Validity of CCRS

Part 1 - How well does average time predict team results?

Tomorrow at the Milaca Mega Meet, Maple Grove will enter the AAAA race as the meet's top ranked team, both in the coaches poll and by CCRS. Their CCRS rating, 108.9, indicates that they have been about 7 seconds faster than tomorrow's nearest challenger, Stillwater (108.2). If they do, in fact, run seven seconds faster than Stillwater, they would of course, you'd think, win. Unless they don't.

The flaw in using team CCRS numbers to predict a future meet is that there does not exist a perfect correlation between average team time and finishing place. Sometimes, teams that run slower manage to beat teams that run faster. Because Stillwater has a tight 1-5 spread and because Maple Grove's CCRS score is buoyed by the front-running Alex Miley, tomorrow could be one of those times. In fact, if Maple Grove runs exactly 7 seconds faster than Stillwater, I expect the meet to come down to only a few points in one direction or the other. This of course raises questions regarding the validity of CCRS as a general metric. Later I'll look at how well the program predicts future performance, but here I'd like to look at its underlying premise - that the faster team wins.

CCRS does not actually assess results, but rather the times that teams run to achieve those results. (Elo ratings are essentially the opposite, taking into account only wins and losses). Its an obvious and intuitive concept - you run faster in order to beat more people, so the faster team usually wins. However, if times fail to predict results often enough, then the entire premise of CCRS would be flawed.

So how often do average times successfully predict results? Luckily, that's easy to observe:

2016 Win % By Time Gap

Time gap(s)	Win %
1	63
2	75
3	81
4	83
5	95
6-10	97.4
11+	99.8
Total	98.4

In 1855 dual meet results this season, the faster team has come out on top 1825 times, or 98.4%. Of the 30 teams that defeated a faster opponent, 25 overcame a gap of 4 seconds or less. What does that mean? Exactly what you would imagine - faster teams win nearly every race. It indicates that average time is fairly predictive even at a gap of 1-2 seconds, and that a gap of 5 or more seconds (corresponding to about 0.5 CCRS points) is nearly insurmountable.

CCRS is simply a reflection of how fast a team has run so far this season, accurate within a fraction of a percent. Of course no team or individual can be expected to be perfectly consistent, so results will fluctuate from meet to meet and unexpected performances will occur. But again, we'll discuss the predictive success of CCRS later. For now, this (in nearly all cases) validates team CCRS as a highly effective metric for assessing performance.

It also means that if CCRS is to be wrong tomorrow, it will mark a rare event. If Stillwater overcomes a 7-second margin to win at Milaca, it will only be the 4th time (in 117 chances) that a team defeats one that runs 6-10 seconds faster. It is entirely possible, though. Elo ratings actually give Stillwater a 67 percent chance of defeating Maple Grove in a dual meet, and individual CCRS scores predict a dual meet score of 27-29 in their favor.

If 6-8 seconds is indeed the margin, Maple Grove will need several runners to finish between Miley and Stillwater's first runner in order for their favorite status to hold. Both Stillwater and Maple Grove are talented teams with aspirations for high finishes at the state meet, so the race should be fantastic regardless.