View Single Post
      04-04-2022, 04:46 PM   #387
Max Well
Colonel
Max Well's Avatar
4151
Rep
2,262
Posts

Drives: '22 BG X3MC, '20 BSM X3MC
Join Date: Jun 2014
Location: Southeast USA

iTrader: (0)

I’ve finally completed the 'Comparative Analysis of Dragy vs Formal Slip Results at GNV Raceway'. I’m including a table showing amount of elevation change per distance measured depending on the slope for ease of reference, and also direct comparisons for 1/8 mile, 1/4 mile and slope & altitude. Disclaimer - I have no financial interest in any of this discussion, and I do not have a formal background in Statistics. I'm just a lay person trying to understand how accurate and precise Dragy is c/w the standard reference for 1/4 mile performance.

As has been suggested by many over the years, Dragy appears remarkably accurate and precise with both time and speed in the 1/8 and 1/4 mile:

-Dragy’s 1/8 ET is on average 0.014s +/- 0.021s slower than formal time and 0.17mph +/- 0.06mph faster in speed. The max value ranges are only 0.11s and 0.32mph, respectively.
-Dragy’s 1/4 ET is on average 0.013s +/- 0.020s slower than formal time and 0.21mph +/- 0.27mph faster in speed. The max value ranges are 0.11s and 1.23mph, respectively.

On the other hand, Dragy has difficulty with 'Precision' in altitude and slope. For this analysis I estimated a 0.00% Slope at GNV Raceway, but as I’ve been unable to find the actual value, it’s not possible to correctly identify Dragy's ‘Accuracy’ with this aspect. If one follows the Strip’s surface on GoogleEarth (which provides elevation data when hovering over a location) it would appear level. But as some papers have estimated GoogleEarth’s 'Accuracy' as only roughly 6-8 feet and that also depends on global location (which varies as some locations have better satellite coverage), one can’t use that as a standard reference vs Dragy. Multiple references can be found online for that topic by searching 'GoogleEarth, Elevation, Accuracy'.

Dragy’s wide variation in slope (even at a set location like a NHRA-sanctioned track on the same day and same lane) results in a large 'Range' of values and lower 'Precision' than it exhibits with it's ET/speed capabilities, and this is independent of it’s 'Accuracy'. In other words, if GNV’s slope was actually -0.1% instead of 0.00% as estimated for this analysis, the average difference for that value would improve, but the values for Range and Standard Deviations would remain exactly the same. From a practical standpoint, this lack of slope Precision doesn’t matter at a known level racing surface. But it does make it hard to trust at unknown locations where one is dependent on that very value for determination of whether the run is ‘valid’ or ‘invalid’.

As can be seen from the slope comparisons, Dragy has the following slope findings (note again that slope ‘Accuracy’ isn’t being discussed as GNV Raceway’s actual slope values for those three segments are N/A):

-In the 0-60mph, the Range of Slope values was 1.95% with a std dev of +/- 0.47%

-In the 1/8 mile, the Range of Slope values was 1.24% with a std dev of +/-0.26%

-In the 1/4 mile, the Range of Slope values was 0.87% with a std dev of +/-0.18%

These results shed some light on why many have reported being confused when getting ‘Invalid’ results with 0-60mph data, even though they were quite sure the short distance ‘had to have been level’. Importantly, though, it appears these slope precision issues lessen as a longer distance is measured. This would imply 1/4 mile data appears to be more resilient to the effects when c/w shorter-distance 0-60mph results.

In sum, it appears this small personal analysis would support Dragy's excellent Accuracy and Precision with 1/4 mile acceleration/performance data. It remains a great tool for serial testing and tuning, especially if used at a known level location (and same starting point for every run) where it's lower reliability slope values and the 'Valid'/'Invalid' descriptors become irrelevant.
Attached Images
    
Appreciate 0