Edward DELL, Petitioner-Appellant, v. UNITED STATES of America, Respondent-Appellee, 710 F.3d 1267

MARCUS, Circuit Judge: Appellant Edward Dell appeals the district court’s denial of his 28 U.S.C. § 2255 motion — one in a series of attempts, some more successful than others, to reduce his sentence for committing several drug offenses in violation of 21 U.S.C. §§ 841(a)(1) and 856. At the heart of this appeal is whether Dell’s defense counsel was constitutionally ineffective for failing to argue, either during sentencing or on direct appeal, for a downward variance based on the substantial disparity between the Sentencing Guidelines’ treatment of cocaine base and cocaine powder.

After thorough review, we conclude that Dell’s defense satisfied the constitutional requirements established in Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), and, therefore, we affirm the judgment of the district court. I. A. Tony Henry ran an operation selling cocaine base, i.e., crack cocaine, with the help of Edward Dell and several others. United States v. Henry, 307 Fed.Appx. 331, 333 (11th Cir.2009). On June 6, 2006, the United States filed a criminal complaint against Henry, Dell, and another co-defendant, Tomiki Jenkins, on various crack cocaine-related charges.

After Dell was arrested, the court appointed a federal public defender as his counsel. On November 14, 2006, the government filed the operative superseding indictment against Henry, Dell, Jenkins, and Henry’s wife, Helena Jones, alleging that this family- and-friends operation had engaged in a criminal conspiracy to possess and distribute crack cocaine.

Dell went to trial, and, on July 27, 2007, a jury found him guilty of: (1) conspiracy to possess with intent to distribute fifty or more grams of cocaine base, in violation of 21 U.S.C. § 841(a)(1); (2) knowingly and intentionally possessing with intent to distribute less than five grams of cocaine base, in violation of § 841(a)(1); (3) three counts of knowingly and intentionally distributing cocaine base, in violation of § 841(a)(1); and (4) opening, using, or maintaining a place for the purpose of distributing crack cocaine, in violation of 21 U.S.C. § 856.

The jury also found Dell’s co-defendants Henry, Jenkins, and Jones guilty of, inter alia, conspiracy to possess with intent to distribute fifty or more grams of cocaine base. On June 11, 2007, shortly before Dell’s trial, the Supreme Court granted certiora-ri to consider whether a district judge’s “sentence ... outside the guidelines range is per se unreasonable when it is based on a disagreement with the [Sentencing Guidelines] disparity for crack and powder cocaine offenses.” Kimbrough v. United States, 552 U.S. 85, 91, 128 S.Ct. 558, 169 L.Ed.2d 481 (2007).

Under the Sentencing Guidelines at the time, U.S.S.G. § 2D1.1 (2007), a drug trafficker dealing in crack cocaine was “subject to the same sentence as one dealing in 100 times more powder cocaine.” Kimbrough, 552 U.S. at 91, 128 S.Ct. 558. The Supreme Court heard oral argument in Kimbrough on October 2, 2007. On November 1, 2007, the district court sentenced Henry to life imprisonment.

On November 5 and November 6, 2007, Jones and Jenkins filed memoranda challenging their Presentence Investigation Reports (“PSR”), which included extensive arguments that they were entitled to downward variances due to the hundred-to-one crack/powder disparity. The memoranda referred to the Supreme Court’s pending decision in Kimbrough. Dell’s counsel, however, did not submit a sentencing memorandum. On November 7, 2007, the district court sentenced Dell to 235 months in prison.

This sentence was at the bottom of Dell’s guideline range of 235 to 293 months, which was based on an offense level of 36 and a criminal history category of III. At the sentencing hearing, Dell’s attorney objected to the factual accuracy of the PSR, requested a minor role adjustment, and asked for a downward variance based on the 18 U.S.C. § 3553(

a) factors, but the district court rejected all of these arguments. Notably, Dell’s lawyer never raised the possibility of a downward variance based specifically on the crack/powder disparity. Later that month, the court sentenced Jones — whose guideline range was 188 to 235 months based on her offense level of 36 and criminal history category of I — to 188 months. The court then sentenced Jenkins — whose guideline range was 151 to 188 months based on his offense level of 34 and criminal history category of I — to 151 months.

During Jenkins’s sentencing, his counsel, after speaking at length about the sentencing disparity between crack and powder cocaine, asked the district court to at least acknowledge that it lacked discretion to vary downward based on that disparity, and the court so ruled. On December 10, 2007, the Supreme Court issued its opinion in Kimbrough and firmly rejected the idea that the disparity between crack and powder cocaine sentences was “effectively mandatory.” 552 U.S. at 91, 128 S.Ct. 558. Two terms earlier, in United States v.

Booker, 543 U.S. 220, 125 S.Ct. 738, 160 L.Ed.2d 621 (2005), the Court had decided that mandatory guidelines were unconstitutional under the Sixth Amendment. See id. at 233, 125 S.Ct. 738. After considering the history of the crack/powder disparity and its interaction with the Sentencing Guidelines, the Court concluded that, like all other types of sentences post-Booker, the guideline range for crack cocaine crimes was merely advisory.

Kimbrough thus invested the district courts with discretion to vary downward if they believed that the crack/powder disparity resulted in greater-than-necessary sentences. 552 U.S. at 91, 110, 128 S.Ct. 558. Kimbrough abrogated this Circuit’s contrary holding in United States v. Williams, 456 F.3d 1353 (11th Cir.2006). Dell timely appealed the district court’s judgment and sentence, and we consolidated his appeal with those of his co-defendants.

Dell’s attorney — the same one who served him during the trial and at sentencing — did not make an argument based on Kimbrough; his opening brief contained only a five-page argument attacking the district court’s factual findings. A panel of this Court affirmed Dell’s conviction and sentence, Henry, 307 Fed.Appx. at 337, but remanded Jones’s and Jenkins’s cases for resentencing in light of Kimbrough, id. at 338-40.

On remand, the district court expressly stated it would treat crack cocaine as equivalent to powder cocaine based on its post-Kimbrough discretion, which reduced Jenkins’s offense level from 34 to 28 and Jones’s from 36 to 28, leaving them both with a guideline range of 78 to 97 months. Because Jones was subject to the mandatory ten-year minimum found in the statute, 21 U.S.C. § 841(b)(1)(A), the district court reduced Jones’s term of imprisonment from 188 months to 120 months.

Jenkins qualified for the safety valve of U.S.S.G. § 5C1.2, so he was not subject to the mandatory minimum and received a reduction from 151 months to 87 months, the midpoint of his recalculated guideline range. B. In 2010, Dell filed a pro se 28 U.S.C. § 2255 motion to vacate, set aside, or correct his sentence. Dell advanced four arguments in his motion, including that his counsel was ineffective for failing to raise a sentencing challenge pursuant to Kim-brough at sentencing or on appeal.

The district court denied Dell’s motion on March 28, 2011, explaining that “Dell’s appellate counsel was not ineffective for not raising a Kimbrough issue because that issue was not raised at Dell’s sentencing.

A defendant may not raise arguments on appeal that were not made to the sentencing judge.” As for Dell’s lawyer’s conduct during sentencing, the district court reasoned that “Dell was sentenced before the Gall and Kimbrough decisions were entered,” and that “[i]t is not ineffective assistance ... for an attorney to fail to foresee a change in the law____” The court opined at the conclusion of its analysis that “[i]t would be just for this Court to treat Dell the same as his two co-defendants in determining his sentence,” but it believed it had no power to do so.

The order granted Dell a certificate of appealability (“COA”) on one claim: “[Wjhether his sentencing counsel was ineffective for failing to raise an argument that the Court should grant a downward variance treating the crack guidelines the same as that used for powder cocaine. Subsequent to sentencing, this argument was accepted by ... Kimbrough as a valid reason to support a downward variance from the guidelines.” Dell appealed the district court’s denial of his § 2255 motion and filed his opening brief pro se.

That brief contained a request for us to expand his COA to include his claim that his appellate counsel was ineffective as well. Subsequently, we directed that counsel be appointed for this appeal and requested supplemental briefing on two issues: first, whether it would be appropriate to expand' the COA since Dell had not filed a separate motion for such expansion; and second, whether appellate counsel was in fact ineffective. During this time, the law surrounding crack cocaine offenses changed substantially. In the Fair Sentencing Act of 2010, Pub.L.

No. 111-220, 124 Stat. 2372, Congress, among other things, reduced the hundred-to-one disparity in the treatment of crack and powder cocaine to approximately eighteen-to-one. The United States Sentencing Commission subsequently promulgated Amendment 750 to the Sentencing Guidelines, which altered the Guidelines’ drug table, U.S.S.G. § 2D1.1(c), to reflect the new eighteen-to-one ratio between powder and crack cocaine. United States Sentencing Guidelines Manual, app. C, at 391-98. A year later, in Amendment 759, the Commission made parts of Amendment 750 retroactive and thereby applicable to Dell’s case.

See United States Sentencing Guidelines Manual, app. C, at 416-21. Based on Amendments 750 and 759, the district court sua sponte reduced Dell’s sentence from 235 months to 188 months, the bottom of Dell’s guideline range after his offense level was reset at 34 based on the new drug table. Dell then moved pro se, pursuant to 18 U.S.C. § 3582(c)(2), for reconsideration of that sentence, arguing that his good behavior during incarceration warranted an additional reduction to 168 months.

In an April 2, 2012, order entered during the pendency of this appeal, the district court further reduced Dell’s offense level by one, yielding an offense level of 33 and resulting in a guideline range of 168 to 210 months’ imprisonment. This one-level reduction was in addition to the reduction already produced after recalculating the offense level using Amendment 750’s revised drug table and was purely at the district court’s discretion.

In explaining its reasons for this additional reduction, the court specifically stated that it was “tak[ing] into consideration the sentences of [Dell’s] co-defendants [Jenkins and Jones],” which had already been reduced after the same judge exercised his discretion post-Kimbrough to take into account the crack/powder disparity. The court ultimately imposed a sentence of 168 months precisely in order “to better align [Dell] with the sentences of his co-defendants” and stated that “a sentence of 168 months [wa]s appropriate” in light of all the 18 U.S.C. § 3553(

a) factors and Dell’s conduct while incarcerated. II. When considering a district court’s denial of a 28 U.S.C. § 2255 motion, “ ‘we review findings of fact for clear error and questions of law de novo.’ ” McKay v. United States, 657 F.3d 1190, 1195 (11th Cir.2011) (quoting Rhode v. United States, 583 F.3d 1289, 1290 (11th Cir.2009)). Ordinarily, “the scope of our review of an unsuccessful § 2255 motion is limited to the issues enumerated in the COA.” Id. Both prongs of the Strickland inquiry present mixed questions of law and fact and therefore receive de novo review upon appeal. Cade v.

Haley, 222 F.3d 1298,1302 (11th Cir.2000). As an initial matter, we grant Dell’s request to expand his COA to include the issue of ineffectiveness of appellate counsel. While we generally consider only requests to expand a COA when made by motion, this Circuit has not established a strict rule that all improperly formed requests for expansion must be rejected. See Jones v. United States, 224 F.3d 1251, 1255-56 & n. 8 (11th Cir.2000). In fact, we have expanded a COA sua sponte on exceptional occasions, even after oral argument. See, e.g., Clark v. Crosby, 335 F.3d 1303, 1307 (11th Cir.2003); Bishop v.

Burnett, 312 Fed-Appx. 252, 253 (11th Cir. 2009). In Hodges v. Att’y Gen., State of Fla., 506 F.3d 1337 (11th Cir.2007), however, we made clear that an appellant granted a COA on one issue cannot simply brief other issues as he desires in an attempt to force both the Court and his opponent to address them. See id. at 1340. As the panel in Hodges explained: It is one thing for an appellate court in an unusual case to be persuaded during its consideration of the merits of a granted issue to expand the COA to include a related issue and to request supplemental briefing on that previously excluded issue.

It is another thing for an appellant to simply ignore the COA order and brief any issue he pleases. We recognize the former practice and condemn the latter. Id. at 1341-42. Dell’s case is an example of just such an unusual case. Although Dell improperly requested an expansion of his COA through his initial brief, which he filed pro se, and not by motion, we subsequently appointed counsel to argue this appeal on Dell’s behalf and directed supplemental briefing both on expanding the COA to include the ineffectiveness of appellate counsel claim and addressing the merits of that claim.

Dell did not attempt to circumvent the need for a COA, unlike the appellant in Hodges. Moreover, the parties have already joined issue on this question at our directive. Generally, the value of the COA requirement is “to weed out non-substantial issues ... before the parties and the court immerse themselves in the more substantial issues which deserve close scrutiny on the merits.” Id. at 1340. Here, that concern is absent because we specifically requested supplemental briefing on the issue.

We expand a COA when “reasonable jurists would find the district court’s assessment of the constitutional claims debatable or wrong.” Slack v. McDaniel, 529 U.S. 473, 484, 120 S.Ct. 1595, 146 L.Ed.2d 542 (2000). In this case, the district court disposed of Dell’s ineffective assistance of appellate counsel claim too quickly, rejecting it in these words: “Since Dell’s counsel had not raised the argument at sentencing, it could not be made on appeal.

Appellate counsel is not ineffective for failing to raise an issue that is barred.” But Dell still had the opportunity to raise his Kimbrough argument for the first time on appeal. The consequence of failing to raise that argument at sentencing was not that it was barred on appeal; rather, the panel hearing Dell’s direct appeal could have reviewed it, if raised, for plain error. See United States v. Doe, 661 F.3d 550, 567 (11th Cir.2011). And such plain error challenges based on intervening Supreme Court precedent have been successful in the past. See, e.g., United States v.

Antonietti, 86 F.3d 206, 208-09 (11th Cir.1996); United States v. Walker, 59 F.3d 1196, 1198 (11th Cir.1995). The district court disposed of the issue of Dell’s appellate counsel’s effectiveness based on the mistaken belief that appellate counsel could not have raised Kimbrough at all, and as a result did not adequately address Dell’s ineffective assistance claim. We therefore grant Dell a COA on this issue and turn to its merits. III. A.

Since Dell’s appellate counsel’s failure to raise Kimbrough months after it had been decided presents the more substantial question under Strickland, we address it first. “Claims of ineffective assistance of appellate counsel are governed by the same standards applied to trial counsel under Strickland.” Philmore v. McNeil, 575 F.3d 1251, 1264 (11th Cir.2009) (citing Heath v. Jones, 941 F.2d 1126, 1130 (11th Cir.1991)); see also Smith v. Robbins, 528 U.S. 259, 285, 120 S.Ct. 746, 145 L.Ed.2d 756 (2000).

Under Strickland, a petitioner must show both that counsel’s perform ance was so deficient that it “fell below an objective standard of reasonableness,” 466 U.S. at 688,104 S.Ct. 2052, and “that there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different,”- id. at 694,104 S.Ct. 2052. “A reasonable probability is a probability sufficient to undermine confidence in the outcome.” Id.

We presume counsel was effective and conduct the Strickland analysis in a “highly deferential” manner, “considering all the circumstances .... from counsel’s perspective at the time,” see id. at 689, 104 S.Ct. 2052. In this case, we need not decide, and in fact have serious doubts about, whether counsel’s appellate performance met the minimum objective standard of reasonableness required by the Constitution. Instead, we hold that Dell has failed to satisfy Strickland’s prejudice prong and therefore cannot establish that his counsel was constitutionally ineffective.

See Strickland, 466 U.S. at 697, 104 S.Ct. 2052 (“[A] court need not determine whether counsel’s performance was deficient before examining the prejudice suffered by the defendant as a result of the alleged deficiencies.”); accord Holladay v. Haley, 209 F.3d 1243, 1248 (11th Cir.2000) (“[T]he court need not address the performance prong if the defendant cannot meet the prejudice prong, or vice versa.” (citation omitted)). In evaluating prejudice, the relevant proceeding is Dell’s direct appeal, which was briefed in 2008 and decided on January 13, 2009.

Dell’s principal claim, embodied in the expanded COA, is that his appellate counsel was then ineffective because he failed to raise the argument that Dell warranted resentencing in light of Kimbrough. It is therefore important to reconstruct the precise circumstances his appellate counsel confronted. By that time, Dell’s attorney- — who had also conducted his trial and represented him at sentencing — had already failed to preserve the claim that Dell was entitled to a downward variance due to the substantial disparity between the treatment of crack cocaine and powder cocaine.

Notably, at the time of Dell’s sentencing, this Circuit had held that district courts lacked the discretion to grant this type of variance. See Williams, 456 F.3d at 1369. Only after Dell’s sentencing, but before briefing on his direct appeal, did the Supreme Court decide Kimbrough, which abrogated Williams and expressly granted district courts the discretion to vary downward based on the erack/powder disparity notwithstanding the Sentencing Guidelines. Thus, had Dell’s appellate counsel raised a Kimbrough argument, the direct-appeal panel would necessarily have applied plain error review.

See Doe, 661 F.3d at 567. To determine whether there was prejudice, therefore, we must judge whether there was a reasonable probability that Dell’s Kimbrough claim would have won the day in 2009, when a panel of this Court decided Dell’s direct appeal. In effect, Strickland requires us to put ourselves in the position of that direct-appeal panel and consider the following issue: did the district court commit plain error by not considering the crack/powder disparity in determining whether to vary Dell’s sentence below the guideline range?

And we may consider that issue only with the aid of all the record evidence Dell’s counsel could reasonably have presented to the direct-appeal panel in 2009. Strickland unambiguously commands us to do so by directing us to evaluate counsel’s conduct “at the time” of the relevant proceeding and to avoid “second-guessfing]” or “the distorting effects of hindsight.” See 466 U.S. at 689, 104 S.Ct. 2052.

This directive also limits our inquiry into Strickland’s prejudice prong, where we must discern whether “the decision reached would reasonably likely have been different absent [counsel’s] errors.” Id. at 696, 104 S.Ct. 2052. It is impossible for the decision reached by the direct-appeal panel to have been different if the evidence needed to convince that panel could not have existed at the time when the panel considered the case. Put simply, when Dell asserts he was prejudiced, what he means is that a competent appellate attorney would likely have won him resentencing on direct appeal by raising Kimbrough.

Common sense naturally limits our inquiry to the record reasonably available at the time of Dell’s direct appeal. The test for plain error, as explicated by the Supreme Court in United States v. Olano, 507 U.S. 725, 732-35, 113 S.Ct. 1770, 123 L.Ed.2d 508 (1993), is a formidable hurdle to overcome. To demonstrate plain error, Dell would have been required to show the direct-appeal panel that during his sentencing proceeding there was (1) error; (2) that it was plain; and (3) that it affected his substantial rights. See United States v. Cotton, 535 U.S. 625, 631, 122 S.Ct. 1781, 152 L.Ed.2d 860 (2002).

Dell would have borne the burden of demonstrating these three prongs. See United States v. Bacon, 598 F.3d 772, 777 (11th Cir.2010). If he had met that burden, then the panel would have had discretion to notice the error “ ‘if (4) the error seriously affect[ed] the fairness, integrity, or public reputation of judicial proceedings.’” Id. at 631-32, 122 S.Ct. 1781 (quoting Johnson v. United States, 520 U.S. 461, 467, 117 S.Ct. 1544, 137 L.Ed.2d 718 (1997)). Dell would have met the first two prongs.

There was error, because the district court sentenced Dell prior to Kimbrough, at a time when this Circuit’s precedent in Williams barred the court from varying downward based on the erack/powder disparity. We confronted this same type of claim in reviewing a host of plain error challenges raised in the wake of Booker, and panels of this Court consistently found error in the sentencing courts’ application of mandatory guidelines even if the district court itself did not explicitly indicate it would have sentenced differently under an advisory guidelines regime. See, e.g., United States v.

Ronda, 455 F.3d 1273, 1304-05 (11th Cir.2006) (defendants all established “statutory Booker error” because they “were sentenced under a mandatory Guidelines system”); United States v. Underwood, 446 F.3d 1340, 1343 (11th Cir.2006) (“The district court erred when it sentenced Underwood because it considered the Guidelines to be mandatory....”); United States v. York, 428 F.3d 1325, 1336 (11th Cir.2005) (“[T]he district court did commit statutory Booker error by sentencing York under a mandatory Guidelines scheme.”); United States v.

Taylor, 417 F.3d 1176, 1183 (11th Cir. 2005) (“Because the district court treated the guidelines as mandatory, it committed an error that is plain.”). The same error occurred in Dell’s proceeding; under our then-controlling precedent in Williams, the crack/powder disparity errjbodied in the Sentencing Guidelines was mandatory, and therefore Kimbrough error necessarily occurred. See United States v. Regalado, 518 F.3d 143, 147 (2d Cir.2008); United States v. Taylor, 520 F.3d 746, 747 (7th Cir.2008). Moreover, this error was plain. In United States v.

Rodriguez, 398 F.3d 1291 (11th Cir.2005), a panel of this Court considered, under plain error review, a sentencing argument based on Booker, which the Supreme Court issued after the defendant’s sentencing but prior to his appeal. See id. at 1297-98. In addressing the second prong of plain error, the panel in Rodriguez explained: The second prong of the plain error test is also met.

Although the error was not “plain” at the time of sentencing, “where the law at the time of trial was settled and clearly contrary to the law at the time of appeal — it is enough that the error be ‘plain’ at the time of appellate consideration.” Id. at 1299 (quoting Johnson, 520 U.S. at 468, 117 S.Ct. 1544). The same circumstances existed during Dell’s direct appeal. Dell was sentenced on November 7, 2007. The Supreme Court decided Kimbrough on December 10, 2007. See 552 U.S. at 85, 128 S.Ct. 558. Dell timely appealed the case, and Kimbrough was the law at the time of his appeal. See Henry, 307 Fed.

Appx. at 338-40 (remanding for resentenc-ing of two of Dell’s co-defendants in light of Kimbrough). The law at the time of appeal was both settled and contrary to the law at the-time of trial, which rendered the error plain. The third prong is the sticky wicket in this case.

Dell would also have had the burden of establishing that the error affected his substantial rights; that is, that the error “affected the outcome of the district court proceedings.” Olano, 507 U.S. at 734, 113 S.Ct. 1770. “The standard for showing that is the familiar reasonable probability of a different result ..., which means a probability ‘sufficient to undermine confidence in the outcome.’ ” Rodriguez, 398 F.3d at 1299 (quoting United States v. Dominguez Benitez, 542 U.S. 74, 83, 124 S.Ct. 2333, 159 L.Ed.2d 157 (2004)).

But if the chances of a greater and a lesser sentence weigh in equipoise, a defendant cannot establish that an error affected his substantial rights. As Rodriguez put it, “[w]here errors could have cut either way and uncertainty exists, the burden is the decisive factor in the third prong ..., and the burden is on the defendant.” Id. at 1300. Rodriguez therefore applied the test in this manner: [I]n applying the third prong, we ask whether there is a reasonable probability of a different result if the guidelines had been applied in an advisory instead of binding fashion by the sentencing judge in this case.

The obvious answer is that we don’t know. If the district court judge ... had the liberty of increasing or decreasing Rodriguez’s sentence above or below the guidelines range, he might have given Rodriguez a longer sentence, or he might have given a shorter sentence, or he might have given the same sentence. The record provides no reason to believe any result is more likely than the other. We just don’t know. Id. at 1301. Thus, Rodriguez failed to meet his burden under the third prong of Olano.

See also Ronda, 455 F.3d at 1305 (“The sentencing record provides no basis for a conclusion that any Appellant has a reasonable probability of receiving a more lenient sentence under an advisory Guidelines system. The district court did not make any comments that the sentences imposed were too severe.”); Taylor, 417 F.3d at 1183. The question, then, is whether there is a reasonable probability that the panel deciding Dell’s direct appeal would have found, based on the record evidence available at the time, that the Kimbrough error had affected the outcome of the district court proceedings.

In doing so, the direct-appeal panel would have had to acknowledge that the burden was on the defendant and that if the record provided no reason to believe one result was more likely than another- — i.e., that a downward variance was more likely than a guideline-range sentence — Dell’s plain error claim could not prevail. See Ronda, 455 F.3d at 1305; Taylor, 417 F.3d at 1183. On the other hand, if Dell could tip the balance in his favor by showing that the district court was reasonably likely to grant a downward variance, then his plain error claim would have succeeded on direct appeal.

In this context, Strickland’s prejudice inquiry comes down to whether there was a reasonable probability that Dell could have overcome the third prong of Olano plain error. If Dell cannot demonstrate a reasonable probability that he would have hurdled that bar in 2009 on direct appeal, then he cannot now show prejudice under Strickland, for the outcome of his direct appeal would have been no different even if his appellate attorney had raised the Kimbrough argument.

There are only two pieces of evidence in the direct-appeal record that could even arguably be read as implicating the issue of the crack/powder disparity. The first is that at co-defendant Jenkins’s sentencing hearing, Jenkins’s counsel asked the district court to acknowledge that it could not take the crack/powder disparity into consideration because of Williams, and the court replied, “I so rule.” Second, Dell was sentenced at the bottom of his guideline range, as were his co-defendants Jenkins and Jones.

We cannot, however, consider two additional pieces of evidence generated later that Dell urges us to incorporate into the calculus. First, Dell points out, accurately, that when the direct-appeal panel remanded Jenkins’s and Jones’s cases for resentencing, both of them received substantially lower sentences. Jenkins had his sentence reduced from 151 to 87 months, while Jones received a reduction from 188 to 120 months. However, both proceedings occurred on May 21, 2009, and postdated the resolution of the direct appeal.

Thus, the direct-appeal panel could not possibly have considered these resen-tencing proceedings when determining whether the district court’s Kimbrough error had affected Dell’s substantial rights. Similarly, Dell notes that the same judge who sentenced him later decided his 28 U.S.C. § 2255 motion.

In the order denying that motion, the judge opined that “[i]t would be just for this Court to treat Dell the same as his two co-defendants in determining his sentence.” While this statement, standing alone and untethered from the district court’s subsequent reduction of Dell’s sentence, would tend to show that the Kimbrough error did affect Dell’s substantial rights, it was made in a March 2011 order denying Dell postconviction relief — more than two years after his direct appeal was decided. Again, there is no way the direct-appeal panel would have been able to consider this order.

Dell could not have met his burden under the third prong of plain error review based solely on the first two pieces of record evidence described above. As we said in Rodriguez, where “we don’t know” what the district court would have done, and “[t]he record provides no reason to believe any result is more likely than the other,” the appellant cannot prevail, 398 F.3d at 1301; accord id. (“[Wjhere the effect of an error on the result in the district court is uncertain or indeterminate — where this Court would have to speculate — the appellant has not met his burden.”).

Or, stated yet another way, “where- the record does not provide any indication that there would have been a different sentence if the court had recognized and exercised its § 3553(

a) discretion and treated the guidelines range as merely advisory, the party with the burden of showing a reasonable probability of a different result loses.” Id. at 1304 (internal quotation marks omitted). Our law could not be clearer: without some contemporaneous indication that the district court would have varied downward if it had the discretion to do so, the direct-appeal panel could not have found that the error affected Dell’s substantial rights.

The problem under Olano’s third prong is that the evidence available at the time of Dell’s direct appeal gave no indication at all that the district court was likely to vary downward if it could have, or even that the court nurtured this ambition.

During Jenkins’s sentencing hearing, when asked to rule that Williams controlled and prevented any discretion in sentencing based on the crack/powder disparity, the district judge said only, “I so rule.” In neither words nor substance does this off-hand rejection of Jenkins’s Kimbrough-type argument suggest that the court was disposed, let alone ready, to sentence differently than it did. The court said only that it lacked discretion, without any indication of how it would have exercised that discretion.

Neither does the fact that Dell, Jenkins, and Jones were sentenced at the bottom of their guideline ranges demonstrate that the court was reasonably likely not only to sentence at the bottom, but to vary downward from the guideline range if it had the chance. The decision to sentence at the bottom of the guideline range could have been based on any number of individualized determinations based on the information contained in the PSRs, including the nature of the offense, the defendant’s own offense conduct, or the presence of a prior criminal history.

Indeed, given that the three participated jointly in a drug conspiracy, the court could have felt that the common underlying criminal conduct warranted a sentence at the bottom of the range but no lower, which would explain the court’s consistency in sentencing. We cannot fairly interpret a sentence within the guideline range as dissatisfaction with the range itself. This Court confronted essentially the same issue in the context of a Booker claim. In United States v.

Fields, the defendant mounted a plain error challenge to his sentence, and a panel of this Court found that he met the first two prongs of Olano. 408 F.3d 1356, 1360 (11th Cir. 2005). With regard to Olano’s third prong, Fields argued that he could show an effect to his substantial rights based “on the fact that the district court imposed the lowest sentence in the guidelines range.” Id.

The panel in Fields squarely rejected this contention and explained “that the fact that the district court sentenced the defendant to the bottom of the applicable guidelines range establishes only that the court felt that sentence was appropriate under the mandatory guidelines system. It does not establish a reasonable probability that the court would have imposed a lesser sentence under an advisory regime.” Id. at 1361.

Applying Rodriguez’s holding, the panel in Fields emphasized that the burden of establishing that the error affected the defendant’s substantial rights belonged to the defendant; “the fact that the district court went as low as it could ..., without more, is not enough to carry that burden,” because that argument “is too speculative, and more than speculation is required.” Id. The logic of Fields applies with equal force in this context. Indeed, in sentencing Dell, the court did not evince any dissatisfaction with the proceedings or the sentence. Cf.

Underwood, 446 F.3d at 1344 (rejecting a plain error Booker challenge on Olano’s third prong when “[t]he record indicate[d] no frustration on the part of the district court with the severity of the Guidelines sentence, nor did the district court indicate a desire to impose a lesser sentence in [Defendant’s] case.”).

The court’s only commentary on the sentence was its statement that it “ha[d] sentenced [Dell] at the bottom end of the guideline range because [it] determine[d] that sentence to be sufficient for the offense behavior in this case.” In addition, when Dell’s co-defendant Jenkins attempted to argue that the Sentencing Commission had grown dissatisfied with the existing guideline ranges, the district court interrupted, “That’s why they just dropped the guidelines two levels as to crack,” presumably a reference to Amendment 706, which the Sentencing Commission had promulgated shortly before the court sentenced Dell, Jenkins, and Jones.

If anything, then, the record indicates that the district court summarily brushed aside the possibility of varying downward due to Williams, and that it felt that the guideline ranges had already been adjusted to reflect the excessive crack/powder disparity. Nor does Kimbrough itself prove that Dell’s substantial rights were affected, despite his assertions to the contrary.

Kim-brough addressed whether a district court could vary from the advised guideline range for a crack cocaine offense or whether the disparity between crack and powder cocaine sentences was “effectively mandatory.” See 552 U.S. at 91, 128 S.Ct. 558.

After considering the history of the crack/powder disparity and its interaction with the Sentencing Guidelines, the Supreme Court ultimately concluded that “it would not be an abuse of discretion for a district court to conclude when sentencing a particular defendant that the crack/powder disparity yields a sentence ‘greater than necessary’ to achieve § 3553(a)’s purposes, even in a mine-run case.” Id. at 110, 128 S.Ct. 558. Kimbrough empowered the district courts with this discretion, but it did not command them to exercise it.

Thus, even taking Kimbrough into account, the direct-appeal panel would still have had to search the extant record evidence for a reasonable probability that the district court would have exercised that discretion. And it simply could not have found anything to overcome Olano’s third prong. But even if we were able to consider everything that happened after (in fact, long after) Dell’s direct appeal, as he urges, it is still altogether unclear this would benefit Dell’s argument.

Notably, in 2012, the district court sua sponte reduced Dell’s sentence from 235 to 188 months, and then, upon Dell’s motion for reconsideration, further reduced the sentence to. 168 months. The first reduction — from 235 to 188 months — was authorized by Amendment 750 to the Sentencing Guidelines.

The second reduction, however, which was the product of lowering Dell’s offense level by one, was a discretionary decision by the district court specifically intended “to better align [Dell] with the sentences of his co-defendants.” At the time when the court twice lowered Dell’s sentence, it had all the relevant information before it, including but not limited to the Supreme Court’s decision in Kimbrough, Congress’s reduction of the crack/powder disparity in the Fair Sentencing Act, the Sentencing Commission’s subsequent alterations to the Guidelines in Amendments 750 and 759, and the district court’s knowledge that it had resen-tenced Jenkins and Jones to 87 and 120 months, respectively.

In choosing the 168-month sentence — which was exactly what Dell asked for in his motion — the court explained that it had considered all the relevant § 3553(

a) factors (which included § 3553(a)(6)’s aim of avoiding “unwarranted sentencing disparities among defendants with similar records”), and that 168 months was appropriate. Unlike in the earlier order denying Dell’s 28 U.S.C. § 2255 motion, the district judge evinced no reservations that this sentence was unfair or unjust to Dell; on the contrary, he indicated that the reduction brought Dell’s sentence in line with those received by Jenkins and Jones.

Thus, even if we could consider all the record evidence to date, this evidence actually undercuts Dell’s speculation that the district judge somehow retained an unspoken desire to lower Dell’s sentence even further, since the judge chose not to go lower than 168 months when he had the opportunity to do so in 2012. Dell urges us to make this speculative leap because he still has not received as much of a benefit from resentencing as his co-defendants have.

But, although it is true that Dell’s sentence remains higher than those of Jenkins and Jones, the starting baseline for each of them was substantially different. Dell had both a higher offense level (36 vs. 34) and a greater criminal history (III vs.

I) than Jenkins. Moreover, Jenkins qualified for the safety valve of U.S.S.G. § 5C1.2 and thus was not subject to the ten-year mandatory minimum sentence that applied to Dell. As for Jones, she had the same initial offense level as Dell, but her criminal history placed her in Category I. The fact that Dell received a sixty-seven month reduction in his sentence belies his assertion that his substantial rights were affected.

It also means that, whatever Kimbrough error existed in Dell’s sentencing proceeding, in light of the entire record it would be hard to fairly conclude that it “affect[ed] the fairness, integrity, or public reputation of [his] judicial proceedings,” Cotton, 535 U.S. 625 at 631-32, 122 S.Ct. 1781 (internal quotation marks omitted). Far from convincing us that there is a reasonable probability that the district court would reduce Dell’s sentence still further if it had still another chance, the post-direct appeal record in fact reveals quite the opposite.

We emphasize once more, however, that this hindsight approach is forbidden under Strickland, and we elaborate upon it only to demonstrate how flawed it would be in any event.

Indeed, if we determined Strickland prejudice for Dell’s direct appeal based on what occurred long afterward, then we would be left with the untenable conclusion that Dell was prejudiced with regard to the outcome of his 2009 direct appeal as of 2011, when the district court rejected his 28 U.S;C. § 2255 motion, but that the prejudice vanished or at least was largely eliminated in 2012, when the district court granted Dell’s 18 U.S.C. § 3582(c)(2) motion and twice lowered his sentence.

This kind of post hoc analysis makes little sense when we remember that the prejudice Dell is attempting to demonstrate-is that the direct-appeal panel would not have denied him relief in 2009 if his appellate counsel had presented a plain error argument for sentencing discretion based on Kim-brough.

It would be impossible even for the most artful appellate lawyer to have constructed an argument to overcome Olano’s third prong based on record evidence that only came into existence some two years later, just as it would be impossible that such an argument could have been undone or undermined by what occurred three years later. Under controlling law, what we are left with is the simple fact that, when evaluating the record at the time of direct appeal, no panel of this Court could possibly have known what this district court would have done if it knew that it had discretion to sentence Dell differently.

And “where the record does not provide any indication that there would have been a different sentence ... the party with the burden ... loses.” Rodriguez, 398 F.3d at 1304 (internal quotation marks omitted). Since Dell was that party, he has failed to show a reasonable probability that his Kimbrough claim would have succeeded on direct appeal when subjected to plain error review. Thus, he has not demonstrated Strickland prejudice with regard to his ineffective assistance of appellate counsel claim, and we affirm the district court’s denial of Dell’s § 2255 motion on this issue. B.

We turn to the issue of whether Dell’s sentencing counsel was ineffective for failing to request a downward variance based on the crack/powder disparity. Dell urges us to find his sentencing counsel ineffective because, although Kimbrough had not yet been decided, a Kimbrough-type argument was nonetheless readily available during his sentencing. After thorough review, we cannot hold that Dell’s attorney was ineffective in this context.

We have never required counsel to anticipate future legal developments — such as the outcome of a pending Supreme Court case — in order to be meet the constitutional minimum for effective advocacy, and we decline to do so here. Dell “ha[d] a constitutional right to effective assistance of counsel at sentencing.” Wilson v. United States, 962 F.2d 996, 997 (11th Cir.1992). To demonstrate ineffective assistance of counsel at sentencing, Dell must show under Strickland that his attorney’s performance “fell below an objective standard of reasonableness.” 466 U.S. at 688, 104 S.Ct. 2052.

Because we presume counsel was competent, Dell “must establish that no competent counsel would have taken the action that his counsel did take.” Chandler v. United States, 218 F.3d 1305, 1315 (11th Cir.2000) (en banc). And we must keep in mind that “[c]ounsel is not required to present every nonfrivolous defense,” since “[sjtacking defenses can hurt a case.” Id. at 1319. Rather, “[g]ood advocacy requires winnowing out some arguments ... to stress others.” Id. (internal quotation marks omitted).

Thus, it is quite difficult to establish that the omission of any particular argument resulted in ineffective assistance, although such a showing is possible if the argument or arguments neglected were stronger than the ones counsel actually offered. See, e.g., Heath, 941 F.2d at 1130-32 (finding deficient performance when counsel in a capital case raised only one argument on appeal).

Dell’s claim fails to satisfy the first prong of Strickland because a reasonable lawyer could have omitted the argument for a downward variance based on the crack/powder disparity at sentencing, a time when the Supreme Court’s resolution of that issue was still uncertain. In Jones, for example, the Supreme Court granted certiorari on October 10, 1989, to review a Second Circuit case involving an issue relevant to Jones’s attempt to suppress wiretap evidence. See 224 F.3d at 1257.

As a panel of this Court explained: From that date, Jones’s counsel was on notice that the Court would be considering the circumstances under which wiretap evidence must be suppressed. Over a month after the grant of certiorari, Jones’s counsel moved to suppress the wiretap evidence. Jones’s motion did not raise the delay in sealing the tapes as a basis for their suppression.... Even two months after the grant of certiorari, when filing a memorandum in support of the suppression motion, counsel did not raise the issue.

Since the district court would be required to follow the law of this circuit until it was overruled by the Supreme Court or an en banc panel of this court, it was not completely unreasonable for counsel to make a strategic decision to forego a claim that was a loser under the then-current state of the law. The highest standards of practice would suggest that Jones’s counsel should have acted to preserve Jones’s rights in light of the Supreme Court’s unequivocal signal that a ruling would be forthcoming.

But we are not prepared to say categorically that counsel’s failure to do so constituted prejudicial, ineffective nonfea-sance while the law was still unsettled. Id. at 1257-58 (footnotes omitted). In short, it generally does not fall below the objective standard of reasonableness for trial counsel to fail to raise a claim in anticipation that undeniably would lose under current law but might succeed based on the outcome of a forthcoming Supreme Court decision. See also Pitts v.

Cook, 923 F.2d 1568, 1574 (11th Cir.1991) (“[L]aw-yers rarely, if ever, are required to be innovative to perform within the wide range of conduct that encompasses the reasonably effective representation mandated by the Constitution.”); accord id. at 1573-74 (“Because law is not an exact science, an ordinary, reasonable lawyer may fail to recognize or to raise an issue, even when the issue is available, yet still provide constitutionally effective assistance.” (internal quotation marks and alteration omitted)).

Dell cannot show that his attorney’s performance at sentencing “fell below an objective standard of reasonableness,” Strickland, 466 U.S. at 688, 104 S.Ct. 2052, since competent counsel could have taken the action that his counsel did take. Even if Dell’s lawyer would have been wiser to include an argument for a downward variance based on the crack/powder disparity, he was “not required to present every nonfrivolous defense,” Chandler, 218 F.3d at 1319.

At sentencing, Dell’s counsel pursued several different objections, including an attack on the factual accuracy of the PSR, a request for a minor role adjustment, and a request for a lower sentence based on the 18 U.S.C. § 3553(

a) factors. This attempt to winnow down the possible arguments fell within the “wide range of conduct that encompasses the reasonably effective representation mandated by the Constitution,” Pitts, 923 F.2d at 1574. Moreover, when we grant counsel’s conduct the presumption of competence, the choice of several plausible factual or legal arguments and the omission of a Kim-brough-type claim may be regarded as having been “a strategic decision to forego a claim that was a loser under the then-current state of the law.” Jones, 224 F.3d at 1258.

At the time of sentencing, unlike the time of appeal, Williams was the law of this Circuit, and it squarely held that the downward variance Dell now claims his counsel should have requested was impermissible. Quite simply, trial counsel’s failure to make a Kimbrough-type claim in the face of what was then the law of this Circuit does not render counsel constitutionally ineffective. Because Dell has failed to satisfy the first prong of Strickland concerning his trial counsel, we have no occasion to address the question of prejudice.

Since we find no merit to either of Dell’s ineffective assistance of counsel claims, we conclude that the district court correctly denied Dell’s § 2255 motion. AFFIRMED. . Dell's offense level was originally calculated as 38 based on the factual finding that the weight of crack cocaine involved in the conspiracy was 1.5 kilograms. However, Amendment 706 to the Sentencing Guidelines, which became effective on November 1, 2007, reduced the offense level for that quantity of crack cocaine by two levels. See United States Sentencing Guidelines Manual, app. C, at 226-31 (2012). .

The government did not appeal this additional reduction in Dell’s sentence. . We do not suggest that a § 2255 movant may simply seek expansion of the COA in his principal brief on appeal. In fact, in most cases, such an attempt will be deemed untimely and rejected because the proper way of requesting an expansion of a COA is by motion. . The government suggested at oral argument that we should not consider statements made during co-defendants' parallel sentencing proceedings while conducting plain error analysis and cited Payne v. United States, 566 F.3d 1276 (11th Cir.2009). That case is inappo-site.

Payne did not address how to conduct plain error review and hence never held that we cannot scour the whole record extant at the time for evidence of whether an error affected a defendant's substantial rights.

We can identify no logical reason why we should be restricted solely to what happened at Dell’s sentencing, especially since the effect to his substantial rights that Dell asks us to find— that the district court would have varied downward if it had known it had discretion to do so, based on its disagreement with the crack/powder disparity' — would be precisely the sort of error that would have affected all of the defendants' substantial rights if it affected one of them. Nor has the government provided us with a procedural or substantive rule of law that bars us from considering co-defendants' proceedings.

We regularly consider co-defendants’ sentencing when determining whether an error occurred at a defendant’s sentencing; in fact, we are obliged to do so in some cases, given 18 U.S.C. § 3553(a)(6)'s aim of avoiding “unwarranted sentencing disparities among defendants with similar records.” See, e.g., United States v. Dorman, 488 F.3d 936, 944-45 (11th Cir. 2007) (in reviewing reasonableness of a defendant’s sentence, this Court examined the co-defendants’ sentencing proceedings and found that they were not similarly situated because they had provided substantial assistance).