concurring in the judgment:
As to the first, “mother-was-raped” issue proffered by the defense, I agree that there was no constitutional error. Moreover, as matters turned out, there was no abuse of trial court discretion in keeping that testimony out of the defense case. I write separately, however, to highlight why, in my view, the court erred in exercising its discretion, and why there was no abuse of discretion requiring reversal only because the error was harmless. See (James) Johnson v. United States, 398 A.2d 354, 366-367 (D.C.1979).
I.
As defense counsel was beginning his opening statement, the trial court sustained the government’s objection to counsel’s allusion to the anticipated “mother-was-raped” testimony. The court cut off counsel’s explanation as to why that testimony would reflect “frame of mind,” ruling that the testimony would not be “relevant.” That abrupt ruling, without allowing further elaboration by counsel, was erroneous because frame of mind is relevant to self-defense, see Howard v. United States, 663 A.2d 524, 528 n. 6 (D.C.1995), and the court as yet had no basis for concluding that the proffered testimony had no such bearing.
Later, after the government had put Johnson’s many brutal acts in evidence, the government was apparently unsure whether the trial court had ruled definitively on the “mother-was-raped” testimony and was concerned that such testimony *305might provide traction for the defense if admitted. The government, therefore, filed a motion in limine to bar the testimony. For the first time, defense counsel had an opportunity to elaborate on why that testimony should be admitted. He opposed the government’s motion by explaining its relevance to establishing Johnson’s “subjective fear,” a frame of mind tending to justify self-defense. See id. This time, the court rejected the defense argument not because it was irrelevant but because, as the government by then was arguing, the proffered testimony — an “attempt to engender sympathy for the defendant” — would be “more prejudicial than probative.”
I suppose we can say that, in moving on to a probative/prejudicial analysis, the trial court implicitly assumed relevance of the proffered testimony, in contrast with its initial finding of irrelevance. Nonetheless, while explicitly referring to possible prejudice (jury sympathy), the court did not discernibly evaluate why testimony attempting to explain fear of rape had relatively little probative value. Apprehension of rape, triggered or intensified by a terrible, highly personal memory, would have added credence to Johnson’s assertion of fear; otherwise, that fear could only have been premised — as Johnson was allowed blandly to testify — on Myers’ putting “his private parts on my private part.” Indeed, the proffered testimony was all the more probative because subjective fear is central to a defendant’s right to an “imperfect” self-defense instruction, which the trial court gave in this case, requiring evidence of a defendant’s “actual belief both in the presence of danger and the need to resort to force.” Swann v. United States, 648 A.2d 928, 932 (D.C.1994). I cannot imagine, and the trial court offered no explanation, why Johnson’s proffered “mother-was raped” testimony would generate so much jury sympathy for this self-confessed, brutal killer that this testimony would prejudice the government more than it would add probative value to the most plausible explanation Johnson could offer for that brutality.
Viewing the entire record, I cannot be satisfied that the trial court adequately exercised discretion in finding more potential prejudice than probative value. Conceivably the court believed that Johnson’s memory of his mother’s rape was too attenuated to have probative value, but the court did not say so. To the contrary, especially because the court initially called the testimony “not relevant,” and because Johnson had learned about his mother’s rape only “about a year or so prior” — not years earlier — I do not believe we can assume that the court relied on attenuation as the basis for its ruling. Nor did the government or the court offer any other rationale. In the decision thirty years ago that governs our review of trial court discretion, we stressed that this court “should inquire whether the trial court’s reasoning is substantial and supports the trial court’s action.” Johnson, 398 A.2d at 365 (emphasis added). We must determine, more specifically, whether the trial court “failed to consider a relevant factor,” whether it “relied upon an improper factor,” and whether “the reasons given reasonably support the conclusion.” Id. (citation and internal quotation marks omitted). We also acknowledged that, “[i]f needs be,” we may “examine the record and infer the reasoning upon which the trial court made its determination.” Id. at 366. But here any such inference would have to go completely outside anything mentioned at trial or inferable from any colloquy between court and counsel. On this record, we cannot be sure what factors the trial court considered, let alone be satisfied that the court employed the “substantial” reasoning required. In short, we cannot tell *306why the court concluded that if the “mother-was-raped” testimony were admitted, the prejudice to the government from jury sympathy for Johnson would outweigh the probative value of that testimony in establishing Johnson’s fear of Myers. Accordingly, I cannot agree with the majority’s conclusion that the trial court did not “erroneously exercise[] its discretion in this case.” Ante at 299.
My colleagues find my error analysis unpersuasive because they perceive an additional factor relevant to — in fact controlling — the trial court’s exercise of discretion. That factor is the “excessive force” limitation on self-defense, which Chief Judge Washington incorporates into review of the trial court’s discretionary analysis. But, as noted earlier, neither the prosecutor — who had the burden of proof once Johnson claimed self-defense, see Rorie v. United States, 882 A.2d 763, 771 (D.C.2005) — nor the trial court ever mentioned excessive force. The majority’s approach, therefore, reads into the equation a component of trial court discretion that is not there. If excessive force has a place in our review, it belongs in an evaluation of harmlessness, not trial court error.1
II.
A trial court’s exercise of discretion in ruling out evidence based on a probative/prejudicial analysis can survive judicial review, even when the court has erred by fading to provide adequate reasoning, if the record clearly establishes that any error in excluding the evidence was harmless. See Johnson, 398 A.2d at 366-367. So, was the discretionary error here harmless? More specifically, because of Johnson’s many acts of brutality, did the “mother-was-raped” testimony lose its currency for the defense because trial counsel, in seeking to use that testimony to establish fear, did not also relate it to the excessiveness limitation on the use of deadly force in self-defense?
Before addressing that question we must resolve a procedural issue. The government, as noted earlier, did not raise excessive force at trial, and the trial court did not touch it. So is that issue properly before us now? The government has argued excessive force on appeal in response to Johnson’s contention on appeal that his “frenzied” mind, attributable to memory of his mother’s rape, excused all his brutal actions in killing Myers — a contention the defense did not make at trial. On appeal, therefore, Johnson has not been willing to rest his case on the probative value of his limited rationale proffered at trial, namely, that his memory caused or enhanced his fear. He has pressed us, somewhat differently, to premise our review on an alleged frenzy his “mother-was-raped” memory produced in him, causing not only fear but also wild frenzy. In short, Johnson has enhanced on appeal the argument he offered at trial. All things considered, therefore, I have no problem with our taking up the government’s corresponding argument that however real the frenzy may have been, it resulted in excessive force nullifying self-defense. Indeed, even *307if Johnson had not elevated “fear” to “frenzy,” the government could have argued the excessive force limitation (explained in the jury instructions) to undermine the “mother-was-raped” testimony.
Now to the merits. This court has noted that “the victim’s subjective perceptions are the prime determinant of the right to use force — and the degree of force required — in self defense.” Fersner v. United, States, 482 A.2d 387, 391-392 (D.C. 1984). But we also have stressed an important limitation: in determining the degree of force required — even when acting justifiably with deadly force in “the heat of passion” — the victim’s perceptions must “be reasonable under the circumstances.” Id. at 392. There are “degrees of deadly force,” id., some reasonable, others unreasonable, on the facts presented. In this case, even though Johnson was entitled to the “perfect” and “imperfect” self-defense instructions he requested and received, it is clear — there can be no doubt — that no reasonable jury, properly instructed as the jury was here, could have found that the degree of force Johnson used in defending himself against Myers was objectively “reasonable.” It was excessive as a matter of law. See id.
Accordingly, even if — contrary to the trial court — we were to grant Johnson’s contention that the proffered testimony was not only relevant but also more probative than prejudicial in showing his subjective fear of Myers, a jury’s finding of such fear could not have reasonably led to an acquittal based on the self-defense instructions the jury received. Even if Johnson had been allowed to testify that his brutal acts were attributable to his going berserk because Myers’s sexual touching had triggered the unbearable memory of his mother’s rape, that testimony — when added to all the other defense evidence admitted at trial — would not have been sufficient to convince a reasonable jury that Johnson’s multiple acts of brutality, viewed objectively, were reasonable, not excessive. See Alcindore v. United States, 818 A.2d 152, 157 (D.C.2003) (quoting Fersner, 482 A.2d at 391-392). Accordingly, because the “mother-was-raped” testimony would have offered little if any help to the defense on the required objective evaluation of excessive force (ultimately the controlling self-defense issue) — and further because the evidence tended strongly to show that Johnson could have escaped from the premises even if Myers was the aggressor threatening serious bodily harm — I agree that the trial court’s discretionary error was harmless. In the words of Johnson, it was not of a “magnitude requiring reversal.” Id. at 366.
III.
My theoretical difference from the approach my colleagues take may seem hy-pertechnical to some, but in reviewing for abuse of discretion we are tempted too often, I think, to merge error and harmlessness together so that review of alleged error is ignored if harmlessness can be found. The very power of discretion gives the trial court great leeway; we have recognized that when the court has “discretion,” it has a range of choice broad enough to be “wrong” up to a point without suffering reversal. Johnson, 398 A.2d at 367. Only when that exercise of discretion reaches a magnitude of error beyond that range do we overturn the ruling on appeal. Id. In this case, the trial court’s error did not reach that level of magnitude, but error there was. And given the broad range of choice that trial court discretion allows, it is important, I think, for all concerned to have error identified when error occurs.
. In reviewing the exercise of trial court discretion, we have articulated the harmless error component as follows: "[W]e must ask, having found error, is it of a magnitude to require reversal." Johnson, 398 A.2d at 366. "[T]he appellate court makes two distinct classes of inquiries when reviewing a trial court's exercise of discretion. It must determine, first, whether the exercise of discretion was in error and, if so, whether the impact of that error requires reversal. It is when both these inquiries are answered in the affirmative that we hold that the trial court "abused” its discretion.” Id. at 367. An abuse of discretion, therefore, is not merely an erroneous ruling; it is a harmful erroneous ruling-one that requires reversal.