I’ve been inspired by Drew Conway’s post about leveraging the wisdom of the crowds for fantasy football. I decided this year to finally sharpen my Python skills, scrape some mock draft data from Fantasy Football Calculator , and learn how to build an R Shiny app! Here’s a link to the full app.
I’m scraping mock draft data every hour and processing the last 7 days worth of data. This data is being stored in a Postgres database and is analyzed and presented in Shiny. The charts are generated using a combination of rCharts and ggplot2. A HUGE thank you goes to the creators and supporters of Shiny and rCharts. I’ve been plastering them with questions, filing github issues left and right and they’ve been insanely helpful.
The main measure I’m displaying is a ‘power index’ on the ‘Power by Position’ tab. This measure is calculated by counting the number of times each player was drafted before another player in every single draft. I then take the max power index, which belongs to Adrian Peterson, and normalize everyone else’s power index by his. This allows me to rank players based on their draft order compared to other players. You can also filter by position and visualize the mean power left by position as you are drafting. This is a useful feature as you can imagine, because your league might be drafting a ton of RBs and a few WRs at first, but you’ll be among the first to notice that WR power is thinning out and you should probably pick someone up soon. I also only included players in this analysis that were in at least 5% of all drafts, which hovers around 205 players daily.
The next tab, ‘Compare Two Players’, allows you to compare players when you’re unsure of which one to draft. The focal measure here is labeled as ‘ADP’ on the y-axis, which is just the average draft position by day for each player. I noticed Kenbrell Thompkins ADP dropping rapidly over the past week, indicating that he’s being picked up earlier in mock drafts. He had a fantastic outing in preseason week 3 against Detroit, which would explain this dramatic change.
If you have a toss-up between which two players to pick, this is a really useful tool. You can compare them both by ADP over time, total number of times picked, and uncertainty, which is the standard deviation in a players’ overall pick number. There’s also a text tip in the bottom left-hand corner that tells you what % of the time one player was chosen before the other. I’ve recently limited the data to the prior 7 days since more recent draft data could be more indicative of potential performance or injuries. I have a draft on Tuesday, Sept. 3rd and will definitely be using this tool! The data will be continuously updating every 4 hours. The link to the full working app is here: Fantasy Football 2013 – Who Should I Draft?
Please leave any suggestions, questions or feature requests in the comments!