concurring in part and dissenting in part.
There is much to admire in the majority’s opinion — its convincing rejection of *570Carter’s aggravating-factors argument and its convincing rejection of his argument that he did not default a claim of ineffective assistance of appellate counsel. But instead of stopping there and resolving the two claims raised by Carter on appeal, the court grants appellate relief on a claim of its own making (and preserving): ineffective assistance at the mitigation phase of the trial. There are several problems with this approach: the claim exceeds the scope of the certificate of appealability; Carter did not raise the claim in the district court; Carter did not raise the claim on appeal; and the claim is meritless. Other than that, it ought to give a court pause to refuse to excuse a procedural default and yet to grant relief on a related, but doubly forfeited, claim. This kind of analysis, needless to say, has a high degree of difficulty, and I am afraid the majority cannot pull it off. I must respectfully dissent.
Exhibit A is the certificate of appealability, a jurisdictional prerequisite to review, see 28 U.S.C. § 2253(c), which does not encompass this theory of relief. The district court granted a COA on one claim: the state trial court’s alleged misuse of aggravating factors at the mitigation stage. It did not grant a COA on today’s theory of relief, and Carter did not seek one. At our court, Carter moved to expand the COA to cover four reasons why we should excuse the procedural default of his ineffective-assistance-of-mitigation-counsel claims: (1) ineffective assistance of appellate counsel established cause and prejudice for the default; (2) Carter had a due process right to effective appellate counsel during the Ohio-law Mumahan proceedings; (3) the Ohio Supreme Court “impliedly” addressed the claims by rejecting the claims he did bring; and (4) a miscarriage of justice would arise if the claims were defaulted. See Mot. to Expand COA at 6-15. Missing from his motion, as the majority effectively concedes, was any argument that he never defaulted the claims in the first place because he fairly presented them in state court on direct appeal. Consistent with what Carter requested, and did not request, we expanded the COA to reach “an additional issue,” noting that “the issue raised in Carter’s present motion warrants further review[:] ... whether the district court properly dismissed Carter’s ineffective assistance of counsel claims because he procedurally defaulted those claims in state court.” COA Order at 2 (emphases added). Neither the district court’s COA nor our expansion of it nor Carter’s COA’s applications raise a claim that Carter in fact fairly presented a claim of ineffective assistance of mitigation counsel in the state courts.
Exhibit B is Carter’s litigation position at the district court, where he forfeited, indeed waived, the argument that he presented this claim in the state courts. His brief in the district court concedes the “failure of [Carter’s] direct appeal counsel to timely present the following meritorious issues to the state court of appeals,” and listed these claims — Claims # 28 and 29— as two of the “meritorious issues” never exhausted in the state courts. R. 124 at 15. Trying to overcome this exhaustion problem, he argued that his “state appellate counsels’ failure to raise” his ineffective-assistance-of-trial-counsel claim “amounted to ineffective assistance of appellate counsel.” Id. at 13. Nowhere did he argue that in fact he presented his claims in the state appellate courts. He instead asked the district court to excuse the default. All of this amounts to a forfeiture of the claim and a waiver to boot. See Akins v. Easterling, 648 F.3d 380, 396-97 (6th Cir.2011).
Exhibit C is Carter’s litigation position in this court, where he also forfeited and waived this argument. The forfeiture: he does not argue in his appellate briefs that *571the district court should have granted the writ on the ground that he fairly presented his ineffective-assistance-of-mitigation-counsel claim to the state courts. The waiver: he says that “ineffective assistance of appellate [counsel] constitute^] cause for his failure to present claims ... 28, 29 ... to Ohio courts.” Carter Br. at 51 (emphasis added). Here, too, he argues only that ineffective assistance of appellate counsel provides cause to excuse this default. He never mentions the argument the majority now makes. At our court, then, this argument also was forfeited and waived. See United States v. Johnson, 440 F.3d 832, 845-46 (6th Cir.2006). Anyone who doubts me should read his appellate briefs for themselves. At the relevant pages, he says not a word about this theory of relief. See App’t Br. at 36-51; Reply Br. 8-10. Indeed, the whole point of his argument is to contrast the “omitted strong claims” (nine of the eleven) with the two (weak) claims that were raised on direct review. Reply Br. 8-9. Carter surely is allowed to make alternative arguments. But to do so, he (and not the Court) must make them.
Exhibit D is the flawed nature of this argument, which helps to explain why the individual with the greatest incentive to raise the point — Carter—never made it himself. In contending that Carter presented this claim — Claims # 28 and 29 — to the state courts, the majority invokes Carter’s Assignment of Error X at the state court of appeals. To “fairly present” a claim — here Carter’s claim of ineffective assistance of counsel at the mitigation phase — the applicant must present both the “factual and legal underpinnings of the claim” to the state courts. Pudelski v. Wilson, 576 F.3d 595, 605 (6th Cir.2009). A claim for federal habeas relief thus cannot turn “upon a different allegedly ineffective action [from] the claim presented to the state courts.” Caver v. Straub, 349 F.3d 340, 347 (6th Cir.2003). Otherwise, state courts would not have a “fair opportunity” to address constitutional claims, defeating the whole point of exhaustion. See O’Sullivan v. Boerckel, 526 U.S. 838, 844-45, 119 S.Ct. 1728, 144 L.Ed.2d 1 (1999).
In Error X, Carter raised two arguments: (1) trial counsel performed ineffectively in using Dr. Chiappone as a mitigation witness; and (2) trial counsel performed ineffectively by not calling Carter’s mother to testify. The state courts rejected both claims on the merits.
When the time came to file his federal habeas petition, Carter remained free to frame his claims as he saw fit. Had he raised just these two issues as independent habeas claims, I could agree that he preserved them. But for reasons of his own — perhaps the prospect of withstanding AEDPA review on the two claims — he expanded his ineffective-assistance claims (in Claims 28 and 29) to a wholesale attack on all aspects of counsel’s mitigation strategy. Habeas Claims 28 and 29 raise eleven theories of relief: Counsel (1) failed to obtain a “real” mitigation expert, (2) failed to obtain a neurological examination, (3) failed to research school records or call an education expert, (4) failed to call school officials, (5) failed to call a drug-addiction specialist, (6) failed to inform the jury that Carter’s mother had applied for disability benefits on his behalf, (7) failed to obtain psychiatric records, (8) failed to call his mother to testify, (9) failed to call his siblings to testify, (10) should not have permitted Dr. Chiappone to testify and (11) should not have given a disjointed closing argument. See App. at 58-63.
Precedent does not allow a habeas applicant to raise two theories of mitigation ineffective assistance in the state courts, then mix in nine new, unexhausted theories when he arrives in federal court. Such “mixed petitions” are not permitted. *572Rose v. Lundy, 455 U.S. 509, 510, 102 S.Ct. 1198, 71 L.Ed.2d 379 (1982). Habeas claimants are masters of their theory of relief, and if they present an “ineffective assistance of counsel claim based on one ground” in state court they may not present “an ineffective assistance of counsel claim based on another ground” in federal court. Bell v. Bell, 460 F.3d 739, 761 (6th Cir.2006), vacated on other grounds, 512 F.3d 223, 237 (6th Cir.2008) (en banc). “[T]he doctrine of exhaustion requires that a claim be presented to the state courts under the same theory in which it is later presented in federal court,” Wong v. Money, 142 F.3d 313, 322 (6th Cir.1998), as the majority seems to acknowledge by relying on this exact case, Maj. Op. at 568-69.
Here, the omnibus ineffectiveness claims in 28 and 29 that the majority gives life rely on a far broader theory — and a far broader set of factual bases — than the claims presented to the Ohio courts. With the exception of Allegations 8 and 10, Carter never presented the rest of these claims — any of them — to the state courts. The majority downplays this, describing the unexhausted claims as merely “some additional detail about the deficiencies of Dr. Chiappone’s testimony and how, in Carter’s view, it was caused by his failure to perform a diligent inquiry into Carter’s background.” Maj. Op. at 569. But that is like saying chess is just checkers with a few additional details: at some point the additions so outweigh the commonalities that you have a new game. Carter’s “entirely new underlying” ineffective-assistance-of-counsel allegations resulted in a “fundamental alteration to the legal claim already considered by state courts,” and thus I “cannot say that the state courts had the first opportunity to review the claim.” Bell, 460 F.3d at 761. That is why Carter acknowledges his “post-trial counsel! ] fail[ed] to properly exhaust claims based on these facts,” App. at 2577-80 (listing facts mentioned for the first time after post-eonviction), and explains why he never made the argument the majority now embraces.
In the final analysis, the majority grants relief based on a claim that we do not have jurisdiction to hear, that was not raised below or at the court of appeals, and that compromises basic habeas principles of preservation and exhaustion. That is not an everyday occurrence, and it is one from which I must respectfully dissent.