rcnewb2004

Hmm... wondering if i got this right...

basically the Error is equal (average lap time) - (time of fastest lap)?

Coming from a highschool stats background... basing this on a crap assumption that lap times are approximately normally distributed, (i guess if we have many lap times, and keeping all else consistent (not really possible) it "should" approach a normal distribution. If someone has a real stats background... do correct me...), then the Error is basically half the "width" of the standard distribution curve?

The goal then is to decrease the error (making the "width" of the peak smaller)?

Could I re-understand this as measuring the standard deviation in lap times as a better indication of "better driver"?

Sorry for the questions... I get lost very easily...

miller tyme

iTrader: (5)

... wondering if i got this right.. I think the standard deviation bell curve and six sigma statistical analysis approach is based on the variation around the mean.

This would work but Bob's variation is based on deviation from the single fastest lap (since that is the ultimate goal)

kn7671

iTrader: (16)

Here are the results from my Stock 13.5 - 2nd place finish yesterday.

http://www.jlapracemanager.com/PC/Mi...ck%20Sedan.HTM

Qual 1:
4 5 Kelly 26 00:05:02.661 00:10.891 23 11.640
Qual 2:
4 5 Kelly 26 00:05:03.287 00:11.044 10 11.664
Qual 3:
5 5 Kelly 25 00:05:03.467 00:10.829 10 12.138
A-Main:
2 5 Kelly 26 00:05:02.723 00:10.992 12 11.643

I think I'm getting to be a pretty consistent driver, but still not consistent enough to not rub flapper or board from time to time. To finish the day with 3-races with almost identical fast laps, avg. laptimes, and total laptime was amazing to me. Most consistent to date for me. Now, how to I go faster and still get more consistent? Practice, Practice, Practice, which I almost never do.

For comparison sake, here is what the Expert Stock 13.5 guys put down on my track yesterday:

A-Main
1 1 Mike Dumas 30 00:05:09.350 00:09.738 2 10.311 Done!
2 6 Steve A Main Marker 28 00:05:01.963 00:10.255 2 10.784 Done!
3 3 Tye Love 28 00:05:02.434 00:10.211 4 10.801 Done!
Qual-3
1 7 Mike Dumas 31 00:05:08.922 00:09.761 2 09.965 Done! <---- This is sick - .204
2 3 Chris Metheny 28 00:05:04.338 00:10.619 12 10.869 Done! <---- This is sick - .250
3 10 Steve A Main Marker 28 00:05:05.432 00:10.350 9 10.908 Done!
Qual-2
1 7 Mike Dumas 30 00:05:01.697 00:09.741 10 10.056 Done!
2 1 Tye Love 29 00:05:05.371 00:10.071 4 10.530 Done!
3 6 Aaron Breuer 28 00:05:00.694 00:10.518 16 10.739 Done!
Qual-1
1 4 Rocco Margiotta 29 00:05:04.729 00:10.202 13 10.507 Done!
2 1 Tye Love 29 00:05:07.363 00:10.180 7 10.598 Done!
3 6 Aaron Breuer 28 00:05:06.707 00:10.310 2 10.953 Done!

trilerian

iTrader: (51)

Without even plotting, I can hazard a guess that lap times will not be normally distributed, even if you drop off the outliers to try and make it normally distributed you would not be able to use the Stormer error as a standard deviation. Alright, the Stormer error is derived from the fastest lap, in a normally distributed bell curve that fastest lap will be 3 standard deviations from the mean. If you are using 1 standard deviation as your indicator you will be 1/3rd the value of the Stormer error. Ultimately however, you are trying to narrow the width of the curve, but that doesn't change the fact that from 1 standard deviation from the mean in both directions, 68% of your laps will fall, from 2, 95% and 3, 99.7%.

Hey Bob, had a bad night Friday, I couldn't stop thinking about trying to turn my best laps for the thread, needless to say disaster struck. So here is my best qualifier:

Laps Time Fast
36 5:3.759 8.028

Stormer error: .291. Error from all the laps: .410

I still managed to qualify second and place second in the A.

TQ run:

Laps Time Fast
37 5:06.182 7.933

Stormer error: .172 Error from all the laps: .342

Bob-Stormer

iTrader: (8)

You say your 3 races and a nearly identical result. What changes did you make to the car, if any, on those 3 runs? How familiar was the track layout? new, old, etc.

I looked at that link, the driver in front of you got you by a lap. Lap times were pretty close for a club scenario. But even though you were second, you can see where he got you. Just more consistent and able to turn more accurate laps. Not all of that is the driver, that's why I like to call it the "package" or putting together your puzzle. An easy to drive car, is easy to drive and you can turn faster more consistent laps with it.

Bob-Stormer

iTrader: (8)

Sounds like you had fun though.

The real advantage to looking at the error is in comparing your result to those around you. To help maximize the package.

Bob-Stormer

iTrader: (8)

The goal is to decrease your error. But the goal is to do it with the fastest possible package. The best use is to compare yourself with drivers around you in any main or qualifier. I find the fastest qualifier is a good spot to look. Mains can get rough.

A lot of this stemmed from people griping about motors... "If I had this guys motor, I'd whip all you guys"...

You could conceivably, and for sake of debate, put your error on his fast lap and see how the car would have been with you at the wheel.

Lot of assumptions, but it works out most of the time. Also, a driver that is a serious wheel would very likely pull a lower error number, driving your car. I hate when that happens to me, but I've seen it happen. What that means is that there is a piece of the puzzle that can still be improved, the driver.

It's a package. It's about maximizing your package and it's potential, and you won't know if it's good enough until you can compare it either to a different setup of your own, or a different driver.

What's faster, a package you can drive to a 10.4 with an error of .560 or a package you can drive to a 10.5 with a .430?

The 10.5 with the .430 is the car you want to take into the main. Easier to drive, and faster (overall).

Setup a car that is loose, and one that is pushy. Run some good laps, watch the error.

rcnewb2004

Thanks to those who are taking my reply seriously. Its good to know when I at least try to understand something here, that some of you do not make fun of me.

trilerian: (just curious) why do you "know" it won't be normally distributed? (sorry, i am not the brightest... not even close). Thank you for correcting me, yes it should be 3 deviations away. Also the Stormer Error correction is much easier to apply at the track than to do some funky standard deviation.

Bob-stormer: Thank you for your reply. It is an excellent way to measure consistency and shows that consistency actually is the difference between a "better" driver and a "worse" driver. I will work on improving my consistency.

~McSmooth~

NOW I remember! I'll have to ask and see if that one still exists!

miller tyme

iTrader: (5)

A few of us were discussing this at the track today so we looked at resulting lap times pretty carefully.

Bob you met Phil Zimmerman at Cleveland this year,
he ran 43 laps with a fast lap of 10.7 and a consistency of .195,
he finished second to a fast lap of 10.7 and a consistency of.190
and I rolled in 3rd also with 43 and a fast lap of 10.7 but only a .245 consistency

and taking my consistency out to 35 laps yielded .295. Studying the #'s if my 12 laps at 11.1 had been 11.0 (excluding a bad marshaling incident) all 3 of us would have finished ~ .8 seconds apart.

Conclusion is it may be easier for me to concentrate more on those dozen laps and not sweat trying to pull a 10.6 second lap.

jiml

iTrader: (3)

You know, now that I think of it, I don't remember anyone ever saying that.

What I have heard is how can I compete with these peope if I can't even match their power?

Bob-Stormer

iTrader: (8)

Phil kept me outta the main! Actually, I kept me outta the main, Phil drove better.

Good close racing there. Once you're to that point, it gets even better. Unless one of you changes something, you will not go much faster. Sure, you'll eventually run some better laps as track familiarity increases, but your error tends to follow you. Some guys dial in quick. I'm not that guy. I'll sometimes struggle the whole weekend and never find something. It happens.

Assuming better driving, for you to knock these dudes off, you need to not only step up your game, you gotta find something else in your chassis. If you assume that your error follows you, you need to up the anti everywhere else you can, so that even with your error, you can win. That's tough. And usually good drivers go faster anyway, so it's hard.

BUT, the more you watch the numbers, the more you'll see guys with super fast cars just barely win, to much error. Or guys with very slow cars knock off a bunch of good drivers, not much error.

Just gotta maximize the package.

Bob-Stormer

iTrader: (8)

Well when you see it next time, whip some math on 'em and let 'em know the real reason they're getting beat... "Dude, we've been looking at the numbers, you suck..."

SWTour

iTrader: (9)

Bob,

this is a great thread...

It's kinda funny, but when I got my son racing about 5 years ago (He's 15 now) I told him to work on being consistant and learn to drive with patience.

After every run - LOOK at the score sheet for 5 things.

1) Your FASTEST LAP
2) Your SLOWEST LAP
3) The TQ's FASTEST LAP
4) The TQ's SLOWEST LAP
5) Your Overall Consistency

Study those times - learn where you've made mistakes...fix those mistakes BEFORE you ask ME to make your car faster.

Work on DRIVING, HANDLING and CONSISTENCY...and slowly we'll keep adding power as needed.

I'm always suprised by people who only compare their FINISH time to the Winners FINISH time w/o comparing laps and consistency and taking into account all the errors in driving...and complain they had a SLOW car (When their lap times sucked)

SWTour

iTrader: (9)

...dang system LAG - DOUBLE POST!!!