Comparing Pitcher-Hitter Split For Various ADPs

After working my way through the first eight rounds of the website ADP draft,  some sites leaned heavy pitcher while others were drafting just hitters. Before continuing the draft, I wanted to determine the split difference for each website. The difference wasn’t as big as I thought considering the results seen to this point in the draft.

Just for reference, I used the ADP data collected at FantasyPros from ESPN, Yahoo, NFBC, CBS, FanTrax, and Real Time Sports (RTS). Next, I assumed people were drafting a 12-team league with 23 players (14 hitters, 9 pitchers) which works out to 276 players. Then, I gave each player an auction dollar amount based on their ADP. Finally, I divided them into two groups and added up their auction dollars. Here are the results

The spread is just under 3% from 62.9% hitters at Yahoo and Fantrax to 65.8% at ESPN. I’m a little surprised the values were so close since some the sites specialize in points, draft-and-holds, and best-ball leagues. In just the first few rounds of my mock draft, Fantrax and RTS have drafted over 50% pitching while ESPN and the NFBC are both under 20% pitching. Just a small difference in the spread makes a huge difference in my experiment. Even Yahoo’s and RTS’s slight edge in reliever valuation has them cornering the closer market.

While the websites differences are close, there are some obvious implications. While drafting off ADP is probably not the ideal method, some owners know that everyone is seeing in the draft software and know that those top (visible) names aren’t going to last long. If more pitchers are near the top and tempting owners, they will get picked earlier. Owners know they will just have reach a bit more for pitching than they would if they were on another website.

Also, some owners are starting to regress player talent to the market when creating their evaluations. Before owners use an ADP for regressing their projections, they should at least have an idea of how their split will change. If the owner desires a 67/33 split but the ADP source pushes the split down to 65/63, the owner might be a dollar or two lower on hitters which can equate a difference in a round or two difference in value.

Besides the pitcher-hitter split, I went a step further with Fantrax (easy data download and player matchup) and found how their preseason ranks influence their current ADP for the top 276 players. And their answer was almost none. The r-squared between the Average and Fantrax ADP was .945. Between the Fantrax’s initial ranks and ADP, it was just .065. When I use them both inputs, the overall ADP jumped to .946. Basically, the initial ranks (at least for Fantrax) have been priced out.

One final test, I found the r-squared between the different sites to see which ones correlate (top 253 players by AVG ADP who was in each sample).

All the sites correlate the best if the Average value with all the r-squares at or over .90. Between the sites, the ESPN and Yahoo matching had the worst single correlation and both weren’t great between other sites. On the other hand, CBS and the NFBC were the most alike and probably could be used for proxies for the overall average with an r-squared of 0.97.

That’s it for now. My kids have been fighting all day and I don’t have the energy to wrap everything up nice and tidy. Just let me know if you have any questions and I’ll answer them once I get my sanity back.