JERRY E. SMITH, Circuit Judge:
This dispute arose as a consequence of a fatal commercial airline crash. We are asked to decide whether the measure of wrongful-death damages awarded to a decedent’s immediate family comports with Texas law; with two notable exceptions, we conclude that it is. We depart with the district court, first, in its assessment of lost inheritance, which, we conclude, erroneously incorporates a measure of double recovery. Second, we find the recovery for lost society and companionship secured by the minor children to be far in excess of what state law tolerates, and such a departure is not explained by any unique facts presented. Accordingly, we affirm in part, reverse in part, and remand.
I.
Michael Douglass was killed in a tragic airplane crash in 1985 at the Dallas/Fort Worth International Airport. His wife sued Delta Air Lines, Inc. (Delta), for wrongful-death damages on behalf of herself and three minor children. Delta agreed not to contest its liability in the disaster as long as the plaintiffs agreed to limit their claims to compensatory damages.
The issue of damages was tried to the court, which awarded the wife and three children (ages 11, 9, and 5 at the time of the accident) $6,901,822, plus $481,759 in prejudgment interest.1 There is little dispute concerning the strong family ties, the anguish that the survivors have suffered, [1338]*1338or the types of damages recoverable in this case: loss of future earnings, mental anguish on the part of the plaintiffs, loss of society and companionship, and loss of inheritance. There is considerable dispute, however, concerning the proper measure of these recoverable damages.
Delta believes that the trial judge erroneously framed the measure of recovery around certain “speculative” estimates offered by the plaintiffs' experts. Their expert on the subject of loss of future inheritance, for example, offered the court three separate growth strategies for the decedent’s then-current net worth, estimated at the time of his death in 1985 to be over $400,000. The various growth models offered at trial were derived from a 1983 Federal Reserve study of consumer finances in the United States that contained demographic data concerning the value of estates based upon age, wealth, and other factors. Further testimony sought to demonstrate that the decedent was an adept and aggressive investor who would have left his family, at his natural death, with a considerable estate.
The statistical average of the estimates for lost inheritance offered by the plaintiffs’ expert, in present value terms, was over $2,900,000. The court found this expert to be much more credible than Delta’s, who discredited himself by proffering testimony to the effect that the decedent would have accomplished nothing to increase his estate during the remainder of his natural life.
The court found the decedent to have been, in fact, a sophisticated saver and investor. However, without explaining its mathematical reasoning the court concluded that the evidence demonstrated that the decedent would have left, in present value terms, at least $1,130,207 to his wife and children. That figure reflects the least ambitious growth prediction of the decedent’s future estate offered by the plaintiffs’ expert.
By so finding, the court valued the plaintiffs’ lost inheritance in accordance with the most conservative growth scenario presented and rejected all objections raised by Delta to minimize the value of the estate. The court stated that there was no suggestion that the decedent would have squandered his estate or disinherited his family. Moreover, at age forty the decedent already had amassed a net worth of over $400,000 and owned several rental properties. Further, at the time of his death he was building a new family home in Florida.
Delta also contests the generous recovery for lost earnings arising from the decedent’s premature death ($3,531,615 in total, excluding prejudgment interest). Experts for the plaintiffs testified as to the decedent’s work history, the prospects for growth in the telecommunications industry in which he worked, likely advancements in employment based upon his reputation, and the documented advancement of his peers.
The experts offered the court three separate “future earnings” scenarios, complete with projected job promotions, dates for such promotions, and expected salaries.2 The three “career paths” offered a full range of possibilities: a conservative scenario of no job advancement for the dece[1339]*1339dent in career path I, as well as an ambitious career projection in path III, in which it was envisioned that the decedent eventually would have attained the management position of chief operating officer in a major telecommunications firm.
There is little dispute that the decedent was well educated, a former military officer and FBI special agent, and a gifted regional vice president for a telecommunications firm. Based upon the evidence presented, the court awarded a total of $3.5 million for pecuniary losses, or the average of the pecuniary losses estimated for career paths II and III. Declaring itself sensitive to speculative career and salary projections, the court nonetheless concluded that the evidence demonstrated rapid future advancement for the decedent and awarded lost earnings to the plaintiffs accordingly.
Lastly, Delta challenges the award of $725,000, in total, for “mental anguish” and $1,515,000 for “loss of society and companionship.” Delta argues that these intangible damages are far in excess of any that this circuit has been willing to tolerate under similar factual situations governed by Texas law. Accordingly, Delta seeks to have the district court’s substantial award reduced under this circuit’s “maximum recovery rule” to the more conservative level reflected in the caselaw authority that it cites for purposes of comparison.
II.
A.
Federal jurisdiction is premised herein upon diversity of citizenship. Sitting as an Erie court, we are bound to apply the substantive law of the forum state in this wrongful-death action. See Smith v. Industrial Constructors, Inc., 783 F.2d 1249, 1250 (5th Cir.1986); Budge v. Post, 643 F.2d 372, 375 (5th Cir. Unit A Apr. 1981). Neither party disputes that Texas law controls the measure of wrongful-death damages in this case.
District courts enjoy “wide discretion” in awarding damages. Wheat v. United States, 860 F.2d 1256, 1259 (5th Cir.1988); accord Wakefield v. United States, 765 F.2d 55, 59 (5th Cir.1985) (citing Hyde v. Chevron U.S.A., Inc., 697 F.2d 614, 632 (5th Cir.1983)). A trial court’s assessment of damages is a finding of fact, which we scrutinize for excessiveness under the clearly-erroneous standard. Wheat, 860 F.2d at 1259; Johnson v. Offshore Express, Inc., 845 F.2d 1347, 1356 (5th Cir.), cert. denied, - U.S. -, 109 S.Ct. 497, 102 L.Ed.2d 533 (1988). Under this standard, we compare damages awarded in factually similar cases, and arising within the controlling jurisdiction, in order to construct an objective framework for comparison. Wheat, 860 F.2d at 1259-60. Excessive awards that are entirely disproportionate to the injury sustained will not be upheld.3
However, we note that the “reassessment of damage awards is not an exact science; rather, it is inherently subjective, involving the interplay of experience, emotions and calculation.” Wheat, id. at 1259. A departure from precedent is merited if unique facts are present that are not reflected within the controlling caselaw. Id. at 1260.
Thus, we look to Texas authority, or those federal cases governed by Texas substantive law, in drawing an initial point of reference for the damages awarded in this case. Where departures from precedent [1340]*1340are evident, and such departures cannot be justified fairly by the novel facts presented by the case, remittitur of the trial court’s award, or a new trial on damages, is appropriate.4
B.
The court-made rule in Texas, only recently adopted, is that recovery for “loss of inheritance” is available in wrongful-death cases and is separate from loss of future earnings. In Yowell v. Piper Aircraft Corp., 703 S.W.2d 630, 633 (Tex.1986), for example, the court defined loss of inheritance “as the present value that the deceased, in reasonable probability, would have added to the estate and left at natural death to the statutory wrongful death beneficiaries but for the wrongful act causing the premature death.” Id. (emphasis added). The court flatly rejected the argument that assessing damages for lost inheritance is inherently too speculative 5 and held that the measure of lost future inheritance is for the trier of fact to decide.6 However, the court did not announce a precise mathematical formula to be applied.
Presumably, no recovery would be available if the factfinder were convinced that the decedent would have consumed his earnings entirely or would have outlived his statutory beneficiaries.7 However, damages are recoverable “where the wrongful death plaintiff shows that the decedent (1) would have enhanced his es-fate by some amount by saving some of his earnings or by prosperous management of his investments, and (2) would, in all reasonable probability, have left that amount upon his natural death to the plaintiff.” Moorhead, 828 F.2d at 290 (emphasis added).
In this case, the court expressly found that the decedent would have bequeathed his estate to the statutory beneficiaries (A will left his estate to his wife.) and that he was a “prudent and successful saver and investor.” It pointed to his $400,000 net worth at age forty and his extensive holdings in rental properties, despite his many years of government service.
Further, the court found Delta’s expert to be completely noncredible. For example, that expert suggested, to the court’s amazement, that had he lived, the decedent would not have accrued any additional assets. By comparison, the plaintiffs’ expert projected a future estate, in present-value terms, of between $1,130,000 and $4,850,-000 based upon demographic data compiled by the Federal Reserve in 1983. The court, although not compelled to do so, adopted the more conservative growth estimate of $1,130,000.8
On appeal, Delta suggests that by awarding that amount, the court concluded that nothing would have been added to the estate. It advances this proposition despite the fact that the district court found the complete opposite to be true and rejected [1341]*1341Delta’s expert as “straining credulity” in his assumptions and estimates for the future estate.
Essentially, Delta points to the $1,130,-000 recovery and asserts that this figure can be calculated by taking the decedent’s 1985 estate and simply investing it at 4.5 percent for the rest of his life expectancy. That being so, the argument advances, the court must have confused the principles of present and future value. Delta suggests that the $1,130,000 lost-inheritance award thus represents, in fact, the future value of the 1985 estate. This, we are told, constitutes double recovery because the plaintiffs have already received the benefit of the 1985 estate pursuant to the decedent’s will.
We are not impressed by Delta’s misconstruction of the district court’s findings of fact. That the award may be similarly calculated by compounding the value of the 1985 estate by a simple rate of interest is fortuitous at best, as the court expressly rejected the proposition that the decedent’s 1985 assets would remain static. It is evident from the court’s opinion that it enjoyed a solid understanding of the time value of money and that, in correctly applying these principles in the instant case, it did not confuse the concepts of present and future value. The judgment for lost future inheritance is fashioned in the court’s opinion in terms of present value, not future value.9
We depart from the district court’s analysis because it is evident that the court misapplied Yowell. Properly construed, Yowell mandates compensation for the loss suffered by the statutory wrongful-death beneficiaries as a consequence of the premature death of a talented and skillful benefactor. But the recoverable loss is limited to the advantage that the estate would have enjoyed from the decedent’s skillful management. Such skill must separate the decedent from the corps of average investors who, at least arguably, share the common attribute of being, in general, risk-averse and capable of earning only modest rates of return on their assets.
We understand Yowell to provide recovery for lost inheritance because, but for the decedent’s premature demise, the statutory beneficiaries would have enjoyed a larger estate to divide had the decedent lived to manage his assets. See generally Note, Yowell v. Piper Aircraft Corp.: Recovery of Lost Inheritance in Wrongful Death Actions, 38 Baylor L.Rev. 1023 (1986) (tracing doctrine’s evolution in Texas). Thus, the plaintiffs in this case are entitled to a recovery, in present value terms, that represents the portion of the future estate traceable to the decedent’s anticipated adroit management.10
The district court, however, ignored the fact that heirs with only average investment skill surely can manage $400,000 in assets in some respect and earn a modest rate of return. It does not follow, therefore, that the plaintiffs are entitled to the present value of the decedent’s entire future estate. The approach of the district court, we conclude, provides duplicative recovery not contemplated by Yowell. The Supreme Court of Texas in fact has stated that “[djamages should not overlap, and no double recovery should be allowed.” Moore v. Lillebo, 722 S.W.2d 683, 688 (Tex.1986).
To be clear, we have no quarrel with the court’s finding of fact that the decedent possessed remarkable financial skill. How[1342]*1342ever, recovery for lost inheritance in Texas does not compensate the heirs of an average investor since, presumably, such heirs could match or exceed the decedent’s investment skill with the property transferred upon death. Where the decedent demonstrates above-average skill, as evident in this case, the heirs are entitled to their lost advantage, reflecting the decedent’s financial sophistication.
As an example, assume that the decedent in this case could have been reasonably expected to manage his 1985 estate so as to earn, on average, a ten percent annual rate of return on the assets for the rest of his natural life. Assume further that by comparison, an average investor with no particular expertise could have invested that same estate in no-risk twenty- or thirty-year treasury instruments and earned, in today’s market, about seven percent. To the extent that the district court’s lost-inheritance award focuses only upon the decedent's likely rate of return with his assets, and ignores the basic investment opportunities of his beneficiaries in low-risk liquid assets, it is, we believe, at odds with the language of Yowell.
The Douglass heirs, who we assume have no particular investment skill, certainly can earn a modest rate of return with the decedent's estate through liquid, risk-free treasury instruments, for example. That being so, the beneficiaries’ damages should be limited to the premium, above the risk-free rate of return, that the decedent likely would have attained with his assets through personal management.
Empirically, the decedent’s advantage can be quantified under Yowell’s rationale by calculating, first, the future value of the decedent’s estate, had he lived to manage it. Thereafter, the future value of this same estate under the management of the heirs, maintaining a risk-free investment strategy for the decedent’s expected life, should be calculated. The difference between these future values is the advantage lost by the beneficiaries and owed to them in presently-valued dollars.11 It is nowhere evident in the record that the district court employed such an approach and, accordingly, we remand for a recalculation of lost-inheritance damages in light of this understanding of Yowell.
C.
Delta also contests the district court’s award for the loss of the decedent’s future earnings. Delta believes that the $3,531,615 awarded to the four plaintiffs is clearly erroneous and should be reduced to the $1,684,585 figure calculated by the plaintiffs’ expert in career path I. That scenario ignored future job promotions for the decedent for the rest of his professional life. Essentially, Delta asserts that the expert testimony concerning Michael Douglass’s future promotions and associated salary increases was entirely too speculative to merit damages.
We reject Delta’s challenge to this element of the award. The court awarded as damages the average of the lost earnings calculated pursuant to career paths II and III and thus regarded future promotions and salary increases to be highly likely to occur. It found the decedent to be an extremely talented professional with a proven record for rapid advancement that would surely continue into the future.
These findings are well supported by the record. The court also found Delta’s lone expert to be “unqualified” to speak upon certain matters relevant in assessing these [1343]*1343pecuniary damages (e.g., decedent’s future career progression) and to have reached conclusions that strained credulity (e.g., no promotions for the rest of decedent’s professional life, which, at age forty, had just begun in the private sector).
In contrast, the court found the experts for the plaintiffs to be more realistic, offering various scenarios and conservative economic forecasts derived from generally accepted economic literature. Significantly, their predictions are firmly anchored by the facts. The court did not, as Delta suggests, blindly adopt faulty expert opinion reminiscent of the experience in In re Air Crash at New Orleans, 795 F.2d 1230 (5th Cir.1986). There, we found certain expert testimony offered by the plaintiffs to be “without basis in the real world” and “completely airborn.” Id. at 1233. We lamented over the rise of the professional expert witness, paid to shape the damage figures as skillfully as any advocate, but maintaining claim before the trier of fact, misleadingly, to impartiality. Id. at 1234. Further, we cautioned trial courts to be wary of expert testimony based upon assumptions that are so outrageous as to transform the expert into an advocate of public policy reform. See id. at 1234-35.
Unlike Air Crash, this case presents a situation in which the expert for the defendant relied upon assumptions devoid of any basis in the real world. In a lengthy opinion, the district court here outlined its factual findings and concluded that the decedent likely would have progressed somewhere between career paths II and III with respect to future advancements and salary. These findings are well supported by the record.
The Douglasses argue in response that the district court properly included the decedent’s likely career advancement within the calculus to determine lost earnings, and they offer substantial caselaw authority to support this proposition.12 Further, contrary to Delta’s suggestion, future advancements need not be ascertainable with certainty to award recovery for lost earnings. Damages are limited by what the factfinder concludes the statutory beneficiaries, in all reasonable probability, would have received from the decedent’s earnings had he lived.13
D.
Texas has adopted the principle that the intangible injuries styled “mental anguish” and “loss of society and companionship” are distinct from one another and therefore merit separate recovery. Moore, 722 S.W.2d at 688. As a general principle, mental anguish compensates claimants for the harrowing experience resulting from the death of a loved one, whereas loss of society and companionship compensates the survivors for the positive noneconomic benefits that the decedent would have provided to his family, had he lived. See id. at 687-88.
The district court awarded the following intangible damages to each plaintiff;
Surviving Spouse Surviving Children Oldest Middle Youngest
Mental anguish $500,000 $100,000 $75,000 $50,000
Loss of society and companion- $500,000 ship _ $270,000 $325,000 $420,000
Total $1,000,000 $370,000 $400,000 $470,000
[1344]*1344For purposes of appellate review, we look to factually similar Texas cases as guidance in determining whether the trial court’s awards for intangible damages are clearly erroneous. See Wheat, 860 F.2d at 1259-60. Such similar cases provide a helpful framework for general comparison, but do not supply any legal standard. Id.) Johnson Offshore Express, 845 F.2d at 1356. Further, we apply the loosely defined “maximum recovery rule” when deciding whether a remittitur is in order. This judge-made rule essentially provides that we will decline to reduce damages where the amount awarded is not disproportionate to at least one factually similar case from the relevant jurisdiction.14
Delta argues that intangible damages should be reduced to $400,000-600,000 for the wife and $200,000-300,000 for each minor child. It cites a number of cases in which lesser awards were granted under Texas law. However, it curiously fails to cite all relevant cases and, in particular, one case arising from this same airplane tragedy in Dallas.
In Larsen v. Delta Air Lines, Inc., 692 F.Supp. 714, 721 (S.D.Tex.1988), a young childless widow was awarded $500,000 for mental anguish and $500,000 for loss of society and companionship. This compares with the recovery that the wife received in this case and, thus, the maximum recovery rule does not operate in Delta’s favor to reduce the intangible damages as to her.15
Delta dismisses its obligation to cite all relevant authority, including adverse caselaw, under the theory that Larsen was never appealed to the Fifth Circuit. That is, until awards are reviewed by the courts of appeal, they are legally insignificant for purposes of the maximum recovery rule. We dismiss this argument as frivolous and reaffirm the ethical obligation that all relevant state wrongful-death cases, and all such federal cases turning upon state substantive law, must be presented to the reviewing court where an award is challenged for excessiveness.16 Awards imposed by district courts do not, as Delta suggests, serve as precedent only upon review.
With little difficulty, we dismiss the proposition that Larsen is irrelevant merely because it is a district court case, not reviewed on appeal. We find no benefit in fashioning a rule that would allow Delta to withhold appellate review of generous awards, as it did in Larsen, and thereby avoid an adverse judgment’s precedential value in subsequent cases such as this. Accordingly, we conclude that the amount of intangible damages awarded to the widow here comports with the maximum recovery rule and Texas law in light of the precedent provided by Larsen.
The intangible damages secured in favor of the minor children are more proble[1345]*1345matic. The recovery for mental anguish by the children (ranging from $50,000 to $100,-000), as a consequence of the sudden and tragic loss of their father, is certainly supported by the caselaw and by the evidence demonstrating that the children suffered considerable trauma. In Grandstaff v. City of Borger, 846 F.2d 1016 (5th Cir.1988), for example, a minor child recovered $250,000, in total, for loss of society and mental anguish when her father was killed by police under the mistaken belief that he was a fugitive. While Grandstaff failed to separate the recoveries for mental anguish and lost society, this $250,000 aggregate judgment supports the considerably lower individual recoveries for mental anguish suffered by the Douglass children here.
However, we depart with the district court in its assessment of lost society and companionship for the minor children. The court awarded damages that dwarf the recovery for mental suffering. While the damages for mental suffering and lost society need not be precisely in balance, huge variations may indicate excessiveness. In this case, the daughter’s recovery for lost society and companionship was over eight times her recovery for mental anguish.
We accept as true the court’s findings of strong family bonds and the heightened sensitivities of the young children. Nevertheless, there are no factually similar Texas cases to be found awarding such generous relief — ranging in this case from $270,-000 for the oldest son to $420,000 for the daughter — for the paternal society and companionship forever lost.
The plaintiffs direct us to Wheat as being dispositive, but that case involved a minor child who witnessed her mother’s painful degeneration with cancer for over four years. Such facts do not compare with the sudden demise of a parent, not witnessed by the minor plaintiffs. Other factually distinguishable cases are cited by both parties, none of which lends support to either side.17 Thus, as in Wheat, we lack the benefit of a closely analogous case. See 860 F.2d at 1262. Grandstaff is, helpful, but it aggregates the intangible damages with respect to the minor child.18
The minor children’s recovery for lost society and companionship is designed to recoup, as best the trier of fact can determine, the loss of positive benefits flowing from the love, comfort, companionship, and society that they, in all reasonable probability, would have enjoyed had their father lived. See Moore, 722 S.W.2d at 687-88. This loss is difficult, if not impossible, to quantify precisely, and we are not qualified to substitute our subjective judgment for that of the district court. Thus, we hold only that such relief for this intangible injury far exceeds any prior judgment in Texas, and we find no novel facts justifying such a grand departure.
Looking to Grandstaff, in which a minor daughter lost her father tragically and recouped $250,000 in intangible damages, and applying the maximum recovery rule to that case, i.e., a 33% increase from the prior maximum recovery in Texas, the loss of society and companionship for the minor children here cannot far exceed $200,000. This figure is not set in concrete; it is offered as guidance only so that the aggregate sum for intangible damages with respect to each minor child does not exceed the $300,000 threshold, which, to date, is the most Texas law might be expected to allow. Accordingly, a remittitur is ordered such that the total intangible damages for the three minor children do not exceed $300,000 for each child. See 6A J. Moore, J. Lucas & G. Grotheer, Jr., Moore’s Federal Practice ¶ 59.08[8] at 59-211 to 59-212 (2d ed. 1989).
[1346]*1346The judgment of the district court is thus AFFIRMED in part, REVERSED in part, and REMANDED for proceedings consistent with this opinion.