NROC - Share Your Ideas for Race Series

[jmac]

Pros race for position, not time

Of course people race for position. That's equally true for road races whether on bikes or on skates. Name a pro bike road race in which people don't race for position? Look at all those furious sprint finishes in TdF stages, with people thrusting their bikes forward at the line, and tell me they're not racing for position!

The claim that basing a road-race series on times takes the racing out of races is simply untrue, as the TdF example makes abundantly clear.

This Series is to support the sport, not morph it.

To create any kind of series at all *IS* morphing the sport!

The purpose of creating a series is indeed to support the sport — *BY* morphing it, in a positive and healthy way that enhances the overall experience of both competitors and fans.

[sxevegan]

And in which the overall standings in the series are based on cumulative time, not placements.

Hmmmmmmmm. . . . . . . .

Seeing as how the most famous and successful pro/elite road-race series on the entire planet bases its standings on time rather than placements, I don't see why we couldn't or shouldn't do the same.

The green jersey has nothing to do with time and everything to do with placement.

[sxevegan]

Strategy and tactics? ...A chess game? ...Reading one's opponents? — Gee, it sounds like you're describing the action in the Tour de France.

Which is, most definitely, road racing.

What Jim described can be said for any bike race, not just the TDF. Whether its a 30 minute crit or a grand tour you'll have strategy and tactics like a chess game.

Yes, the grand tours go by overall times. Other stage races (omniums) go by points, similar to how the green jersey works in the tour.

What you could do is have each person take their 3 or 4 fastest times and average them together. That way as long as they make 3 events they can be competative. The problem i see with this is that even when a race is called a marathon it's hardly ever exactly 26.2 miles. Texas Road Rash is about 28 IIRC and Chicagoland is about 25. That 3 mile difference adds up to a couple minutes. The better alternative would be to go by pace.

With all that said, I still don't think going by time would be the best solution. I think the points system (regardless of how it is done) is better.

[jmac]

Yes, the grand tours go by overall times. Other stage races (omniums) go by points, similar to how the green jersey works in the Tour.

Quite true. My point was simply that there's plenty of examples of successful road-race series whose overall standings are based in one way or another on competitors' finish times in the individual races. There is no inherent incompatibility between road racing and time-based series.

What you could do is have each person take their 3 or 4 fastest times and average them together. That way as long as they make 3 events they can be competitive. The problem i see with this is that even when a race is called a marathon it's hardly ever exactly 26.2 miles. Texas Road Rash is about 28 IIRC and Chicagoland is about 25. That 3 mile difference adds up to a couple minutes. The better alternative would be to go by pace.

This illustrates yet another benefit of the calibrated time-based points system I've proposed: the number of points awarded to any competitor is mathematically independent of not only the number of competitors in the race, but also of the length of the race, and of the relative difficulty of the course or the racing conditions. All that matters is how each competitor performs, relative to the winner of that race. Any factor that affects all competitors equally, such as distance and weather, is automatically neutralized. This makes all events equally "modular", meaning that it's very easy to fairly allow for "X best results of Y races" accommodations for people who can't attend every single race in the series.

This doesn't mean that different events couldn't be given different "weights" within the series. In fact, this sytem would make it as simple and easy as possible to do so. Because all events would already be equalized by the points system, it would be as easy as multiplying the points awarded in a given race by a simple weighting factor, like 1.2 (or, to put it another way, setting the value of Pwin to something other than 100). The question of whether and why to weight some races more than others could then be discussed strictly on the merits, without the clutter of logistical and mathematical complications.

With all that said, I still don't think going by time would be the best solution. I think the points system (regardless of how it is done) is better.

I agree — a strict cumulative-time system would not be workable for our purposes, and a points system would definitely be better. In particular, a calibrated points system based on times from each race appears to me to be the best and most elegant solution, for the reasons I've described.

[tdellaringa]

With that more detailed explanation of the proposed point system, I think I'm really digging the idea. It seems to be flexible enough for the challenges presented. And you make a good point about being able to weight a race if you want to. So maybe the regional finals and the championship race have heavier weights - like a race like Duluth for example.

[Robert]

Thanks everyone for continuing to post your ideas for the series.

As I've mentioned before, one thing we are trying to do — in addition to creating a point system for pros — is to come up with a ranking system for non-pros. We are considering a system based on NASTAR, the National Standard Race for skiers.

In NASTAR, a "par time" is established for each race (course). The par time is considered the best possible time. Then skiers are given a handicap based on how they do in comparison to the par time.
For an explanation of NASTAR, read this:
http://www.nastar.com/index.jsp?pagename=rules

I'm looking for some suggestions on how to make this work in the skate world. ... Any thoughts?

- Robert

[tdellaringa]

To me that's a really cool idea. It would be a way to gauge myself against other skaters, which is very hard to do right now (other than people I skate with). Don't really have any ideas right now on it, but just wanted to say good idea.

[jmac]

Here's a detailed explanation of how times would be converted to points in the system I'm proposing.

The winner of a race gets a predetermined maximum number of points. Every other competitor who finishes that race would be awarded a lesser number of points, based on the percentage by which their finish time exceeds the winner's time. This would be calculated using a simple exponential function so that every additional X% in time difference behind the winner would earn half as many points. (I call this X the "half-points threshold".) After doing a lot of numerical analysis, I'm hovering at 15% as being the best choice for X, with the range of 10%–20% seeming plausible.

For example, let's say the winner of a marathon finishes in exactly 75 minutes (1:15:00) and is awarded 1000 points. [Using 1000 allows more precision than 100, while avoiding decimal points.] And assume we're using a 15% half-points threshold. Then someone finishing 11.25 minutes later (1:26:15) would be 15% off the winner, and would earn 500 points. Someone finishing another 11.25 minutes later (1:37:30) would be 30% off the winner, and would earn 250 points. And so forth.

Here's a table showing how this would play out for various finish times. (There are three possible point equivalents given for each time; these show the result of using a half-points threshold of 10%, 15%, or 20%.)

1:15:00 = winning time = 1000 : 1000 : 1000 pts
1:15:01 = 0.02% back = 998 : 998 : 999 pts
1:15:02 = 0.04% back = 996 : 997 : 998 pts
1:15:05 = 0.11% back = 992 : 994 : 996 pts
1:15:10 = 0.22% back = 984 : 989 : 992 pts
1:15:15 = 0.33% back = 977 : 984 : 988 pts
1:15:30 = 0.67% back = 954 : 969 : 977 pts
1:15:45 = 1.00% back = 933 : 954 : 965 pts
1:16:00 = 1.33% back = 911 : 940 : 954 pts
1:16:30 = 2.00% back = 870 : 911 : 933 pts
1:17:00 = 2.67% back = 831 : 884 : 911 pts
1:18:00 = 4.00% back = 757 : 831 : 870 pts
1:19:00 = 5.33% back = 690 : 781 : 831 pts
1:20:00 = 6.67% back = 629 : 734 : 793 pts
1:22:00 = 9.33% back = 523 : 649 : 723 pts
1:24:00 = 12.0% back = 435 : 574 : 659 pts
1:26:00 = 14.7% back = 361 : 507 : 601 pts
1:28:00 = 17.3% back = 300 : 448 : 548 pts
1:30:00 = 20.0% back = 250 : 396 : 500 pts
1:35:00 = 26.7% back = 157 : 291 : 396 pts
1:40:00 = 33.3% back = 099 : 214 : 314 pts
1:45:00 = 40.0% back = 062 : 157 : 250 pts
1:50:00 = 46.7% back = 039 : 115 : 198 pts
1:55:00 = 53.3% back = 024 : 085 : 157 pts
2:00:00 = 60.0% back = 015 : 062 : 125 pts
2:10:00 = 73.3% back = 006 : 033 : 078 pts
2:20:00 = 86.7% back = 002 : 018 : 049 pts
2:30:00 = 100% back = 001 : 009 : 031 pts

And here's a chart that shows the same information, from the winning time to +50% back. The 10% curve is drawn in orange, the 15% curve in red, and the 20% curve in green. In addition, each curve has a straight dashed blue line intersecting it at the half-points threshold, that shows what would happen if points were deducted at a uniform linear rate instead of on an exponential curve.



Please study the table and chart for a bit, try to get a feel for how the calculations work, and see if you can judge which choice of half-points threshold (10%, 15%, 20%) you think would be the most appropriate.

(And yes, you can now tell me to turn off the dang computer for a while and go skate!)

[jmac]

[O]ne thing we are trying to do — in addition to creating a point system for pros — is to come up with a ranking system for non-pros. We are considering a system based on NASTAR, the National Standard Race for skiers.

In NASTAR, a "par time" is established for each race (course). The par time is considered the best possible time. Then skiers are given a handicap based on how they do in comparison to the par time.

More specifically, in NASTAR a benchmark time is established for each course every day, so that varying course and weather conditions are accounted for.

This is quite similar to the calibrated times-to-points system I've been discussing. In essence, the finish time of the winner of a race becomes the benchmark time for that course on that day, and everyone else's performance is compared to it. That makes individual performances more-or-less comparable across different races and conditions.

Another great thing about the calibrated times-to-points system is that it's equally applicable to Pros and non-Pros... just like NASTAR.

I should mention that Robert told me about NASTAR a couple of weeks ago, and I researched it, and its key feature of neutralizing the differences among courses and conditions through a system of calibration was a major influence in my design of the system I'm proposing for us now. So, thanks to Robert for bringing NASTAR to my attention, as I had never heard of it before he mentioned it!

[maroonloon]

I don't really know how much emphasis should be put on time. Unless as Maj said, you are doing a time trial, time is not really that relevant. Alot of times, you go as fast as the crowd you are with. The easiest half marathon I ever did was a bit over 35 minutes in Baxter, MN.

I sat at the back sucking the wheels of Randy Plett, Mike Anderson, and Jeff Terwilliger and hung on for the ride. Should this hold more importance than the race where serious pace setters don't show up and I skate 40 minutes instead?

I always look more at where I finished compared to people I know more so than looking at my watch at the end because of the cat and mouse.

To me, the finish time is pretty irrelevant because of drafting but thats just me.

I absolutely love the concept and don't want to be negative. I hope you brainiacks figure it out. I also would really like to see somewhere that there is a finale of some type.

[Jim White]

The present series, the Minn. Inline Grand Prix, system would rank non-pros on a percentile basis (winner gets 1000, the person half way down the finisher list gets 500, the last place person gets 1). This substantially eliminates absolute times, as Maroonloon suggested be done. It also substantially eliminates effects of course difficulty, including rain and headwind.

Tdellaringa suggested that comparison against skaters he hasn't raced is important; this system allows that. If I finish in the 60th percentile (60% of skaters are faster than me, and 40% slower), and someone else finishes in the 50th percentile, that gives me an idea that I need to train harder if I want to equal him. As always, comparing absolute times is also useful, as is looking at the placements (in the race I didn't skate) of skaters against whom I have competed.

(For pros, the point system which Robert posted earlier continues to seem to show what is important to those skaters; it gives much more weight to a first place finish, as is generally done at that level.)

[jmac]

The easiest half marathon I ever did was a bit over 35 minutes in Baxter, MN. I sat at the back sucking the wheels of Randy Plett, Mike Anderson, and Jeff Terwilliger and hung on for the ride. Should this hold more importance than the race where serious pace setters don't show up and I skate 40 minutes instead?

No, it shouldn't. That's why absolute times won't work for us — and it demonstrates why calibrated relative times will.

The present series, the Minn. Inline Grand Prix, system would rank non-pros on a percentile basis (winner gets 1000, the person half way down the finisher list gets 500, the last place person gets 1). [...] Tdellaringa suggested that comparison against skaters he hasn't raced is important; this system allows that. If I finish in the 60th percentile (60% of skaters are faster than me, and 40% slower), and someone else finishes in the 50th percentile, that gives me an idea that I need to train harder if I want to equal him.

Actually, a placement-based system does not allow comparisons any more meaningful than simply knowing whether someone finished ahead of you or behind you in a particular race that you both attended. The number of people who finished between you is irrelevant, because it depends on the number of competitors who show up, and on the distribution of skill levels among those competitors — neither of which is consistent from race to race, and neither of which you have any control over. In a placement-based system, your own result is more dependent on what other people do than on what you do. But in a calibrated-time system, none of that matters; you are evaluated only on the basis of your own time relative proportionally to the winner, just like everyone else. It's the only way to make results truly reflect your own performance. And that, in turn, is the only way you can make meaningful comparisons of your performance with those of other skaters — especially those at other races, where different people show up.