There is no subjective adjustments made to the speed factors.
From the 2003 ProRally rule book pg 20, same rule as 2004.
Here is the formula:
A Stage Speed factor for each stage run is calculated.
Competitors stage speed factor = Fastest Time/competitors time
All of the stage speed factors for the event are added up dropping the fastest and the slowest and the rest are then averaged by the number of stages run by the competitor. This is now your event speed factor. At least 4 stages must be completed.
The event speed factor is then compared to the competitors previous 1/2 season history and the higher number is used. There is an aging component as well of 0.02 for each 1/2 season that is out of date. So your speed factor would drop 0.02 for every 1/2 season that you did not complete enough stages to get an event speed factor.
Regarding calculating the 0.02 speed factor reduction for every six months of inactivity:
Has this reduction been applied retroactively (back to the person's last rally) or only back as far as when Raly-America began calculating the speed factors (last year)?
OK, as I read it, the current speed rating is basically a straight average of your stage speed factors (throwing out the highest and lowest score.) Ultimately, it means that long stages are treated the same as short stages. And, the number reflects your performance against the competitors of the events you participate in, not the pool of competitors.
Now, we use speed factors to allow us to better position cars on the roads. The concept here is that speed factor gives you a comparative figure for each competitor. So if Joe has a speed factor of .85, he must be faster than Steve who has a speed factor of .83.
The problem with this system is that speed factors are not always comparative. We are using them as a tool to predict future (current event) speed related to others. The only way this system works with certainty is if the winner of each stage was constant, meaning that each competitor was compared to the same (or similar) driver.
So our first consideration is who you are running against. Now our second consideration may be the type of event you are running. For example, an event such as Sno*Drift will reward drivers with lower HP cars yet these comparison times would be useless at a faster event such as STPR.
Now, the current system also treats every stage the same. A 1 mile stage is the same as a 20 mile stage. The spread between the top and bottom drivers widens on longer stages yet remains small on short stages. It would be easy to weight stage speed factors (multiply by the miles and divide the sum by the total miles covered) if you want to keep cars more spread out over the longer stages where passing would be more of an issue.
For example, on a 2 stage rally we have the following for our competitor Joe (we're assuming the high and low scores have already been trashed here):
Stage 1, 5 miles, .8 SF
Stage 2, 20 miles, .9 SF
under the current system, Joe would have a SF of .85. Now, if we multiplied by the miles we would have ((.8 * 5) + (.9 * 20) / (20 + 5) = .88. Is that better? Beat me.
I'm not sure how to normalize times across competitors and frankly I'm not even sure that is possible, but I'm sure there is someone here who took more statistics than me who can come up with the appropriate formula taken to, let's say, a confidence interval of .8.
>>Regarding calculating the 0.02 speed factor reduction for
>every six months of inactivity:
John McArthur emailed me when I questioned my factor and said that the speed factors are updated based on the last ProRally attended (even if you are not running the PRO event). So, if you ran Rim, then skipped events in conjuction with Pro events, your speed factor could go down for "inactivity".
We have an additional issue here in CRS-world, where a lot of rallies are run with non-SCCA sanction. So, Joe or Josie Newbie Competitor could run 1 SCCA rally, get their factor, then run 4 or 5 "other" events. They presumably get significantly faster, yet end up being placed too low on a start order at their next SCCA event.
This issue will start to manifest itself in the east this year too, as a number of rallies will be run there with non-SCCA sanction.
>This issue will start to manifest itself in the east this
>year too, as a number of rallies will be run there with
Its already happened because there have been only one NEDiv SCCA ClubRally in the past year, many people have choosen to compete in Canada.
Look at Randy Zimmer's speed factor. He is no slouch yet his speed factor is slower than many production class drivers and drivers he'd beat with an arm tied behind his back.
Inactivity in SCCA events in not necessarily rally driving inactivity.
Speedfactor also seems to advantage ClubRally drivers who have ProRallies in their division over those whom do not.
Maybe this is a good time to bring up the feasability of a cross-national speed ranking system for both CARS/SCCA/NASA events? It would certainly address many of the issues mentioned here. I'm certainly for it. Can't see how it'd hurt, other than the *logistics* (er, egos) of a shared body between CA and US.
>>ps - Jake, thanks for stuffing off the road enough (on the
>>ranch) so we could get by without hitting you (and glad to
>>see you got out).
>Yeah, how did you get into that position at that spot anyway
Dear wise asses
You will not be poking such insensitive fun at Wounded Sisu Racing for much longer when you read in the press release forum about my lucritive sponsorship package with Toro portable Snowblowers for next year - complete with Toro babes and all. You wisecracking backmarkers who feel the need to poke fun at drivers who go in the bank on long straights better just hope I don't fill the road with snow from my port-o-Toro when I'm unstucking the Sentra at Sno Drift 2005.
As far as that spot, you would laugh if I told you, real hard.
Too bad you guys missed our off on Saturday, though. It was collasal. It must have looked like a car hitting a 1000lb bag of white flour. The Sentra was running strong, hungrily chasing after the smell of smoked whitefish trailing from the scent-filled car of Dave Cizmas. We were trying to win the whole stinkin' rally then and there and were in the process of setting a land speed record on Meaford ice over a R6 120 L5- when POOOF!!!, the road...like... mysteriously turned into a snowbank. What a wierd way to plow roads. AND everyone else made it though OK.
Anyway, I'm glad for Rally America comparing my speedfactor from Sno D to my sf from the last 6 months.
>-Rally America? Well they've got a relationship with SCCA,
>so are kind of beholden to them.
I think "Beholden" is a poor choice of words. Rally-America works with the SCCA because they sanction the ProRally Championship Series.
The earlier suggestion that there be a combined ranking system across series is a good one, but I fear unlikely. Everyone needs to keep in mind that the SpeedRanking only establishes the PRELIMINARY start order. The Stewards are free to make sensible adjustments as they see fit.
Over all I really like speed factors, it is such a better system than anything else used in seeding. Thank you whoever came up with the system - it really is one of the best improvements of the last few years.
There may be some small improvemnts to be made to the math. For example I do think the aging factor is missguided, I now have a better speed factor than Eric Eaton and he might get a minute out of me in one stage if he ever got his nice Subie out to play. This is a guy who has won stages at the national level but not rallyed since 2000.
The distance issue above is not an issue if I am 10 seconds a mile slower than Higgins I get the same speed factor result on a 6 mile stage as I do on a 12 mile stage in both cases .857 with Higgins = 1
The current system does not need correcting for distance unless I did the math wrong - always possible.
>>-Rally America? Well they've got a relationship with SCCA,
>>so are kind of beholden to them.
>I think "Beholden" is a poor choice of words. Rally-America
>works with the SCCA because they sanction the ProRally
>The earlier suggestion that there be a combined ranking
>system across series is a good one, but I fear unlikely.
>Everyone needs to keep in mind that the SpeedRanking only
>establishes the PRELIMINARY start order. The Stewards are
>free to make sensible adjustments as they see fit.
Well why don't we start small then. A Co-2 Clubrally with a ProRally competitor with an established speed factor, it should be reasonable for other competitors to get a speed factor based on a percentage of the "Known" ProRally speedfactor.
Say someone is A .89 and wins the co-2.
Say I run .89 against that competitor in a co-2.
My speedfactor for that event should be .89 x .89 = .7921
Co-2 is enough stages/miles to get a good comparison.
And don't even go to the "the big boyz don't drive as hard at Clubrally's". Go to NW region, the 4wd boys at the front of our fields take no prisoners. If a talented driver can still compete (and win) against our NW guys with NO STINKING DRIVING LIGHTS that shows a certain level of comittment to me.
I just did the math using 3 cars vs David Higgins at Maine Forest Rally and the stage distance had no effect on speed factor. Basicaly everyone scored about the same on every stage vs Higgins as they did over all. In fact all the sample drivers did better vs Higgins on the longest stage than they did overall in the rally.
I used Brad Hawkins, Brian Scott and Celsus Donelly. Ok I added a GpN car and the same thing holds True.
At Maine Forrest at least there was no coralation on the distane vs speed compaired to Higgins.
Some drivers get tired, others get into a rythem. Some short stages have lots of fast, slow fast slow. Some long Stages have lots of open road where driver skill has little to do but hold the pedal down. Distance does not seem to play a roll. We can do more math however to make sure this is ture in all cases.