Data Import

Race results are imported by a group of specially trained monkeys who browse the web and racing publications to find the results. The rest of the numbers are crunched by a much more trained group of monkeys in the basement of my house. Please forgive me and let me know if something is incorrect, there is only so much one man can do to train a group of monkeys.

Power Rankings

Like many different forms of competition here at Race Stat Central use a modified version of the ELO rating system for generating our rankings and ratings. We decided on using the ELO system because not only does it produce a rating for the driver's finishing position it adjusts the rating based on the finishing positions of other drivers in the race. You can read in detail elsewhere on the internet how it works, but I will summarize below.

At the beginning of the season a driver will carry over a percentage of the previous year's rating while the remainder is averaged with the base of 1500. All new drivers will start at a rating of 1500.

Each time a driver starts a feature their rating will either increase or decrease based on their finishing position and the ratings of the other drivers. That rating will be their starting rating going into the next race that they start.

The ratings are calculated once per week and updated on Friday morning.

Weather Forecast

The weather forecast is pulled from the National Weather Service specifically for the area around the track. Forecasts are updated every three hours.

Series Predictions

The series prediction and methodology was developed during the 2021 season and is expected to morph as we understand how our predictions align with actual values.

The forecasts are based on the Power Rankings, a revised system that took place during the 2020 season. We have updated and incorporated our new rating system with data back to 2002.

How we forecast a season. Racing, and Dirt Track Racing specifically is much more difficult to this sort of thing with compared to stick and ball sports. Prediction methods for stick and ball sports have the benefit of a fixed schedule with fixed competitors. In auto racing different competitors can come and go from event to event. To give a similar relative strength of field, I use the top drivers in each series and simulate the remaining schedule with those divers.

Each time an update is made we simulate the remaining races in a season 10,000 times to give a wide range of possibilities that you see on the page. We also do the simulations "hot", so in a given season if a simulated driver starts to perform better their rating for the next simulated race will be better and the same for a string of poor performances.

In our standings we do our best to simulate the season for each series correctly. In the case of the Short Track Super Series, we follow their drop race and attendance bonus system. With the BRP Modified tour we award heat points based on randomly generated heats, which are also simulated to award points, and also use their attendance bonus system.

As we see actual data against our simulations, we will make tweaks, but this does provide a nice visual of each series potential final standings with all of the complexity of drop races and other bonuses. Please note this assumes all of the drivers shown for each series will show up to each race and participate. Conflicting dates and drivers’ appearances are impossible to simulate.