Most experts seem to have Louisville in their preseason Top 10 next year. And because people need to justify that ranking, between now and November you are going to be hearing a lot about Louisville recruit, point guard Chris Jones. Jones is the top junior college player in the country, and everyone is saying he will be an instant impact recruit. (Besides Chris Jones, some of the other top JUCO recruits this year include SMU’s Yanick Moreira, Marquette’s Jameel McKay, and Missouri’s Keanau Post.)

But what is a reasonable expectation for a top JUCO recruit? For many of us, we simply want to see Jones play before we draw any conclusions. Scouting is hugely important, and I wouldn’t be watching the Jordan Brand Classic and writing about it, if I didn’t see value in watching players and picturing how they will fit into an offense.

But if you want to look at the historical stats and use those to predict how Chris Jones might play, that is what today’s column is all about.  But before I share those numbers, let me point out that the biggest problem with junior college players is attrition. JUCO players are systematically more likely to have problems with eligibility, systematically more likely to withdraw at the start of the school year, and systematically more likely not to make it through the entire season. I thought this post by John Templon of NYBuckets summed it up rather nicely. Below I am only presenting the numbers for players that enrolled successfully and played more minutes than a typical walk-on. Only 77% of Top 100 JUCO players met this criteria, meaning my sample suffered a 23% attrition rate. Because of this, the most important hurdle for Louisville may be making sure Chris Jones is on the court on November 1^st.

Second, I think it is important to set a baseline for expectations for incoming players. I’m going to use freshmen as that baseline. The next table is a recreation of a table Luke Winn made a few years ago, measuring the first year impact of various freshmen. The only difference here is that I am using 11 years of data, from 2003 to 2013. Remarkably, the expectations are quite similar in the expanded data set.

Occasionally, a team of freshmen come together and dominate from Day 1, but this year’s Michigan squad was definitely an outlier. More realistically, only the highest rated freshmen tend to come in and dominate the first year they step on campus.

Now, in last year’s preseason rankings, I didn’t do much with the JUCO data. About the only fact I used was this one: From 2005 to 2013, JUCO transfers into Power programs have performed better than JUCO transfers into smaller programs.

I typically view JUCO players as emergency fill-ins. A JUCO player can come in and get you a 97.0 ORtg on average, which is better than the 94.3 for an unranked freshman. But it isn’t a lot better.

When I published my preseason rankings last year, one of the initial suggestions I received was that I should use the JUCO Top 100 data and see if it provided some additional information with which to form expectations. As it turns out, over 80% of the JUCO players enrolling in Power 6 conferences were ranked in the JUCO Top 100. (An example of a JUCO player outside the Top 100 at a major program would be Cincinnati’s David Nyarsuk.)  I have now coded that Top 100 data, and the next table shows how the JUCO Top 100 have fared in their debut seasons. For 2011 and 2012, I am using the JUCO rankings found on Jucorecruiting.com. From 2007-2010, I am using the JUCO rankings created by Juco Junction, a division of rivals.com.

I didn’t break out the Top 100 further because the variance for these players is huge. Using six years of data, none of the tiers of data are statistically significantly different from one another. To put it another way, when you plot efficiency relative to the JUCO ranking, the data is virtually a scatter plot. There are players at all points in the rankings with efficiency ratings below 90 and above 110. I’m not completely sure why the JUCO rankings have so little predictive power. It could be the differences in competition, the lower quality of scouting, or simply the different circumstances these players walk into. (I.e. a JUCO player matriculating at Texas Tech and Marquette probably faces a substantially different probability of success.)

That said, I think it is fair to say the Top 10 JUCO recruits look very similar to freshmen recruits ranked 21-30 out of high school. If you want to set your expectation for Chris Jones at 101.5 based on this data, that seems reasonable.

You might argue that my Top 10 cut-off is too wide. Maybe if I just focused on the Top 5 or Top 3, we would get an even higher expectation for a player like Chris Jones. But the data don’t really support that. If we cut off players ranked 6-10, we cut out some of the best JUCO players to go D1. You would be dropping Mississippi’s Marshall Henderson (#10 in 2012) and LSU’s Marcus Thornton(#6 in 2007) among others. And unfortunately, a number of players near the top of the JUCO rankings have bombed badly. Do you remember that Mario Little and Tyrone Appleton were ranked #1 and #3 in the JUCO rankings before attending Kansas in 2009? They did nothing. The #2 JUCO recruit in 2007 Nemanja Calasan had a dreadful 88.1 ORtg for Purdue. The #2 JUCO recruit in 2011 Kansas St.’s Freddy Asprilla was little better at 91.2.

Wichita St.’s Cleanthony Early (#2 in 2013), Missouri’s Ricardo Ratliffe (#1 in 2011), and Baylor’s Pierre Jackson (#1 in 2012) were all brilliant college players. So there are clearly some success stories. But do not let the positive outcomes for those players cause you to be overly optimistic about Louisville’s Chris Jones. Because for every Pierre Jackson, there is 2012 #2 JUCO Nurideen Lindsey, who posted an 83.8 ORtg before transferring out of St. John’s. I should also add that PG is notoriously the hardest position to play well immediately.

Probably the best thing Louisville’s Chris Jones has going for him is that Rick Pitino believes in him. If Pitino thinks he can compete and possibly start immediately, expectations should be pretty high. At this point, I see Jones as having a slightly higher expected value than Louisville’s other incoming PG, Terry Rozier. Rozier was the #75 RSCI recruit out of high school before spending a year at prep school. But I would expect them both to play. My model has Jones playing 55% of the minutes and Rozier playing 45% of the minutes.

Without Russ Smith, my predictions model currently has Louisville 15^th nationally. With Russ Smith, my model would now have Louisville 3^rd, behind only Kentucky (with the greatest recruiting class of all time) and Michigan St. (assuming Adreian Payne and Gary Harris return.) 15^th may seem unreasonably low, but remember that Russ Smith was Ken Pomeroy’s player of the year as the most important tempo free player, so obviously his loss is going to hurt in any model that relies on tempo free stats. Personally, I think 15^th may be a little low. But it all comes down to what you are trying to measure. My model has Iowa slightly ahead of Louisville right now. Iowa almost certainly has less of a chance of being an elite team or national title contender than Louisville. But Iowa had the 23^rd best margin-of-victory numbers last year, returns basically its entire rotation, and has little chance of a disastrous season. I think most people tend to make their predictions at least partly based on upside, and I won’t argue with Louisville in the 8-12 range in most people’s rankings.

Overall, the top JUCO recruits can be dominant players. But much like freshmen recruits outside the Top 20, there are no guarantees.