You Are Here > > BBR Blog > NBA and College Basketball Analysis

SITE NEWS: We are moving all of our site and company news into a single blog for We'll tag all Basketball-Reference content, so you can quickly and easily find the content you want.

Also, our existing Basketball-Reference blog rss feed will be redirected to the new site's feed. // Sports Reference

For more from Neil, check out his new work at

Layups: Derrick Rose, Plus/Minus MVP?

Posted by Neil Paine on January 25, 2011

In the absence of a runaway choice, there's an ever-growing push among traditional media members in favor of Derrick Rose's MVP candidacy -- and to be totally honest, the advanced boxscore-based stats don't see it. Rose is having a tremendous season, without a doubt, but he's currently 9th in Win Shares, 17th in WS per 48 minutes, 14th in Player Efficiency Rating, and 14th in Statistical +/-... Not exactly the most impressive MVP resume from the stathead's perspective.

However, there is one advanced metric that does validate the love for Rose: Adjusted Plus/Minus (via Sure, the standard errors are huge, and Mike Dunleavy Jr. shows up as the 2nd-best player behind Rose (yikes!). But at least there is some numerical evidence that Rose is making Chicago better in ways that aren't being detected in his box score numbers.


56 Responses to “Layups: Derrick Rose, Plus/Minus MVP?”

  1. huevonkiller Says:

    Adjusted plus minus also has the goofiest rankings.... Especially this half-season.

  2. Walter Says:

    I wouldn't put too much stock into that adjusted stuff this early. I know by most advanced measures Kobe isn't the best in NBA, but he certainly isn't poor and currently those stats show Kobe as the 2nd worst in the NBA!!!!

  3. AYC Says:

    Traditional point guards are undervalued by box score stats and the advanced metrics derived from them. PG's typically score less, rebound less, turn the ball over more, and shoot less efficiently (yes, Stockton lovers, I know there are exceptions). They get more assists, and maybe more steals, but usually not enough to offset these other factors.

    Rose has a remarkably high scoring avg and usage rate for a PG. But he's also having a career year as an assist man, and he's shooting career bests from 3 and the FT line. Like Larry Bird, Rose is undervalued offensively by advanced stats (AS) because he doesn't get to the line enough, and most AS overvalue FT rate as I have pointed out before. Rose is also hurt by the fact that he plays alot of minutes (38.0/mpg). Like usage rate, I think MPG gives you an idea of how valuable a player is to his team, but playing more minutes hurts your per minute stats like WS48 and PER.

    Finally, I want to point out that Rose's AS production is within spitting distance of several other MVP candidates. His PER of 23.0 is within one point of Kevin Durant, Amare, Blake Griffin, and Dirk. His WS48 is within .025 of Deron Williams, Kobe, Dirk and Dwight Howard. Going beyond the stats, Rose has led the Bulls to the 3rd best recoed in the east (ahead of Dwight's Magic) despite the injuries that have plagued Boozer and Noah, the two best players on the team after him.

  4. Greyberger Says:

    Hah if you give it to Rose at +17.9 with a 10.8 standard error, does that make Dunleavy co-MVP with +17.6 and 6.3 SE?

    I think Plus-Minus king for this year is still Dirk, with Chris Paul right on his heels.

    Dirk is out in front in the latest 1-year RAPM batch and rates highly in APM. In unadjusted +/- he's also the biggest net effect on his team and it's not even close - The Mavs are +11 with him and -9 without!

    Of course some of that is getting confused with his teammate's contributions. Maybe sometime during the 2011-2012 season we'll have enough games in the log to properly disentangle Dirk from his teammates.

  5. Neil Paine Says:

    Hey, I'm with you Greyberger. But with the "Rose for MVP" coalition growing in strength, I thought it was appropriate to point out that there is at least one (incredibly noisy) advanced stat out there which considers Rose an MVP front-runner.

    Btw, AYC -- SPM is actually pretty pro-PG, and most metrics consider Chris Paul to be far ahead of Rose. So that's no excuse. Also, I wouldn't go so far as to say all advanced stats overvalue FT rate (maybe TS% does... maybe not).

  6. Greyberger Says:

    Point guards aren't as efficient on offense, but that doesn't mean they'll be under-appreciated by advanced stats... just box-score based mash-up advanced stats.

    For Rose specifically, few linear mash-ups will take into account how great the Bulls are on defense and mediocre they are on offense. Noah, Boozer and Korver are more efficient than Rose, but all three are limited by the opportunities that Rose can create for them, especially Korver and Noah.

    It's kind of the Allen Iverson effect if not as extreme - Thomas, Bogans, Asik and Brewer mostly just play defense while Rose is responsible for most of the offense (even if it ends with a Boozer or Korver shot attempt). Rose is the Bulls' offense, and the Bulls offense is mediocre at best, but few players are as important to the team.

  7. Greyberger Says:

    How could you 'overrate' FT shooting? It seems to me that a made free throw should be 'rated' at one point, against made two-pointers at, let's say, two points, and three points in our model for a three-pointer.

    Wait, that's what TS% does. If Adrian Dantley can score 29 points on 25 possessions per game like he did in '84, why does it matter where he shoots it from?

  8. Jason J Says:

    I'm with AYC here as far as point guards getting a general underrating with comparatively low usage, low FT rates, and such little emphasis being placed on assists. I'd actually like to see someone come up with a point guard specific metric where the valuation was on distribution and efficient scoring from the field.

  9. Greyberger Says:

    I don't get it. Why do point guards get a pass at getting to the line/not getting to the line?

  10. DSMok1 Says:

    Check out the RAPM numbers for this year, though:

    He's down about where the statistical ratings have him. RAPM is much better than APM for small samples like this.

  11. Jason J Says:

    It's about distribution of labor. John Stockton gets Karl Malone the ball in scoring position. That's his job. Karl then uses that position to get to the line. It works. If they switched it, that wouldn't work. Karl Malone cannot run the offense because he doesn't have the handle or the vision (though he became an excellent high post passer). Personally (and this is just me) I want my PG to be concerned with setting up everybody else first, and I'd like to see a metric that rewards that behavior.

    I have a similar problem with measuring wings by the same standards as bigs. Prof Berri had to position adjust his WP because the basic truth of his system is that no one should ever bother to cover a perimeter shooter or run out on the break because defensive rebounds are as valuable as made shots or steals. So if Kobe ignored Ray Allen and attacked the defensive glass in an effort to take rebounds, his WP goes up, and his team loses in 4 games.

  12. AYC Says:

    Neil, I was thinking specifically of WS and PER.

    Anyway, it's an objective fact that TS% overvalues FT rate. As evidence, I will point to Wilt, Shaq and Dwight Howard; those 3 are just the most well-known players who have a TS% higher than BOTH their eFG% and FT%. This is because the full value of made free throws is recognized, but the full value of an attempted FT is not. You get credit for FTA even when you MISS.

    As for PGs being properly rated, I think assists are undervalued, and TOs overvalued. It annoys me that a guy like Dirk gets credit for a low TO%, when he never passes the damn ball.

  13. Greyberger Says:

    Oh. I'd better add that to my list of objective facts, then.

    You do know what TS% is, though, right? It's a scoring metric with points as the numerator and FGA and FTA as the denominator. .44 is used to weight FTA to approximate the possession cost relative to FGA - not .50 because not every 2 FTA end a possession thanks to and1s, techs and flagrants.

    How well you shoot free throws is as much a part of it as getting to the line or field scoring. I don't know how you came to a conclusion like "You get credit for FTA even when you MISS" when you don't get a point out of it and the partial possession counts against you.

  14. Anon Says:

    That may be true Jason, but isn't the pass the "safer" action to make on offense anyway? Even if you get the ball in position to score as a result of the pass, there's still a (usually) coin flip chance that you miss the shot. I don't see how assists are undervalued in relation to the players that actually make the basket.

    Also AYC, Dirk isn't someone who catches and shoots the ball. He is the focal point of the Dallas offense, and defenses work to pressure him into making bad decisions as he creates his own shot. It's an asset for his team that he takes care of the ball pretty well and not create extra possessions for Dallas opponents.

    Just some thoughts about these topics.

  15. AYC Says:

    Greyberger, do you understand what "approximate" means?

    TS% gives you shooting efficiency relative to estimated possessions. But what does determining the number of "possessions" have to do with efficiency anyway? As an independent measure of shooting efficiency, TS% is LESS accurate than it needs to be. Looking at efficiency relative to the actual number of shots taken would be more accurate.

    PS I notice you ignored the indisputable evidence I pointed out. How can TS% be considered accurate when a player's TS% can be higher than his eFG% AND his FT%?

  16. Greyberger Says:

    "But what does determining the number of "possessions" have to do with efficiency anyway?"

    This is as close to comedy gold as you'll get out of a APBRmetric blog.

  17. Neil Paine Says:

    True Shooting % isn't really a "percentage", AYC. It's points per shooting possession (what Hollinger used to call PSA) multiplied by 50 because somebody -- and not Justin or me, so don't blame us -- thought it should vaguely look like FG%.

    If I had my choice, everyone would go back to PSA, but TS% has become the standard way of looking at the metric.

  18. Greyberger Says:

    Re: 17, you know, I had wondered about that.

  19. AYC Says:

    You can mock me all you want; I still have the facts on my side. The silence is deafening regarding TS% and Wilt/Shaq/Dwight....

  20. AYC Says:

    Thanks for confirming what I was saying Neil. The fact is that the casual fan like greyberger is going to assume TS% is telling him something different from what it actually does.

  21. Greyberger Says:

    Ah, the internet.

  22. Neil Paine Says:

    It doesn't really seem like that was what you were saying. It seems like you were arguing that TS% is flawed because, as a "percentage", it shouldn't be possible for it to be higher than eFG% and FT% for some players. And I'm telling you that's not an apples-to-apples comparison because, terminology be damned, TS% isn't really a percentage. It's just a points-per-shot metric that has been scaled to "look" like a shooting percentage.

  23. Greyberger Says:

    "How can TS% be considered accurate when a player's TS% can be higher than his eFG% AND his FT%?"

    The last part of the formula, which I didn't mention when breaking it down, is to divide the 'scoring possessions' part by two so it provides ranges like 45-60 instead of 90-120. It's purely cosmetic; it's so the numbers look like FG% or eFG% or whatever.

    TS% eFG and FT% do not have the relationship you think they do.

  24. Jason J Says:

    Anon – Assists are a highly debatable commodity, no doubt. I have in fact argued on the other side saying essentially that an assist is nothing more than a potential turnover that some talented teammate caught and turned into two points. I ONLY think that assists are undervalued as regards point guards. IMO (and only IMO), the value of each assist is completely context-based.

    When Shaq played in LA, the offense was based on dumping him the ball in the deep post, waiting for the double team to drop down, getting the pass out and swinging the ball to the open shooter as the defense rotated. Shaq got some assists in the process when the defense failed to rotate, but for the most part the assists went to Fox or Horry or Kobe or Fisher as they swung the ball around to the open man. Taking nothing away from those players who made the proper play and helped their team to score that basket, the playmaker in that scenario was Shaq, who gets no credit whatsoever. So whatever assist the perimeter player got for passing to last open man before the shot is not indicative of his actual impact in the play.

    On the other hand, when Steve Nash curls off a pick and roll, keeps his dribble alive deep into the paint, and hits a wide open wing shooter in his shooting pocket for a rhythm jumper as the defense collapses, that’s a valuable play that can only be measured by crediting the assist.

    Both plays result in open made jump shots, but in one example the player credited with the assist might as well have been a bumper for Shaq to skip the ball off of to the shooter, while in the other Nash created the shot for the shooter. In both cases the shooter should be credited with the majority of play, and he is with the points from the made FG. In the case of Nash, that assist where he actually created the opening for his teammate to shoot as well as getting the ball should be worth more than the Rick Fox pass it along method. Again IMO.

    That’s just how I see it. And since we can’t usually (though more and more we’re getting to where we can) measure HOW each player gets each assist, my inclination is to weight a dedicated passing PGs assists more highly because that is his role and for the most part nobody else on his team is being asked to create such a high proportion of shots for others. The pure point guard has less opportunities to get other forms of credit for his play and more responsibility for producing assists.

  25. AYC Says:

    Neil, I understand that. TS% is just points per shooting possession cut in half to look like a %. (I assumed earlier you meant multiply by 0.50, not 50). And that's been my complaint all along; it claims to be a measure of overall shooting %, but that's not what it really is.

  26. Greyberger Says:

    Here's the problem, 24 - what makes a point guard a point guard? Is it just the shortest guy on the floor or the one who brings the ball up? They're usually the same person, but not always. Is it the guy who assists the most, however you want to measure it? This way would correctly identify the shooting guards and forwards who are the lineup's true 'point' (Lebron, Grant Hill in Detriot, Tracy McGrady in '07). But it has its own problems.

    Is it always one guy and then nobody else in that lineup is the point, or is every player to some degree pointy and to some degree as beneficiary of pointedness? It's not as easy to say, point guard assists matter, when really it's the creator's assists that matter and that player is only usually the point guard.

  27. AYC Says:

    I have to agree with greyberger for once; a guy like Lebron, Bird, or Wilt in the late 60's can be his team's primary playmaker without being a PG. I don't want to segregate players at different positions; I want to see assists weighted more heavily; perhaps exponentially?

    Regarding #23, you have failed to realize that even if we recognize TS% for what it really is (points/shot, not a %), the numbers for Wilt/Shaq/Dwight are still clearly bad approximations; multiplying TS% by two does nothing to change that. TS% is neither true, nor a percentage....

  28. Greyberger Says:

    Okay, fine. I'll take this on.

    TS% is awesome. It's great. We all understand that it's a scoring efficiency metric and not a shooting percentage. It measures scoring against scoring possessions.

    You're so sure that the numbers for Shaq and Howard are just plain wrong. But if anything TS% is selling these players short - they surely get more 'bonus' free throws via and-ones than the .44 approximation charges them. So it makes perfect sense that an estimate of their scoring efficiency would be higher than either their FG% or FT%, since they get more than 2.0 free throws per possession used and probably more FTs per possession used than the average player used to construct the TS% weight.

  29. Anon Says:

    Those are good points Jason. I think I would argue though that when you think about it, the guy who sets up the assist chance isn't as penalized as much as his teammate who misses the shot. You don't get the assist; he doesn't get the made bucket AND also gets docked in the % column for the extra FGA. So I think in terms of the penalty the weighting is fair in this regard.

  30. Greyberger Says:

    I'll put it this way - in 2010 Howard attempted 11.7 FG and 11.5 FTs per 40 minutes. He shot 61% on FGs and 59% on FTs. But he also got one and-one opportunity per 40 minutes, so one of those FTs didn't cost a possession and shouldn't be counted in the denominator of a scoring metric that measures scoring against possessions.

    So adjusted for the fact that FTs cost Dwight not .5 of a possession but something more like .44, his FT%-against-possessions scoring rate is better than the 59% FT shooting suggests. Just a tick better, but enough to make his TS% higher than either eFG or FT% and drive you crazy.

  31. Anon Says:

    Also, going back to what I was talking about earlier about the "safer" play - is the guy who can only set people up for assists but not shoot the ball more valuable than the guy who can shoot but not create his own shot? I don't think so. By making that pass you could turn the ball over, but it doesn't happen as often as missing the shot which can go to the other team via the defensive board. So I think weighing the assist more and not keeping these things in mind inflates its actual value.

  32. Greyberger Says:

    Actually Dwight's recent years are more like a .45 or .46 possessions used per free throw, even with all the and-ones he gets. It's Shaq who out-performs the TS% estimate without ever taking three free throws or technicals (and ignoring flagrants). Shaq's personal FT per real TSA was 43.5 last year and 42.5 this year.

  33. Jason J Says:

    Defining what a point guard is definitely is a whole other can of worms, true, true.

  34. AYC Says:

    To have a TS% higher than your eFG% and your FT%, you have to A)have a high free throw rate, and B) have a FT% lower or not much higher than your FG%. Why should that player get credit for attempting more free throws if he doesn't hit them? We don't say that when it comes to FG attempts. Does the skill curve phenomenon apply to FT shooting as well as FG's? If that's the case, why do metrics like WS that don't factor in skill curves use the 0.44 value?

  35. Jason J Says:

    31. That's sort of why I think pure (and I mean as the driven snow) point guards need to be looked at a little differently. You just described Rondo and Ray Allen in a sense - a pure passer and pure shooter. Which is more valuable isn't really what I'm driving at, I'm just trying to say (or unnecessarily insisting) that it doesn't necessarily make sense to rate them along the same statistical lines.

    If Rondo stops passing the ball and insists on trying to score every time he has it... it's bad for everyone. If Ray refuses to shoot the ball and insists on making plays (cough Mike Miller cough), that's bad as well. Not because it hurts the team for Rondo to score or for Ray to drop a dime, but because the offense is maximized by one using his skills as a creator and the other using his skills as a shooter - division of labor like I said above.

    I'm just not sure a metric that properly assesses the value of one is going to be the best bet to assess the value of the other. Maybe something like SPM as Neil mentioned above. It tends to value both versatility and efficiency more highly than other box score metrics. And I'm not saying there's no place for universal metrics. I'd just like to see something that focused more on the value of a player playing his role, but I'm not statistically gifted enough to do that myself.

  36. Greyberger Says:

    You're assuming that free throws and two-point field goals give you an equal opportunity to score if you have the same FG% and FT%. One chance for two points versus two chances for one point, right?

    But you have to consider bonus free throws too. It's usually two chances for one point but sometimes an even better deal. .44 instead of .50.

  37. Walter Says:


    I think I understand what AYC is saying.

    Assume for simplicity a player makes 50% of his field goals (all 2P attempts) and 50% of his free throws. Additionally, assume their are no and-1 situations or technical foul shots taken.

    His eFG% is 50% and his FT% is 50% by definition.... now let's vary the amount of the 20 possessions that were used on field goals and free throws.

    16 possessions used on field goals and 4 on free throws:
    8-16 from the field = 16 points, 4-8 from the line = 4 points, total result is 20 pts on 20 possessions. TS% = 56.3%

    10 possessions used on field goals and 10 on free throws:
    5-10 from the field = 10 points, 10-20 from the line = 10 points, total result is 20 pts on 20 possessions. TS% = 69.4%

    4 possessions used on field goals and 16 on free throws:
    2-4 from the field = 4 points, 16-32 from the line = 16 points, total result is 20 pts on 20 possessions. TS% = 90.6%

    As you can see, we created a situation where the points generated remained the same and only the distribution between shot type (FG vs FT) changed but the TS% changed drastically!

  38. Walter Says:

    To continue my post at #37 further...

    You can relax the standard that the eFG% = FT% and it doesn't matter. Assume an extreme like the FG% = 70% and the FT% = 40%... in this case the points generated per possession by FG's are expected to be 1.4 while FT's only generate 0.8 and thus are much less efficient.

    How does TS% look if we consider a 15/5 split between possesions used on FG and FT and then the opposite?

    15 FGA generates 21 points and 10 FTA generates 4 points so we have 25 points on 20 possessions. TS% = 54.3%

    5 FGA generates 7 points and 30 FTA generates 12 points so we have 19 points on 20 possessions. TS% = 73.1%

    So while fewer points are generated per possession in the second scenario, the TS% actually increases solely because they took more FTs relative to FGA (even if less efficient).

    Basically, any difference between two players TS% could be due solely to the difference FTA/FGA ratio and not really the true "efficiency" by which one scores. The problem occurs when a player has a significant departure from the 0.44 FT rate that is assumed.

  39. Walter Says:

    Sorry... please disregard my previous post as I had an error in calculating the TS%.

  40. Anon Says:

    Jason, it could be a slippery slope to go down if we "award" extra credit for players who only focus on assists though. Do we also award players who are just there to concentrate on rebounding? Defense and blocking shots? Making threes? Etc etc - and then it goes back to the reason why these guys are specialists in the first place, because we wouldn't want them doing the other things that they are not as adept at. Seems a little odd to do this especially when there are players in the league who are pretty good at doing a bit of everything (and are harder to find by nature). Then you also have to properly weigh the factors that are important to winning ball games - is it assist or the basket scored more important? Are they necessarily equal to other? Etc.

    I think this is why the Oliver stats are useful in this regard, because you can look at poss % rate and see who your specialist players are. They still get credit though because these are the players who have the great offensive efficiencies for being specialists - you can find a handful of point guards among the all-time list.

  41. AYC Says:

    Thanks Walter for the effort. 0.512, .532, and .553 are the TS% for his first three examples if I calculated correctly. Those differences are big enough.

  42. AYC Says:

    Note that in all three examples, TS% overstates the shooter's efficiency. But the higher the FT Rate, the greater the overstatement....

  43. DKH Says:

    Err, but the assumption was that no foul shots are the result of and-1 situations or technicals. If that were the case, then the 0.44 coefficient is of course inaccurate. The coefficient should be 0.5. With that coefficient, each TS% would be the same. It sounds like your objection is more that 0.44 is not an accurate coefficient for some players, and you would rather it was individually measured.

  44. Jason J Says:

    Anon - I'm totally with you there, and I guess what I'm failing to convey is that I wouldn't necessarily view these specialist metrics as being comparable to each other. If someone came up with a pg scale that went 1-10, I wouldn't expect that to be useful to say measure Rondo against Durant, but maybe to measure fulfilling the point guard role against Westbrook?

    I don't want to reinvent the wheel here. I'm just spitballin'.

  45. AYC Says:

    That's one objection. Another is that the full value of FTA should be recognized in a stat that purports to measure shooting accuracy.

  46. Greyberger Says:

    It only works if your example is a player who doesn't get any bonus free throw attempts. Bigs and slashers get and-ones, three-point shooters get fouled on threes, and players with a high FT%

    Player rates of bonus free throws per field goal attempt vary a bit. Is that your problem in all this? With play by play parsing we can count exactly how many are bonus and how many are possession-using and don't have to rely on the coefficient. True True Shooting percent if you will. You could parse the data and compare to box score TS to find if there are any kinds of players that consistently are under or over valued by the .44 method (technical free throw shooters, probably), if that's what it would take to convince you that your concerns have been overstated.

  47. Greyberger Says:

    *players with a high FT% shoot Technicals.

  48. Greyberger Says:

    Scoring, not shooting. It's a scoring metric.

  49. EvanZ Says:

    If we go by a combination of box score metrics and Adj. +/-, then Chris Paul is the clear MVP. It's a shame that he plays for a team that nobody cares about. A strong case can also be made for LeBron James...every year. But Rose is the soup de jour, and there's something to be said for that, too, I guess.

  50. Poohdini Says:

    Put down the damn calculator and watch a Bulls game for once.. If anyone is saying Rose isn't MVP, doesn't understand the game of basketball.. Being a stat geek, and number crunching doesn't always show the players importance and impact to that team.. A number can't tell you how many games Rose has showed up big for the Bulls with the absence of Noah and Boozer.. Just last week the Bulls had a starting unit of Bogans/Deng/Taj Gibson/Kurt Thomas.. The Bulls went 3-1 that week on the coat tail of Derrick Rose. Not many players in the league could have carried a team with that type of starting cast.. Thats the purpose of MVP.. Not number crunching but who's the MOST VALUEABLE PLAYER to their team in the league.. He currently has the Bulls 30-14.. If someone told you before this season that Rose would carry the Bulls to that record with Noah/Boozer missing a combined 39 games. You would have laughed.. Like I said before stop the damn number crunching and watch a basketball game or two..

    Overr analyzing things like this is the reason why Kobe only has 1 MVP while Steve Nash has two. Kobe has been robbed so many damn times due to stat geeks or people over analyzing, but is somehow a 5x champion, and 2x Finals MVP..

  51. Jay Says:

    "Overr analyzing things like this is the reason why Kobe only has 1 MVP while Steve Nash has two."

    Actually, that's not true at all. The reason nash won those 2 MVPs was due to your main argument for rose. The suns success completely hinged on nash's play. When he missed a game, they basically had no chance out there. So, in effect, he was the "most valuable player", and had the (some of the) stats to back it up.

  52. Neil Paine Says:

    Yeah, Poohdini's final paragraph is unintentionally hilarious because in 2006, the stat support was much, much stronger for Kobe Bryant than it was for Steve Nash. Nash's two MVPs are still considered by "stat geeks" to be among the most controversial MVP selections ever.

  53. Sean Says:

    NBA has an interesting article proposing a new award that better fits Rose's performance: Offensive Player of the Year.

  54. Neil Paine Says:

    What does it say about the MVP's pro-offense bias that we have a DPOY but not an Offensive Player of the Year Award? It's almost like the MVP is assumed to be basically the OPOY, when it should be a reward for all-around impact.

  55. Kelly Says:

    Paul Pierce crushes Howard or James in adjusted +/-. Where's the cry for him to be MVP?

  56. kev Says:

    If only there were some way we could figure out by talking to actual people if a player busted his butt all summer, comes to every game with a winning attitude and is money in crunch time. We could then combine that with some sort of "Win-Loss" stat to see if that has had an impact and then make a decision.