AutoScout is a scouting application designed for FIRST Tech Challenge tournaments. AutoScout Utilizes Machine Learning and statistically driven calculations to predict the outcome of any inputted meet. AutoScout is typically around 90% accurate when run before a meet. AutoScout has its own statistic, Calculated Average Points (CAP), and on average, using CAP is 5-10% more accurate than running with OPR alone.
AutoScout Utilizes Statistics such as OPR, TBP1, and TBP2 to calculate CAP. CAP allows schedule prediction to be more accurate, allowing your team to scout more efficiently and optimize your time spent scouting. AutoScout allows your team to estimate which teams and which matches to scout. Instead of using the classic approach of "watch it all," you can now sit back and watch AutoScout do the work, making it easier to focus on winning your tournament and awards.
The statistic CAP(Calculated Average Points), used in AutoScout, is the result of years of development and refinement. Unlike traditional models, such as OPR (Offensive Power Rating), CAP consistently delivers more accurate schedule predictions—typically 5–10% more accurate on average. Additionally, the predicted match scores closely reflect real-world outcomes.
For example, during the 2024–2025 FTC season, a match with a CAP-predicted score of 168.8 ended with an actual score of 167.8. In contrast, OPR predicted a significantly higher score and incorrectly favored the winning team, largely due to inflated values from prior events. This discrepancy highlights a key flaw in OPR: it can be overly influenced by historical data from matches involving different team compositions. For instance, a team with a near-zero score can still inflate your OPR if they were paired with high-performing alliances in the past, leading to misleading metrics.
While this may benefit your individual OPR, it creates inaccurate evaluations of other teams—potentially leading to poor scouting decisions. AutoScout addresses this issue by providing data that more accurately reflects current team performance, helping teams see past statistical noise and make better-informed choices.
Some features of AutoScout are still in development and may not function properly. Some examples include Loading speed calculation, Strength of Schedule, Schedule Performance, etc. As seen in the image below, the accuracy is 70%. It is only so low because this is a meet from early in the season, and by the time it was run through inaccurate for this image, the data was very inaccurate.
AutoScout's Alliance Selection Optimization tool, ALSO, is out. ALSO helps minimize the amount of scouting you need to do at a meet by forming a list of teams that your team could win with if in an alliance together in finals and semifinals.
The link is below: