In the 2010 BPR Correlation Analysis post, I alluded to a forthcoming prediction model based off the BPR rating. As demonstrated by the 2010 correlation analysis, the BPR Power rating and its tracking of three-race and full-season performance trends is itself a highly effective predictor of future results. However, there is one notable deficiency when applying the POWER rating to predict future results: individual track-type differences. Here’s what I wrote in the 2010 correlation post:
With the foregoing in mind, I intend to adapt the existing BPR formula to create a prediction model for the 2011 Formula 1 season. At this stage, the model will incorporate the POWER rating as well as a separate formula based on track-type performances which positively correlate to other rounds. Including the second calculation is intended to address the simple fact that there are completely different tracks on the Formula 1 calendar which tend to produce different results based on individual car and driver characteristics. Although it’s relatively obvious that the results of the Monaco Grand Prix are unlikely to reflect the results of say, the Italian Grand Prix at Monza, result relationships amongst other tracks aren’t as obvious. Therefore, I decided to run a correlation of all the BPR scores for each round on the 2010 Formula 1 calendar to find which events’ results correlate. Remember that the BPR incorporates a lot of data for all 24 participating drivers at each round and therefore the correlation results should be fairly dependable.
Following that post, I was able to run similar correlation analysis for the 2007, 2008, and 2009 seasons as well. Armed with four seasons worth of correlation data, I have developed a prediction model utilizing a weighted mean and the historical correlation coefficients between specific tracks on the Formula 1 calendar. For more on what a weighted mean is, see this explanation; in short, a weighted mean averages a set of numbers based on associated figures which determine the ‘weight’ each number should have in the final average. Therefore, the BPR prediction model, unlike the BPR POWER rating, accounts for the demonstrated differences between each track on the F1 calendar. Unlike the POWER rating, the prediction model will not incorporate a Reliability Rating because it is impossible to predict whether a driver will finish the upcoming race. For this reason, the prediction model only predicts the likely BPR score of a driver if he finishes the race un-delayed. As the season develops it is possible that changes will be made to the model, including incorporating the Reliability Rating or three-race performance trend featured in the BPR POWER rating. However, if any changes are made I will be sure to inform you of them.
The BPR Predictor table looks much like the regularly-posted BPR table with just a few exceptions. The PREDICT column displays the predicted BPR score of each driver for the upcoming round, and the table is sorted based on the PREDICT outputs. The shades of blue denoting levels of performance are identical to those in the BPR table. To the immediate right of the PREDICT column are the individual round BPR scores for each entry. You will notice that the label for each round’s column has a decimal number below it; this decimal is the correlation coefficient for that particular round’s track and the upcoming round as demonstrated over previous seasons.
2011 FIA Formula 1 World Championship – BPR Predictor Rd. 3: Chinese Grand Prix
posted by Trey Blincoe