term has, in fact, been made less "swingy." The problem that I have now is that it's gone too far the other way. I think this is pretty much the right weight for the "on track" term, but you can't leave everything up to finishing position. This, of course, brings us up to what we'll be discussing on Friday. How to find a term that shows us "how they ran," without me having to input thousands of data points by hand. So...

Stay Tuned

On Friday, I'll be back trying to nail down the newest edition of our formula: a little term that I'm calling "Representative Running Position." See you then!

-- Guido