Opinion for the Court filed by Circuit Judge BROWN.

BROWN, Circuit Judge:

This case features two natural gas pipeline companies, Petal Gas Storage, L.L.C., and High Island Offshore System, L.L.C., that challenge ratemaking orders from the Federal Energy Regulatory Commission. The chief issue — the only one Petal and *698 HIOS share in common, and the one that brings the Interstate Natural Gas Association of America (representing most of North America’s natural gas transportation companies) into this case as interve-nor — is whether the Commission erred in its selection of the “proxy groups” used to calculate petitioners’ gas transmission rates, along with its placement of petitioners within those proxy groups. We hold that the Commission did err, by failing to explain how its proxy group arrangements were based on the principle of relative risk. In addition, HIOS presents three claims of its own, arguing the Commission should have approved the settlement it presented; selected a faster depreciation rate for its pipeline system; and awarded it a higher management fee. We reject all three claims, concluding that the Commission was well within the considerable deference we show it in ratemaking cases.

We begin with the question of whether the Commission erred in its selection of proxy groups and placement of petitioners within those proxy groups. The Administrative Procedure Act’s arbitrary and capricious standard governs our review, entitling the Commission to substantial deference, particularly in the rate-making context, E. Ky. Power Coop. v. FERC, 489 F.3d 1299, 1304 (D.C.Cir.2007) (citing Administrative Procedure Act, 5 U.S.C. § 706(2)(A)), but also imposing on it a duty of reasoned decisionmaking, Nat'l Fuel Gas Supply Corp. v. FERC, 468 F.3d 831, 839 (D.C.Cir.2006), and requiring that “we ... reverse a decision that departs from established precedent without a reasoned explanation,” Exxon Mobil Corp. v. FERC, 315 F.3d 306, 309 (D.C.Cir.2003).

The Commission has a duty under § 4 of the Natural Gas Act to ensure “just and reasonable” rates in the natural gas industry. 15 U.S.C. § 717c. Proxy groups are a tool the Commission uses, by policy, to determine just and reasonable rates. Since the Supreme Court has held that rates “should be sufficient to assure confidence in the financial integrity of the enterprise, so as to maintain its credit and to attract capital,” FPC v. Hope Natural Gas Co., 320 U.S. 591, 603, 64 S.Ct. 281, 88 L.Ed. 333 (1944), rates are typically based on a pipeline’s costs, including the cost of capital, Canadian Ass’n of Petroleum Producers v. FERC, 254 F.3d 289, 293 (D.C.Cir.2001) (CAPP I). As the cost of equity capital for a private natural gas company cannot be read from the market, the Commission estimates this figure based on a proxy group of publicly traded, but otherwise comparable, natural gas companies. Id. at 293-94.

For natural gas pipelines, the Commission has traditionally relied on a proxy group of publicly traded companies with a high proportion of their business in pipeline operations. Transcon. Gas Pipe Line, 90 F.E.R.C. ¶ 61,279, at 61,933 (2000). But the industry is changing. Acquisitions, financial mishaps, and other factors have left, by one count, just three companies that fit the old requirements (too few for a proxy group), Williston Basin Interstate Pipeline Co., 104 F.E.R.C. ¶ 61,036, at 61,103 P 35 (2003), and all parties to this case agree the Commission’s traditional approach must change. See High Island Offshore Sys., L.L.C., 112 F.E.R.C. ¶ 61,-050, at 61,356 P 56 (2005) (“[A]ll the parties agree ... only one corporation ... meets our historical proxy group standards and need not be excluded for other reasons.”). Controversy about how it should change has been bubbling up in a number of recent cases, see, e.g., Kern River Gas Transmission Co., 117 F.E.R.C. ¶ 61,077, at 61,342-49 PP 123-59 (2006); Williston Basin, 104 F.E.R.C. ¶ 61,036, at 61,103-04 *699 PP 34-43, but this case seems to represent an arrival point of sorts for the Commission, see Kern River, 117 F.E.R.C. ¶ 61,-077, at 61,345 P 138 (reversing an administrative law judge for deviating from the HIOS proxy group). The instant case is also this circuit’s first opportunity to weigh in on the issue. Cf. Williston Basin Interstate Pipeline Co. v. FERC, 475 F.3d 330, 333 (D.C.Cir.2006) (dismissing without reaching the proxy group issue); Canadian Ass’n of Petroleum Producers v. FERC, 308 F.3d 11, 14-16 (D.C.Cir.2002) (CAPP II) (finding no jurisdiction to consider the proxy group issue).

Petal and HIOS dispute the design and implementation of the Commission’s proxy groups in their cases on three grounds. First, they claim the Commission improperly included in their proxy groups low-risk, diversified natural gas companies with most of their business in distribution rather than pipelines — essentially LDCs (local gas distribution companies). Second, they claim the Commission improperly excluded risk-comparable MLPs (master limited partnerships) with a high proportion of their business in natural gas pipelines. Finally, given a proxy group of companies mainly engaged in gas distribution, Petal and HIOS object to the Commission placing them in the middle rather than the high end of the range of returns.

That proxy group arrangements must be risk-appropriate is the common theme in each argument. The principle is well-established. See Hope Natural Gas Co., 320 U.S. at 603, 64 S.Ct. 281 (“[T]he return to the equity owner should be commensurate with returns on investments in other enterprises having corresponding risks.”); CAPP I, 254 F.3d at 293 (“[A] utility must offer a risk-adjusted expected rate of return sufficient to attract investors.”). The principle captures what proxy groups do, namely, provide market-determined stock and dividend figures from public companies comparable to a target company for which those figures are unavailable. CAPP I, 254 F.3d at 293-94. Market-determined stock figures reflect a company’s risk level and, when combined with dividend values, permit calculation of the “risk-adjusted expected rate of return sufficient to attract investors.” Id. at 293; see also Williston Basin Interstate Pipeline Co. v. FERC, 165 F.3d 54, 57 (D.C.Cir. 1999). The Commission itself has recognized the principle (and in quite a pointed way) by excluding, until recently, gas distribution companies from gas pipeline proxy groups, sometimes expressly for risk-based reasons. See, e.g., Trailblazer Pipeline Co., 106 F.E.R.C. ¶ 63,005, at 65,-037 P 94 (2004) (rejecting inclusion of LDCs in a proxy group “because they face less risk than a pipeline”); Wyo. Interstate Co., 96 F.E.R.C. ¶ 63,040, at 65,262-63 (2001) (“The flaw in BP’s position is that the proxy group advocated by it is heavily influenced by several distribution’ companies that are less risky than a transmission company such as WIC.”); Mountain Fuel Res., Inc., 28 F.E.R.C. ¶ 61,195, at 61,370 (1984) (“[T]he distribution companies ... are not really comparable to the level of risk involved in Resources’ operations.”).

On the record before us, we do not find adequate support for the contention that the Commission’s proxy group arrangements were risk-appropriate. In its HIOS ruling, the Commission states that changes in the gas pipeline industry compel a new approach to proxy groups. High Island Offshore Sys., L.L.C., 110 F.E.R.C. ¶ 61,043, at 61,157-58 PP 131-32 (2005). We accept this. The Commission rejects the use of MLPs on the ground that they issue distributions rather than dividends (and thus might provide returns of equity as well as returns on equity), and justifies the use of what are primarily gas *700 distribution companies as “the best [option] available ... on the current record.” Id. at 61,157 P 131. While these propositions are not self-evident, we accept them for the sake of argument. Still, nothing in the Commission’s explanation tells us why the companies selected are risk-comparable to HIOS. (Put differently, when the goal is a proxy group of comparable companies, it is not clear natural gas companies with highly different risk profiles should be regarded as comparable.) The Commission does address the issue of relative risk when it places HIOS in the middle of the proxy group in terms of return on equity. But in doing so, the Commission expressly relies on the “assumption that pipelines generally fall into a broad range of average risk ... as compared to other pipelines” — an assumption that is decisive only given a proxy group composed of other pipelines. Id. at 61,161 P 154. If gas distribution companies generally face lower risks than gas pipeline companies (as seems likely), a risk-appropriate placement would be at the high end of the group. Finally, as to the Commission’s ruling in Petal’s case (which involved a somewhat different proxy group from the one later used with HIOS), the same analysis applies: The Commission included gas distribution companies without an adequate analysis of relative risk, and placed Petal in the middle of the proxy group as if it were comparing Petal’s risk profile to other gas pipeline companies. See Petal Gas Storage, L.L.C., 106 F.E.R.C. ¶ 61,325, at 62,282-83 PP 26-30 (2004).

We therefore vacate the Commission’s orders with respect to the proxy group issue. On remand, we do not require any particular proxy group arrangement. Perhaps it would be best to include gas distribution companies and exclude MLPs, but to put petitioners’ rates of return on equity at the top of the range. Or perhaps including MLPs and excluding gas distribution companies, while putting Petal and HIOS in the middle of the range, would be best. The Commission might even acceptably return to this court with just the same arrangements it has chosen, albeit explained and justified in very different terms. What matters is that the overall proxy group arrangement makes sense in terms of relative risk and, even more importantly, in terms of the statutory command to set “just and reasonable” rates, 15 U.S.C. § 717c, that are “commensurate with returns on investments in other enterprises having corresponding risks” and “sufficient to assure confidence in the financial integrity of the enterprise ... [and] maintain its credit and ... attract capital,” Hope Natural Gas Co., 320 U.S. at 603, 64 S.Ct. 281.

We turn now to HIOS’s three individual claims. In 2002, HIOS sought to raise its shipping rates. An administrative law judge rejected the rate increase in 2004, and HIOS appealed. While the appeal was pending, HIOS worked out a settlement with at least some of its shippers and filed the settlement with the Commission for approval. The Commission rejected the settlement and, on the merits, depreciated HIOS’s pipeline system at a slower rate than HIOS favored, awarded HIOS a smaller management fee than HIOS requested, and, of course, used a proxy group to determine HIOS’s rate of return on equity. Besides the proxy group argument (addressed above), HIOS claims the Commission should have accepted its settlement proposal. Failing that, HIOS disputes the Commission’s choice of depreciation rate and management fee. As before, the arbitrary and capricious standard governs our review.

*701 A

The Commission’s stringent, merits-based rejection of its uncontested settlement, HIOS argues, violated FERC’s settlement regulations and departed without explanation from the precedent of Stingray Pipeline Co., 101 F.E.R.C. ¶ 61,365 (2002). In the proceedings below, some controversy swirled over whether to regard the settlement as uncontested (as HIOS contended it should be). But we need not join the fray: The opinion under our review for this issue proceeds on the “assumption that] the settlement may be treated as uncontested,” High Island Offshore Sys., 110 F.E.R.C. ¶ 61,043, at 61,140 P. 25; FERC’s brief never contends otherwise; and we, proceeding on the same assumption, affirm.

Under FERC’s regulations for uncontested settlements, the Commission “may” give a settlement its approval “upon a finding that the settlement appears to be fair and reasonable and in the public interest.” 18 C.F.R. § 385.602(g)(3). The implication—as we have stated in dicta before and affirm now—is that the Commission may adopt an uncontested settlement only after finding it “fair and reasonable and in the public interest”; that is, the Commission has a duty to disapprove uncontested settlements that are unfair, unreasonable, or against the public interest. Mobil Oil Corp. v. FPC, 417 U.S. 283, 314, 94 S.Ct. 2328, 41 L.Ed.2d 72 (1974) (“If a [settlement] proposal enjoys unanimous support ..., it could certainly be adopted ... if approved in the general interest of the public.” (emphasis added) (internal quotation marks omitted)); NorAm Gas Transmission Co. v. FERC, 148 F.3d 1158, 1165 (D.C.Cir.1998) (“Even if ... customers had unanimously supported the proposed settlement, the Commission would still have the responsibility to make an independent judgment as to whether the settlement is ‘fair and reasonable and in the public interest.’ ”); Tejas Power Corp. v. FERC, 908 F.2d 998, 1003 (D.C.Cir.1990) (“Commission may approve uncontested settlement only upon a finding that the settlement appears to be fair and reasonable and in the public interest.” (internal quotation marks omitted)).

Here, the Commission rejected the settlement for two reasons. First, it dictated rates half again as high as the administrative law judge had approved. Second, it awarded a three-million-dollar payout from HIOS to the active parties to the settlement—a payout that the Commission concluded would, in the circumstances of this case, undermine the usual assumption that a settlement’s active parties will protect the interests of its inactive parties. HIOS complains this level of scrutiny goes to the merits in a way properly reserved for contested settlements, 18 C.F.R. § 385.602(h)(l)(i), but we see only the independent consideration of fairness, reasonableness, and the public interest the Commission is duty-bound to give. HIOS claims the Commission approved a very similar settlement proposal in Stingray. But in Stingray, the rates dictated in the settlement were lower than those the administrative law judge had approved. True, Stingray featured a payout provision like the one here. Faced with the discrepancy, the Commission explains that “upon further reflection” it is “increasingly concerned about the unduly discriminatory nature of such arrangements.” High Island Offshore Sys., 110 F.E.R.C. ¶ 61,043, at 61,142 P 33. In our view, so slight a change of course, for so ample and well-established a reason, does not qualify as “departing] from established precedent without a reasoned explanation.” Exxon Mobil, 315 F.3d at 309.

*702 B

Determining a gas pipeline’s depreciation rate requires forecasting “the probable useful life of the specific pipeline systems in question,” based both on wear and tear and on the exhaustion of natural resources. Memphis Light, Gas & Water Div. v. FPC, 504 F.2d 225, 232 (D.C.Cir. 1974). In this case, we have a clash of experts. HIOS’s expert, relying on certain specific reserves in the Gulf of Mexico (where HIOS operates), gave the pipeline a reserve life of 10 years. FERC’s staff witness, basing his estimate on a large swathe of the Gulf called the Western Planning Area, gave a reserve life of 17.5 years. The issue for the Commission was which of these two estimates and experts to believe.

That is not, however, the issue facing us. Our question is whether the Commission’s choice was rational or arbitrary. The context gives further shape to our arbitrary and capricious standard of review. First, “[t]he finding of the Commission as to the facts, if supported by substantial evidence, shall be conclusive.” Natural Gas Act § 19(b), 15 U.S.C. § 717r(b). Second, where an agency is “making predictions, within its area of special expertise, at the frontiers of science ... as opposed to simple findings of fact, a reviewing court must generally be at its most deferential.” Balt. Gas & Elec. Co. v. Natural Res. Defense Council, Inc., 462 U.S. 87, 103, 103 S.Ct. 2246, 76 L.Ed.2d 437 (1983). Third, where expert witnesses dispute a factual issue “ ‘the resolution of which implicates substantial agency expertise,’ ” our role is only to verify that the agency has “relied upon sufficient expert evidence to establish a rational connection between the facts and the choice made.” Wis. Valley Improvement Co. v. FERC, 236 F.3d 738, 746, 747 (D.C.Cir.2001) (internal quotation marks omitted) (quoting Marsh v. Or. Natural Res. Council, 490 U.S. 360, 376, 109 S.Ct. 1851, 104 L.Ed.2d 377 (1989)). Here, the FERC staff expert, relying on data from multiple sources, cited four facts in support of his conclusion that there is “a significant amount of undiscovered gas remaining in HIOS’[s] supply area”: an independent source’s estimates, historical discoveries of new reserves, substantial exploratory drilling in HIOS’s supply area, and proposed incentives for further explorations. HIOS makes much of the expert’s “area-wide approach to determining] the total reserves,” but at no point, on our reading of the testimony, does the expert actually assume that HIOS can access all the gas in the Western Planning Area. Rather, he takes “representative” samples from the whole area to give an indication of what HIOS can expect from the reserves it can access. The work shows every sign of care, and the Commission’s reliance on it falls well within our norms of deference.

Typically, a pipeline receives a return on its net investment, which is measured by the pipeline’s rate base. In the anomalous situation where a pipeline’s assets have become fully depreciated, FERC policy provides that the pipeline may receive a “management fee” in lieu of such a return. See Tarpon Transmission Co., 57 F.E.R.C. ¶ 61,371, at 62,240 (1991). Here, both HIOS and the Commission agree that HIOS should be awarded a management fee on the Tarpon model, and agree that the Tarpon formula strictly applied would produce an unreasonably low result. The issue for the Commission, then, was how to appropriately modify the Tarpon formula. The issue for us — again, at the apex of our deference under Baltimore Gas, 462 U.S. at 103, 103 S.Ct. 2246 — is whether the Commission’s modification was arbitrary.

*703 HIOS argues its approach to modifying the Tarpon formula is “more reasonable” and “straightforward” than the Commission’s, and would “best match[] the return that the fee is designed to substitute for.” Petitioners’ Reply Br. 31-32. But FERC is not required to choose the best solution, only a reasonable one. See Deaf Smith County Grain Processors, Inc. v. Glickman, 162 F.3d 1206, 1215 (D.C.Cir. 1998) (explaining the arbitrary and capricious standard). HIOS claims its own method is better, but provides no evidence that FERC’s approach is unreasonable. Therefore, even if true, HIOS’s claims would not invalidate the Commission’s action.

For the reasons above, we vacate and remand on the issue of the proxy groups, and otherwise affirm.

So ordered.