AccuPredict Driver Projection Results
NASCAR Cup Series
This is a historical results' page analyzing Accupredict's accuracy in driver projections. A novel idea, right? Checking how accurate an experts projections are.
Kyle Larson
288 Cup Series races with projected and actual finish position for Kyle Larson are available.
Averaged statistics of all race projections Accupredict has made for Kyle Larson.
Tracks
| Track | Races | AVG FIN | AVG RNK | RNK +/- | AVG PRJ | PRJ +/- |
|---|---|---|---|---|---|---|
| Atlanta Motor Speedway | 10 | 14.60 | 7.60 | 7.00 | 12.85 | 1.75 |
| Auto Club Speedway | 8 | 13.00 | 11.75 | 1.25 | 14.68 | -1.68 |
| Bristol Motor Speedway | 13 | 13.38 | 8.62 | 4.77 | 13.15 | 0.23 |
| Bristol Motor Speedway (Dirt) | 2 | 16.50 | 3.50 | 13.00 | 10.36 | 6.14 |
| Charlotte Motor Speedway | 13 | 16.77 | 15.15 | 1.62 | 16.16 | 0.61 |
| Limited Accupredict View |
For full access to Accupredict and other subscriber-only content, please sign up for a Crew Chief or Owner subscription.
Already a subscriber? Please log in.
Not sure which subscription you have or when it expires? Need to renew? Look in the top navigation under ACCOUNT > Subscription Details.
Track Types
| Type | Races | Description | AVG FIN | AVG RNK | RNK +/- | AVG PRJ | PRJ +/- |
|---|---|---|---|---|---|---|---|
| Flat | 83 | The 'Flat' NASCAR tracks are Indianapolis, Martinsville, New Hampshire, Phoenix, Pocono and Richmond. We originally grouped Indianapolis and Pocono as 'Shallow' Tracks. But, the pair provided little statistical data at only three races per year. Therefore, we grouped them into the 'Flat' Tracks category. As a result, we found that results for eight-race averages show improvement in all Flat Track correlations. Pocono in particular is better correlated with this new grouping of Flat Tracks. | 13.48 | 9.57 | 3.92 | 13.32 | 0.17 |
| Large Oval | 88 | The 'Large Oval' NASCAR tracks are Atlanta, California, Charlotte, Chicago, Kansas, Kentucky, Michigan and Texas. This grouping of tracks show a fairly tight clustering of the curves and improved correlations for each of the member tracks with average finishes over other groupings these tracks get placed in. | 14.49 | 11.18 | 3.31 | 14.32 | 0.17 |
| Limited Accupredict View |
For full access to Accupredict and other subscriber-only content, please sign up for a Crew Chief or Owner subscription.
Already a subscriber? Please log in.
Not sure which subscription you have or when it expires? Need to renew? Look in the top navigation under ACCOUNT > Subscription Details.
Seasons
| Season | Races | AVG FIN | AVG RNK | RNK +/- | AVG PRJ | PRJ +/- |
|---|---|---|---|---|---|---|
| 2022 | 26 | 13.96 | 4.08 | 9.88 | 11.06 | 2.90 |
| 2021 | 37 | 8.89 | 4.43 | 4.46 | 10.14 | -1.25 |
| Limited Accupredict View |
For full access to Accupredict and other subscriber-only content, please sign up for a Crew Chief or Owner subscription.
Already a subscriber? Please log in.
Not sure which subscription you have or when it expires? Need to renew? Look in the top navigation under ACCOUNT > Subscription Details.
Dabble in Draftkings or Fan Duel?
- My OWNER Subscription has these DFS tools:
- Dominator Points Cheatsheet.
- Salary Cap Calculator.
- Season Salary Trends.
- Points by Track.
- Points by Start Position.
- Season Points Trend.
- Points by Race.
- Season Boxscore.
Select a Result
View Accupredict's Projection History Performance by a helpful variety of groupings to better identify the tracks and drivers that it projects the best.
Accupredict Pages
Accupredict is my exclusive NASCAR driver finish position projection tool for Crew Chief and Owner subscribers. The projection algorithm is the work of research scientist Cliff DeJong using traditional and loop data statistics. I've had it online since the 2012 Daytona 500.
NASCAR CUP SERIES
NASCAR XFINITY SERIES
ACCUPREDICT ARTICLES by Cliff DeJong
The driver, track and race links in the tables link to more historical Accupredict data.
How I Calculate Difference (+/-)
Differences are calculated as: Actual Finish minus Projection
Which 'direction' the difference 'moves' is important. A difference that is negative means the driver finished better than projected. A difference that is positive means he finished worse than projected. A zero means Accupredict got the drivers finish position spot on.
Driver was projected to finish 12th. If the difference is calculated using the Absolute Value of the number (the number of positions 'off' without classifying the number positive or negative) then you're getting a range of eight positions. Four of which are worse than the projected finish.
In my experience, differences between +2 to -4 consistently, are the 'sweet spot' for favorable use of a driver based on Accupredicts projection.
When I show the Absolute Value (distance from zero that a number is on the number line, without considering direction. Absolute Value is always a positive number.) value of a number in addition to Real value.
The MEDIAN and the MEAN numbers in the sidebar both use the Absoulte Value of the number difference between Rank/Predicted Finish and Actual Finish. Otherwise, the positive and negative differences cancel each other out and calculate to 0. I show the STANDARD DEVIATION and the FREQUENCY TABLE using both the absolute and real numbers and note there which is being shown.
I've watched over the years, Accupredict is more conservative in projecting a drivers finish and thus has more negative differences. That is, projecting a driver to finish 18th - they finish 14th - resulting in a difference of (ACTUAL - RANK) 14 - 18 = -4. Whereas, the other way around is a +4. To me, knowing that I missed the projection of a driver by 4 spots...and they were 4 spots better, is a lot better than being 4 spots worse. One way I get more points, the other way I get fewer.
Also, the large positive numbers are the result of a 'top-tier' driver ranked in the top-10 being the victim of a wreck, mechanical failure or unfortunate timed caution.
USING ABSOLUTE VALUES
9.67 Mean (Average) number of positions 'off' that the projections were.
8.00 Median (the middle number when all are sorted, descending) number of 'positions off' that the projections were. Median helps account for outliers in the sequence that might skew the Mean with such a small population of data.
8.85 Standard Deviation measures the dispersion of each driver 'positions off' in relationship to the Mean (Average).