GPRM Scheduling - Distribution

[dna1990]

We have some Groups (Ranks) that are much larger than the others. The biggest was 46 this year.

I am sure it has been answered for both GPRM and the various scheduling theories....but does GPRM try to schedule a given racer over the duration of that 46-car group?

Meaning that in a perfect world, it would race somewhere in the first 10, second 10, third 10, and last 10...?

The question comes up about using timed racing and how stable a track and timer will be throughout the day. In a small Rank, the time is short enough between first and last race, where we feel comfortable that nothing major has changed. But for 46 (about 45 minutes for us) is a long time for the timer not to heat up, the IR to do somethign odd, the spring on the start gate to wear differently, etc.

Racing once per lane helps combat lane bias. Is there something in the scheduling to combat general timing bias over a longer period of heats?


We thought about a Finals By Group race, to bring back the top four within each (so they all race once more while conditions are more equal). But when nothing has changed (which is the usual case)...the cars finish in the exact same order as you would predict and it doesn't make for a good feeling race. And our turnaround time is very slow as the cars are walked back to the starting gate.


Or perhaps I ask all this, not truly understanding what goes into a PerfectN chart, and it is already well taken care of...?

[dna1990]

I have gone back and read http://grandprix-software-central.com/g ... format.php, so I see that maximize opponents is there.

But that still leaves me wondering if you could run all four lanes against all new opponents and still race only in the first half of the overall group/rank boundary (given a large enough population).

[davem]

Don't know the algorithm underneath the code.
But my practical experience over 6 races is that it does a pretty good job of spreading the racers' heats out evenly across the race.

One or two racers will "feel" like they have gone a long time between heats.

[gpraceman]

If you are that concerned with track and timer changes over teh course of the race, then you should use points scoring. Of course, then you are much more likely to get ties, unless you can do more racing to reduce those chances.

[dna1990]

No, our StdDev says all seems good and we like Times much better.

I was just looking for a fact about distribution, if I need to reply to someone that wants to make an arguement for going back to points.

And the staging crew of course loves there to be several heats between when a car runs next. As to not wait on a entry from the return pile. So an added benefit there of spreading them out.


Is is not practical at all, but from the math/logic type in me would like to run the heats and then take a quick study of the StdDevs, grouped in quartiles of the clock time it took to run that group. If there is a spread > x, then run a Final Round, otherwise declare winners as they stand. But crowds don't like "if" statements. And running a finals only sometimes, then causes suspicion in why it was needed, etc. Like I said, not practical.

But I don't think it is worth the boring re-run on the top N cars either.

Thanks all.