Our Scout Scheduler is a one-of-a-kind application that makes scheduling an organization's scouts/coaches a breeze! Currently developed for NBA organizations, it is easily adaptable to any sport at any level. Within the application, users can search any game across multiple seasons with easy filters such as conference, team, dates, games with prospects and much more. One can easily click on players, bring up their season schedule, bio information, rankings, notes and lots of extra goodies! Finally, the cream of the crop with this tool, it has a built in scheduling algorithm that optimizes travel for any number of scouts in the organization. Input up to hundreds (or use defaults) of parameters including dates, locations, allowable driving times/distances, games must see, blocked dates, sleep preferences and more. When finished, simply click a button and seconds later and optimized travel schedule will be provided. No longer are you stuck buying tickets the night before - you can now plan months in advance! Don't completely like the schedule? That's ok - change it - and the algorithm works around your inputs! You can even view an example web prototype for the 2013-2014 season online.
Do you ever wonder how one coach can grade a player's speed as a 3 and another coach grades that same player as an 8? We developed a system that eliminates these issues that commonly occur when evaluating players. In short, we have taken the subjective and made it quantitative! Now scouts/coaches can grade players however they want (example: that player is a serious black hole!) - we take those subjective comments, develop a system specific to your organization and then provide you with an easy step-by-step tool that can then be utilized to grade any player at any time. In return, each player is scored a value between 0 and 1 to represent how well that particular player will fit your specific organization. Combined with useful visualization, we have masked the geeky math to only show you what really matters - how do different players stack next to eachother in terms of value to your organization!
We've all seen the great pitchf/x graphs and heatmaps in MLB. Now we've taken that complex dataset and have provided you an easy to use application that allows you to understand a pitcher's tendencies in different situations (outs, count, runners) and be able to PREDICT where the next pitch will be thrown. We have gone way beyond simply just adding and dividing to show you historical stats - we implemented high level analysis to actually predict both where the next pitch will cross the plate and what type of pitch it will be. Two applications were developed based on the user's preference: The first is a traditional filter selection in which you can choose different pitcher/catcher combinations, the current count, the batter stance, the last pitch thrown type and location. A quick heatmap is shown to demonstrate where the next pitch will likely be thrown and what pitch type to expect. The second application is more of a simulator that allows you to simulate past games to understand the accuracy of the analyzer. For example, we ran a past game of Jake Peavy (in 2011) and acheived 60% accuracy in predicting both type and location for all pitches not on count 0-0.
Working with a Division I NCAAF team, we were asked if one could predict if a high school football player would be an All-American in college based on results from several tests (Combine or SPARQ drills). This analysis was conducted a couple years ago so high school data was very sparse - especially in the sense of knowing how those same high school football players turned out in college. To approach this problem, we decided to implement analysis by working backwards - from NFL to NCAA. For example, we took athletes' Combine scores and developed algorithms to determine if they were All-Americans. The benefit of doing it this way allowed us to test accuracy of the algorithms since we already knew the answers. We applied multiple techniques such as Cluster Analysis, Factor Analyis, K-Nearest Neighbor and Artificial Neural Networks across all possible football positions. Using only athletes that participated in all 6 tests over the last 10 years, we saw accuracies in the range of 80%-92%.
Working with an NCAA Hockey organization, we created new metrics that would assist in recruiting new players that optimized lineup configurations. As an example, we created new metrics to measure a player's total impact (TotalIMpact*) they have on a roster. Each position was separated and given a unique metric that applied a combination of goals, assists, hits, blocks and play situations. These metrics were created through the use of Learning Trees and other analytical techniques. In the end, we scored every NHL team, their players and their lineups. We conducted analysis on rookies, veterans and even collegiate players. In addition, we scored the Top 3 utilized lines for every team in the league. Results showed that teams with higher scores in their 3rd-5th lines tended to perform at a higher level than the rest of the league.