ORDER
Reginald Davis and Anthony Knox appeal their sentences for a second time. They pleaded guilty to conspiracy to possess with intent to distribute crack cocaine, 21 U.S.C. §§ 846, 841(a). Knox also pleaded guilty to possession of a firearm by a felon, 18 U.S.C. § 922(g)(1). The defendants were subject to a possible life sentence for the drug conspiracy because the amount of crack was at least 50 grams, 21 U.S.C. § 841(b)(1)(A)(iii) (2006), and they *651also qualified as career offenders under the sentencing guidelines, U.S.S.G. § 4B1.1. The district court initially sentenced the defendants to prison terms within the guidelines range, Davis to 220 months, and Knox to a total of 262 months.
After sentencing, however, the Supreme Court held in Kimbrough v. United States, 552 U.S. 85, 91, 128 S.Ct. 558, 169 L.Ed.2d 481 (2007), that a district judge has discretion to impose a prison term below the guidelines range based on the judge’s own view about the appropriate differential in sentences for crimes involving like amounts of powder cocaine and crack. The differences in base offense level that exist in U.S.S.G. § 2D1.1 also carry over to the career-offender guideline, and thus we held in the defendants’ first appeal that sentencing courts have discretion to disagree with the imprisonment ranges resulting from the application of § 4B1.1 to a defendant convicted of a drug conspiracy involving crack. United States v. Knox, 573 F.3d 441, 450, 454 (7th Cir.2009). (We have since held that defendants convicted of substantive drug crimes involving crack also may challenge, on the basis of Kim-brough, a career-offender imprisonment range. United States v. Corner, 598 F.3d 411, 415-16 (2010).) We ordered that Davis and Knox be resentenced.
The district court resentenced Knox in October 2010 and Davis in November 2010. A few months before, Congress had passed the Fair Sentencing Act of 2010 (“FSA” or “Act”), Pub.L. No. 111-220,124 Stat. 2372, which increased from 50 to 280 grams the amount of crack necessary to trigger a possible life sentence under 21 U.S.C. § 841(b)(1)(A)(iii) (2006 & Supp. IV). Five kilograms is still the amount of powder cocaine needed to authorize a life sentence. See id. § 841(b)(1)(A)(ii). The FSA thus reduced the crack/powder ratio in § 841(b)(1)(A) from 100:1 to 18:1. Because the career-offender guideline is based on the offense statutory maximum, see U.S.S.G. § 4B1.1(b), the FSA would have lowered Knox’s offense level under § 4B1.1 from 34 to 31, and with a criminal history category VI, the low end of his imprisonment range also was reduced from 262 months to 188 months. As with Knox, the FSA would have lowered Davis’s offense level under § 4B1.1 from 34 to 31, and the low end of his imprisonment range from 262 months to 188 months. Davis also argued that the district court should go further and, in exercising its discretion under 18 U.S.C. § 3553(a), employ a 1:1 ratio and sentence him as if the conspiracy involved powder, rather than crack, cocaine.
The district court concluded that the FSA did not apply retroactively to defendants whose criminal conduct occurred before the Act was signed into law. Nevertheless, the court considered that if the 18-to-l ratio applied, the Guideline range would be lowered to 188 to 235 months. The court stated that in sentencing Knox, it “looks, in part, to the factors in Section 3553. And the sentence the Court is going to impose will be sufficient, but not greater than necessary, to achieve the goals set forth behind those factors.” The court also said that it was taking into consideration “[t]he disparity between powder and crack cocaine” and “considering where this case would be under the 18-to-l ratio, which is the law now, which would put [Knox] in the 188- to 235-month range.” Then the court observed that Knox was convicted of a “very serious offense” — “not just the drugs and dealing the crack cocaine, but the violent nature of it with the firearms that were involved.” In addition, the court considered the Section 3553 factors, including Knox’s history and characteristics. The judge stated that “I am happy to see the progress that you have made” and “I do see a different person than stood before me previously at the *652sentencing.” But the judge also said that she “was concerned originally and still continue to have some concerns about [Knox’s] violent past....” Explaining that it was “taking the disparity into account and using a different Guideline range rather than the higher one,” the court sentenced Knox to 188 months, at the bottom of the range that would have applied under the FSA, for the drug conspiracy. The court imposed a concurrent sentence of 120 months on the firearm offense, to be served concurrently.
As for Davis, the court noted our precedent that the FSA did not apply retroactively, but announced that “in terms of the disparity issue and in order to address that, I am going to ... use a different Guideline calculation as if the 18-to-1 ratio applied, with a Level 31 and a Criminal History Category of IV, for a corresponding Guideline range of 188 to 235 months.” The court added that “[g]iven the law as it is now, the 18-to-1 is appropriate. And I have used that with respect to the other defendant [Knox] that I have resentenced in this case.” The court emphasized that it was “going to ... use—or at least start from—that particular Guideline range [188 to 235 months] and deviate. I know the calculation for the case is different, but under 3553, given the disparity, I am going to work off of the 31 and VI or 188-to-235 range....”
Acknowledging that the court would be “using the 188-to-235 Guideline range,” the government requested a 188 month sentence. The government explained that a 42-month deviation, what the court gave Knox at his original sentencing, was not necessarily appropriate “now that we’ve moved the Guideline range.” After considering the 3553 factors, including Knox’s history and characteristics, and stating that Knox “appeared] to be a different man standing before the Court today than [he was] at [his] original sentencing” and that he “certainly [did] seem to have turned a corner somewhat and that counters your prior recidivist behavior to some extent,” the court imposed a sentence “sufficient but not greater than necessary to comply with the purpose of [the Section 3553] factors,” namely, 150 months, 38 months below the low end of an FSA-determined range. But the court rejected Davis’s argument that he should be sentenced as if there is no distinction between crack and powder offenses. The court stated that “for all the 3553 factors I noted, whether the mandatory minimum in this case were five years or ten years, I would impose the same sentence here.”
Knox now argues that the district court erred in calculating his guideline range because it did not apply the Act. Davis argues that on resentencing, the court should have applied the FSA and that his prison sentence is unreasonable.
Knox
When Knox filed his appeal, his argument that the district court erred by not applying the FSA was foreclosed by circuit precedent. See, e.g., United States v. Holcomb, 657 F.3d 445, 445-46 (7th Cir.2011); United States v. Fisher, 635 F.3d 336, 340 (7th Cir.2011). Since then, however, the Supreme Court decided Dorsey v. United States, — U.S.-, 132 S.Ct. 2321, 183 L.Ed.2d 250 (2012), in which it held that the FSA’s more lenient penalties apply to pre-Act offenders sentenced after the Act took effect. The government concedes that the FSA’s reduced mandatory minimums should have been applied to Knox, but contends Knox waived his right to challenge the non-application of the FSA at resentencing. Knox responds that he preserved his objection to the court’s calculation of the guideline range for appeal, and at worst forfeited the issue. His reply asserts that the government waived the *653waiver argument by failing to develop it. Knox also claims that the government waived any reliance on plain error review by arguing only that the error in calculating the guidelines range was harmless.
Waiver principles are construed liberally “in favor of the defendant.” Swanson v. United States, 692 F.3d 708, 716 (7th Cir. 2012). “[W]e will find waiver when there are’sound strategic reasons’ explaining why counsel would ‘elect to pursue’ a route’ as a matter of strategy.’ ” Id. We can assume forfeiture where the government proffers no strategic justification for a defendant’s decision to bypass an argument. United States v. Johnson, 668 F.3d 540, 542 (7th Cir.2012); United States v. Anderson, 604 F.3d 997, 1001-02 (7th Cir. 2010).
Before resentencing, Knox filed a sentencing memorandum, objecting to the presentence report’s guideline calculations and arguing that the FSA should apply such that his guideline range would be 188 to 235 months. He also argued that the district court should consider the effects of the FSA under the § 3553 sentencing factors. After stating that it had received Knox’s sentencing memorandum and expressing its view that the very recent decision in United States v. Bell, 624 F.3d 803 (7th Cir.2010), “answers some of the questions” (Bell held that the FSA did not apply retroactively), the district court asked Knox’s counsel if he had any “comments, corrections or objections” to the updated presentence report?” Counsel answered, “No.” The court questioned whether Knox was objecting to the guideline calculation in the updated report, and counsel said, ‘Well, I disagree with it. I do object to that because it’s the exact same of what it was before.” Then the court explained that it was “trying to find out if you are asking for some difference on the Guidelines or if you are just arguing variance and the other arguments under 3553.” Counsel said, “We’ll leave it ... just as a variance with the Guidelines and 3553, not as a technical [objection to the guidelines] calculation.” Later when defense counsel said he was going to summarize Knox’s sentencing memorandum, the court stated, “[Y]ou do not need to reiterate the arguments about the [FSA] because the Seventh Circuit has addressed that definitively in Bell ” and “you do not need to reraise with me orally the [FSA] arguments.”
Although defense counsel said that Knox would leave his argument as a variance with the guidelines, it seems that counsel did so in the face of the controlling authority in Bell that foreclosed any argument that the FSA applied. The government proffers no strategic reason why Knox would have bypassed his argument that the FSA should be applied in calculating his guideline range. Given the record made at the resentencing hearing, defense counsel’s failure to state that Knox was preserving the argument for appeal does not preclude our review. Furthermore, the government’s waiver argument is undeveloped and the government urges us to apply harmless error review. Thus, the government has waived the waiver argument and has waived any argument that the issue was forfeited. See, e.g., United States v. Tichenor, 683 F.3d 358, 363 (7th Cir.2012) (government waived waiver where it urged court to view issue as forfeited).
So we consider Knox’s argument that the district court erred in determining the guideline range. The FSA would have lowered Knox’s offense level under § 4B1.1 from 34 to 31, and with a criminal history category VI, the low end of Knox’s imprisonment range would have fallen from 262 months to 188 months. Knox maintains that by applying the pre-FSA sentencing guideline range, including the *654mandatory minimum term of supervised release, the district court proeedurally erred. This error, he claims, was not harmless since the court did not indicate that it would have imposed the same sentence even if the FSA applied.
An error in calculating the sentencing guidelines range is a procedural error that requires remand unless the government proves that the error is harmless. United States v. Abbas, 560 F.3d 660, 666-67 (7th Cir.2009). “To prove harmless error, the government must be able to show that the Guidelines error did not affect the district court’s selection of the sentence imposed.” Id. at 667. When we are convinced that the sentence would have been the same absent the error, we deem the error harmless. United States v. Hill, 645 F.3d 900, 906 (7th Cir.2011). We have found procedural errors harmless where the district court made it clear that it would have imposed the same sentence even if the error had not been made. Hill, 645 F.3d at 912; Abbas, 560 F.3d at 667.
The government argues that the error in calculating the guidelines range had no effect on the sentence imposed because the district court also calculated the guidelines range under the FSA. It asserts that the court relied on the lower guideline range in imposing sentence, noting the court’s statement that “I am taking the [crack/powder cocaine] disparity into account and using a different Guideline range rather than the higher one” and the fact that the court sentenced Knox to 188 months, at the low end of the FSA guidelines range. Yet the government acknowledges that the court did not state that it would have imposed the same sentence even if the FSA applied to Knox.
However, as the government points out, at the original sentencing, the district court imposed a sentence at the low end of the pre-FSA guideline range. At resentencing, the court used instead the lower guideline range that would have applied under the FSA and imposed sentence at the low end of that range. The court stated that it was taking into consideration “[t]he disparity between powder and crack cocaine” and “considering where this case would be under the 18-to-l ratio, which is the law now, which would put [Knox] in the 188- to 235-month range.” Moreover, just before imposing the 188-month sentence, the court explicitly stated that it was “taking the disparity into account and using a different Guideline range rather than the higher one.” Thus, the court in effect applied the FSA in resentencing Knox.
Even though the court did not expressly state that it would have imposed the same sentence assuming the FSA applied, the record convinces us that the sentence would have been the same. The court explicitly stated that it was using the different range-meaning the FSA-determined range — rather than the higher one. Knox’s suggestion that the court might impose a lower sentence at resentencing would run head on into the court’s statement that the sentence it “is going to impose will be sufficient, but not greater than necessary, to achieve the goals set forth behind [the Section 3553] factors.”
Knox argues that the court found at resentencing that he had changed significantly for the better, which makes it impossible to know whether the court would have imposed the same sentence. But the record shows that the court considered that for four years while in the Bureau of Prisons Knox had not had any disciplinary problems and he appeared to be trying to improve himself so he could be more productive. The court also considered what impact those positive factors should have on Knox’s sentence. However, the court also took account of Knox’s entire history and characteristics, noting his “extreme *655criminal history,” which was “very concerning” to the court. The government had argued that Knox’s positive factors “should have some impact” but not “a great impact” given his complete history and characteristics. After considering the arguments and Knox’s evidence, the court said that it was “happy to see the progress” that Knox had made while in the Bureau of Prisons and hoped that he “follow[ed] through on what he said [about stripping himself of the ‘bad stuff and becoming a’productive citizen’].” Still, the court voiced that it had “some concerns about [Knox’s] violent past” and noted the “seriousness of [Knox’s] offense.” Thus, Knox’s positive factors already are reflected in the sentence imposed.
This case is unlike United States v. Love, 680 F.3d 994, 998 (7th Cir.2012), where, even though the defendant “received a sentence that was significantly below the guidelines range,” there was “no indication that the district court would have imposed the same sentence” had it correctly calculated the guidelines range. Here, we have the district court’s statement that it was using the FSA-determined guideline range rather than the higher range, applicable under then-circuit law.
Nonetheless, a partial remand is appropriate. The FSA lowered not only the mandatory minimum term of imprisonment for defendants like Knox, it also lowered the mandatory minimum term of supervised release from five years to four years. 21 U.S.C. § 841(b)(l)(B)(iii). At resentencing, Knox again was given a five-year term of supervised release. In discussing the new FSA sentencing range, the court made no mention of supervised release. The government requests that if we conclude that Knox did not waive his objection to the non-application of the FSA, we remand for the limited purpose of allowing the district court to determine whether it would have imposed the same term of supervised release had it known that the FSA applied. The government cites United States v. Taylor, 520 F.3d 746, 748-49 (7th Cir.2008), in which we employed a similar remand procedure. But that type of procedure has been limited to forfeited claims where we conduct plain error review, see United States v. White, 582 F.3d 787, 799 (7th Cir.2009), and here we review for harmless error. The change in the mandatory minimum term of supervised release was not discussed at resentencing, and we have no way of knowing whether the district court would have imposed the same term had it known that the FSA applied. It is possible that the term of supervised release “would have been the same, but it is not certain.” United States v. Zahursky, 580 F.3d 515, 528 (7th Cir.2009). Therefore, we must remand for reconsideration of the term of supervised release is also necessary. Such a reconsideration “requires only a brief resentencing procedure.” United States v. Trujillo-Castillon, 692 F.3d 575, 580 (7th Cir.2012).
Davis
As for Davis, he also argues that the FSA was applicable. We have explained already that the FSA is retroactive and thus applicable. As with Knox, the FSA would have lowered Davis’s offense level under § 4B1.1 from 34 to 31, and the low end of his imprisonment range from 262 months to 188 months.
The government concedes that the FSA’s reduced mandatory mínimums should apply to Davis, but contends that we should affirm because the district court explicitly stated that it would impose the same sentence whether the mandatory minimum was five years or ten years. Davis asserts that the court’s statement was not binding and we should accord it no *656weight. Given the district court’s demonstrated understanding that the guidelines are advisory, its accurate computation of the FSA range, its consideration of the § 3558(a) sentencing factors, and the imposition of a sentence of incarceration below that range, we are confident that we can take the court at its word.
Yet, as with Knox, no mention was made about the term of supervised release. As stated, the FSA lowered not only the mandatory minimum term of imprisonment; it also lowered the mandatory minimum term of supervised release from five years to four years. 21 U.S.C. § 841 (b)(1)(B)(iii). At the original sentencing, the district court imposed a five-year term of supervised release, which was the low end of the advisory range. But at resentencing, the court imposed the same five-year term of supervised release that it had imposed previously, prior to the enactment of the FSA. We have no assurance from the record that the court would have imposed the same term whether the mandatory minimum was five years or the FSA’s four-year mandatory minimum. Therefore, a remand only for the purpose of considering the supervised release portion of the sentence is appropriate. As with Knox, such a reconsideration “requires only a brief resentencing procedure.” Trujillo-Castillon, 692 F.3d at 580.
We briefly address Davis’s other arguments for completeness. He argues that the 18:1 ratio adopted through the FSA still suffers from a lack of empirical support and will lead to unwarranted sentencing disparities. See 18 U.S.C. § 3553(a)(6). He also argues that the district court did not understand that the guidelines and the 18:1 ratio are advisory, not mandatory.
After Kimbrough it is understood that sentencing courts have discretion to disagree with the Sentencing Commission’s approach for any guideline. See United States v. Aljabari, 626 F.3d 940, 951 (7th Cir.2010), cert. denied, -U.S.-, 131 S.Ct. 2164, 179 L.Ed.2d 945 (2011); United States v. Reyes-Hernandez, 624 F.3d 405, 415 (7th Cir.2010). But that discretion does not have to be exercised in favor of the defendant’s take on things; a sentencing court may yet agree with the Commission or, as the court did in sentencing these defendants, strike its own balance. See United States v. Curb, 626 F.3d 921, 927-28 (7th Cir.2010); United States v. Scott, 555 F.3d 605, 610 (7th Cir.2009). We have repeatedly held that there are adequate justifications for treating crack offenses more seriously than crimes involving powder cocaine. See United States v. Moore, 644 F.Sd 553, 556-57 (7th Cir. 2011); Scott, 555 F.3d at 610. Here, the district court considered Davis’s argument for a 1:1 ratio but decided that 18:1 was appropriate. The court was not required to consider Davis’s argument that the 18:1 ratio is not supported by empirical evidence, see United States v. Aguilar-Huerta, 576 F.3d 365, 367-68 (7th Cir.2009), and his argument that the court did not understand the guidelines are advisory lacks merit: The court sentenced Davis below the range that would have applied under the FSA.
Finally, Davis argues that the district court erred by sentencing him as a career offender. But we previously considered and decided this issue, Knox, 573 F.3d at 452-53, and arguments that were made or could have been made in the last appeal cannot be raised here, United States v. Swanson, 483 F.3d 509, 515 (7th Cir.2007); United States v. Husband, 312 F.3d 247, 250-51 (7th Cir.2002).
* * *
Accordingly, we Remand for resentenc-ing on the term of supervised release for *657both Knox and Davis, but otherwise Affirm their sentences.