This petition arises out of a proceeding before the Federal Energy Regulatory Commission to set rates for the transmission of electricity over lines operated by a regional transmission organization. Over *55a century ago, the first Justice Harlan noted that regulated rates must ensure just compensation, but confessed that “[h]ow such compensation may be ascertained, and what are the necessary elements in such an inquiry, will always be an embarrassing question.” Smyth v. Ames, 169 U.S. 466, 546, 18 S.Ct. 418, 42 L.Ed. 819 (1898) (quoted in Duquesne Light Co. v. Barasch, 488 U.S. 299, 308, 109 S.Ct. 609, 102 L.E.2d 646 (1989)). For our part, we have recognized that “agency ratemaking is far from an exact science,” Time Warner Entm’t Co. v. FCC, 56 F.3d 151, 163 (D.C.Cir.1996), and that it involves “complex industry analyses,” Ass’n of Oil Pipe Lines v. FERC, 83 F.3d 1424, 1431 (D.C.Cir.1996), and “[¡Issues of rate design [that] are fairly technical,” Town of Norwood v. FERC, 962 F.2d 20, 22 (D.C.Cir.1992). For these reasons, and because ratemaking “involves policy determinations in which the agency is acknowledged to have expertise, our review thereof is particularly deferential.” Time Warner, 56 F.3d at 163 (internal quotation marks omitted).
Given the deferential standard, we uphold FERC’s decisions to calculate the pertinent rate of return on equity in this case by reference to a particular “proxy group” of publicly-traded companies, and to base the rate of return on the midpoint, rather than the median or mean, of the rates in that group. But FERC is entitled to deference only if it plays fair, and we conclude that the Commission failed to give adequate notice that it would add 50 basis points to the rate of return generated by its calculations, to encourage participation in regional transmission organizations. We accordingly grant the petition in part.
I.
Midwest Independent Transmission System Operator, Inc. (MISO) is a regional transmission organization (RTO) — a company that combines multiple power grids into a single transmission system. In recent years, FERC has promoted the formation of RTOs as a means of increasing competition and driving down the price of electricity. According to the Commission, RTOs provide a large and stable transmission system that reduces regional pricing disparities and creates an efficient market for new power generators. See generally Regional Transmission Organizations, Order No.2000, 65 Fed.Reg. 809 (Dec. 20, 1999); Order No.2000-A, 65 Fed.Reg. 12,088 (Feb. 25, 2000). MISO, the first such organization in the nation, came into being when a series of midwestern utilities placed their grids under its centralized control. See Midwest ISO Transmission Owners, Inc. v. FERC, 373 F.3d 1361, 1365 (D.C.Cir.2004).
The rates charged by electric utilities such as MISO are regulated by FERC to ensure that they are just and reasonable, and not unduly discriminatory. See 16 U.S.C. §§ 824d, 824e. Utilities themselves initiate the ratemaking process by submitting proposals to the Commission, but FERC retains authority to modify such proposals to ensure compliance with the statutory standards. Id. §§ 824d(c)-(d), 824e(a).
A major component of the rates charged by MISO is the return on equity (ROE) paid to its member utilities. This rate compensates the utilities for the capital cost of the grids they placed under MISO’s control. FERC derives the rate by estimating the annual return an equity investor in the utility would expect on such capital, had the utility continued to operate the grid outside the RTO. See generally Canadian Ass’n of Petroleum Producers v. FERC, 254 F.3d 289, 293-94 (D.C.Cir.2001). Calculating this rate would be relatively easy if a utility’s interest in its grid — its business as a transmission owner *56(TO) — were publicly traded, but “there are no publicly traded independent pure electric transmission companies.” MISO Initial Decision, 99 FERC ¶ 63,011, 65,040, 2002 WL 32056864 (2002). The Commission must therefore resort to more roundabout estimations.
In December 2001, MISO and certain of its member TOs petitioned the Commission to increase the ROE component of MISO’s charges from a previously approved level of 10.5 percent to 13 percent. The Commission set the matter for hearing, at which all interested parties were allowed to present evidence. Among those availing themselves of this opportunity were the petitioners in this case — the Public Service Commission of the Commonwealth of Kentucky (PSCKY) and a group of private consumers and municipal entities (the Intervenor Group) — who appeared on behalf of ratepayers and argued against any rate increase or, in any event, for a more modest one.1 At the hearing, an administrative law judge selected a proxy group of public companies to use in estimating the appropriate return on equity for the MISO TOs. The judge chose a group consisting of the parent companies of certain MISO TOs themselves — which, unlike the transmission-owning subsidiaries, are publicly traded. The judge rejected several other proposals, including one submitted by the Intervenor Group. Id. at 65,038-42.
Once she had made her choice, the judge sought to extract a single ROE value, representative of the proxy group as a whole, to be applied to all the MISO TOs. She chose the midpoint of the range — the average of the highest and lowest data points— which yielded a return of 12.38 percent. She rejected the recommendation of FERC staff to use the mean — the average of all values in the proxy group (11.74 percent) — as well as the Intervenor Group’s proposal to use the median — the middle data point in the group (11.85 percent). Id. at 65,047-52. The judge also rejected competing pleas from MISO and the Intervenor Group to derive ROE using only part of the proxy range, such as the top or bottom half. Id.
FERC affirmed the ALJ’s choice of the proxy group and her use of the midpoint. Speaking to the latter, the Commission noted that its electrical industry precedents — unlike its oil and gas cases — had relied on the midpoint as a measure of central tendency. The Commission, however, also decided sua sponte to increase the final return by 50 basis points, to 12.88 percent, as an incentive for companies to join a regional transmission organization. MISO Order Affirming Initial Decision, 100 FERC ¶ 61,292, 62,313-15, 2002 WL 31986513 (2002). FERC explained that it “will be clarifying [this] incentive rate policy in the near future.” Id. at 62,315. Several months later, the Commission indeed issued a proposal to give “any entity that transfers operational control of transmission facilities to a Commission-approved RTO ... an incentive adder of 50 basis points on its ROE for all such facilities transferred.” Proposed Pricing Policy for Efficient Operation and Expansion of Transmission Grid, 102 FERC ¶ 61,032, 61,065, 2003 WL 245747 (2003).
In the meantime, petitioners sought rehearing of FERC’s order, which the Commission denied. Besides affirming its earlier findings, FERC rejected petitioners’ contention that they had not been given sufficient notice of the possibility of an *57incentive-based adder. FERC explained that petitioners should have been aware of the possibility of such an adder, given the Commission’s statutory power to amend proposals to ensure just and reasonable rates. FERC also noted that the final value of 12.88 percent was less than the 13 percent initially requested by MISO, and thus presumably within the range of petitioners’ expectations at the outset of the proceedings. See MISO Order Denying Rehearing, 102 FERC ¶ 61,143, 61,395, 2003 WL 257395 (2003).
Petitioners then sought review in this court, but the Commission moved for a voluntary remand to allow further consideration of the rate of return. On remand, the Commission maintained its position, but offered a different rationale for the use of the midpoint in this case. FERC explained that the question here “is not what constitutes the best overall method for determining ROE generically (i.e. the midpoint versus the median or mean).” MISO Order on Remand, 106 FERC ¶ 61,302, 62,192, 2004 WL 598168 (2004). Rather, the Commission observed that under the “unique ... circumstances” of this case, in which the chosen value will apply “to a diverse group of companies,” the midpoint provides the best measure because it emphasizes the endpoints of the proxy group range, ensuring that outlier as well as average TOs receive just and fair compensation. Id. at 62,192-93. The Commission also found that the proxy group was not too skewed to permit a midpoint analysis. Id. at 62,193.
PSCKY and the Intervenor Group now seek review of the FERC orders, challenging the selection of the proxy group, the use of the midpoint, and the incentive-based adder.
II.
The “arbitrary and capricious” standard of the Administrative Procedure Act governs our review. See 5 U.S.C. § 706(2)(A). Under this standard, FERC must consider relevant data and “articulate a rational connection between the facts found and the choices made.” Williston Basin Interstate Pipeline Co. v. FERC, 165 F.3d 54, 60 (D.C.Cir.1999) (citations omitted). The Commission must also respond meaningfully to the arguments raised before it. See Canadian Ass’n of Petroleum Producers, 254 F.3d at 299. The Commission’s factual findings are conclusive if supported by substantial evidence. See 16 U.S.C. § 8251 (b).
A. Petitioners first question the proxy group chosen to represent the MISO TOs’ expected return on equity. They assert that the ALJ improperly dismissed the Intervenor Group’s proposal, which would have considered (and heavily weighted) four generation-divested electric utilities in addition to the companies ultimately chosen. They also maintain that the Commission failed to engage their petition on this issue, and erroneously saddled them with the burden of proof in the group selection process. We find these arguments unavailing.
The judge chose a group submitted by MISO, consisting of the publicly-owned parent companies of the TOs themselves. She reasoned that “[t]his is the best proxy group since it involves companies that are currently in the MISO; the group includes comparable risk companies, similar in business profiles and size.” Initial Decision, 99 FERC at 65,041. She also relied on a prior FERC decision that approved a proxy group in part because that group contained parent companies of the utility whose rates were being set. Id. (citing Southern California Edison, 92 FERC ¶ 61,070, 2000 WL 1100260 (2000)). She dismissed the Intervenor Group’s proposal because “it consisted of distribution companies, not transmission companies.” Id. *58at 65,042. FERC summarily affirmed these findings. Order Affirming Initial Decision 100 FERC at 62,312.
Petitioners provide us with insufficient reason for overturning the selection of the proxy group. At the hearing before the ALJ, the Intervenor Group’s own expert conceded that MISO’s proposal was one of “the two most appropriate groups to use in establishing the range” and that “[t]he TOs or their parent companies have rates at issue in this proceeding and are therefore an appropriate group to use as a reasonable starting point.” Testimony of J. Bertram Solomon at 67. On appeal, petitioners note that the parents’ businesses extend well beyond transmission, and maintain that the judge underestimated the transmission component of the additional companies in their own proposal.
Given the deferential standard of review, this is just nibbling at the margins. The Supreme Court explained long ago that ratemaking is a pragmatic exercise, see FPC v. Hope Natural Gas Co., 320 U.S. 591, 602, 64 S.Ct. 281, 88 L.Ed. 333 (1944), and more than second-guessing close judgment calls is required to show that a rate order is arbitrary and capricious. Id. FERC chose reasonably from the options presented, based on the testimony offered and the Commission’s own precedent.
Petitioners’ remaining challenges to the proxy group also fall short of the mark. Their contention that FERC failed to engage their petition finds no support in our case law. While we have held that FERC may not ignore an argument presented to it, our cases addressed situations where neither the Commission nor an ALJ addressed the issue. See Canadian Ass’n of Petroleum Producers, 254 F.3d at 298-99; NorAm Gas Transmission Co. v. FERC, 148 F.3d 1158, 1165 (D.C.Cir.1998). Here, by contrast, FERC reviewed the ALJ’s extensive discussion of proxy groups and expressly adopted her rejection of petitioners’ proposal. See Order Affirming Initial Decision, 100 FERC at 62,312 (“For the reasons set forth in the Initial Decisions, we find that the Intervenor Group has failed to set forth convincing evidence [regarding its proposal].”).
As for the burden of proof, petitioners rely on a handful of isolated statements in the rulings under review. At one point in her decision, the judge noted that the Intervenor Group’s witness did not “prove[ ]” that the companies in its proposal had risks comparable to the TOs. Initial Decision, 99 FERC at 65,042. In affirming the ALJ, the Commission observed that, the Intervenor Group “has not demonstrated that the [chosen proxy] group is unrepresentative,” and “has failed to set, forth convincing evidence” of the superiority of their own proposal. Order Affirming Initial Decision, 100 FERC at 62,312. Based on these statements, petitioners argue that FERC improperly saddled them with the burden of proving the validity of the rate increase at issue — a burden statutorily assigned to the proponent of the increase, in this case MISO. See 16 U.S.C. 824d(e) (“the burden of proof to show that the increased rate or charge is just and reasonable shall be upon the public utility”).
In proceedings before the ALJ, however, MISO expressly assumed the burden of justifying the proposed rate increase. Prehearing Conference at 25 (“PRESIDING JUDGE: And you have the burden of proof, don’t you? MR. SMALL [counsel for MISO]: Yes, your Honor .... ”). In selecting MISO’s proposal, the judge explained that she found it “consistent with Commission precedent,” and described the Intervenor Group’s alternative as “more speculative and statistically less reliable.” Initial Decision, 99 FERC at 65,042. MISO acknowledged and carried its burden of proof before the ALJ. As for the *59statements in FERC’s order, these are best read as merely affirming the reasonableness of the judge’s decision. See Town of Norwood, 962 F.2d at 26 (interpreting similar phrasing as “a rather perfunctory restatement of the Commission’s finding ... and not an indication that the FERC improperly shifted the burden of proof to the petitioner”).
B. PSCKY and the Intervenor Group next challenge the Commission’s use of the midpoint. They question the cogency of FERC’s reasoning in general, and dispute its application on the facts of this case. They also point to language in FERC’s order on remand suggesting that the Commission chose the midpoint simply because it yielded the highest rate of return.
In the order on remand the Commission stepped back from its prior reliance on electrical industry precedents, and even acknowledged that the median, and not the midpoint, may be “the most refined measure of central tendency.” Order on Remand, 106 FERC at 62,192. The Commission rested its use of the midpoint on a wholly different ground: it distinguished between “cases in which a ROE is set for one gas pipeline or electric utility” and cases where “applicants proposed setting a single ROE for across-the-board application.” Id. In the latter situation, where “the ROE will apply to a diverse set of companies,” FERC reasoned that the range of results becomes as important as the central value. Id. The midpoint — unlike the other measures of central tendency — “fully considers that range,” because it is derived directly from the endpoints of the range. Id.
This justification provides a reasoned approach that lends itself to consistent application over a series of cases. While petitioners highlight the shortcomings of the Commission’s method, they fail to debunk its fundamental premises. For instance, petitioners are correct in noting that all measures of central tendency “consider” the entire proxy group range, in the sense that all are influenced — at least indirectly — by each data point in the range. Pet. Br. at 23. But only the midpoint emphasizes that range, as it is equally placed between the top and bottom values. Likewise, petitioners rightly suggest that using the midpoint undermines the statistical sampling of the proxy group process, id. at 28-30, as only the mean places equal weight on every point in a sample. But the Commission made clear that it was less interested in particular data points than in the full range covered by the group.
Petitioners’ more technical challenges fare no better. First, they contend that the Commission erred in calculating the midpoint of a range containing two values for each company — a high and a low estimate — because FERC meant to “fully consider” the range of companies in the proxy group, and high or low estimates by themselves do not reflect a company’s ROE without considering the corresponding low or high estimate. Pet. Br. at 26-27. The Commission’s use of a high estimate to mark the high end for its midpoint calculation was, however, balanced to some extent by its use of a low estimate to mark the low end. It is true that this balance is imperfect, particularly on this record, where the company whose high estimate marks the high end also had a low estimate among the lowest values in the range. But it is certainly consistent with FERC’s rationale to calculate the midpoint of the entire range, rather than — as petitioners now seem to suggest — averaging each company’s high and low estimates and then determining the midpoint of that narrower range. See Pet. Br. at 27. At any rate, no such alternative proposal was ever presented to the Commission. *60Second, petitioners maintain that the data in this case were too skewed for a midpoint analysis. Pet. Br. at 33-87. This too is a matter squarely within the Commission’s expertise. The proxy group clearly has some skew — otherwise all measures of central tendency would have produced the same result — but this is part of the reason FERC used the midpoint in the first place. FERC acknowledged that some distributions are too skewed for such an analysis, but the Commission evaluated the distortion here and found it not to be excessive. Order on Remand, 106 FERC at 62,193; compare Northwest Pipeline Corp., 99 FERC ¶ 61,305 (2002) (employing a similar method for evaluating skew). As the Commission explained, this is not a case where one end-point is considerably out of line with the rest of the range; ratemaking is not an “exact science,” Time Warner, 56 F.3d at 163, and FERC was not required to adopt petitioners’ mathematical model for determining when some skew becomes too much skew.
All this notwithstanding, the order on remand does contain language clouding the clarity of FERC’s midpoint rationale. In a footnote, the Commission observes that:
The results of our proxy group yielded a midpoint of 12.38, a median of 11.85, and a mean of 11.74. Relying on the median or mean in this case would result in an unreasonably low ROE in light of the high end values in the proxy group and could substantially under-compensate those utilities at the upper end of the range.
106 FERC at 62,192 n. 14. Elsewhere, FERC notes that “the midpoint does not have as deleterious an effect as the median on those [MISO] TOs whose returns are higher than what we allow for [MISO] TOs here,” and that this “offers additional incentive for such companies to join RTOs.” Id. at 62,193-94. According to petitioners, these passages indicate that FERC chose the midpoint simply because it yielded the highest result.
If FERC’s orders premised a policy of using the midpoint on an effort to obtain the highest rate of return, the orders could not withstand APA review. This is because there would be no logical connection between the rationale and the result: nothing about the midpoint ensures it will be higher than the median or the mean in any particular case. Of course, FERC could have explained that it would always choose the measure that yields the highest result, and tried to defend that rationale here. But the Commission cannot justify a commitment to the midpoint on the ground that it produced the highest return, because that is pure happenstance.
We think the most reasonable reading of FERC’s order on remand as a whole is that it selected the midpoint because the rate of return would apply to a diverse group of companies, and that the language petitioners point to simply reflects FERC’s view that the resulting rate gave it confidence that it was not undercompensating MISO TOs. The bulk of the order sets forth the “diverse companies” rationale. The contrary language appears in a footnote and a single paragraph near the end of the opinion. The Commission emphasized the rate’s across-the-board applicability to MISO TOs — rather than the greater value of the midpoint — in both its brief and at oral argument. In light of all this, we read the order as resting on its principal rationale. Should the Commission elect to abandon that rationale to rely on a higher median or mean in the next case involving applicability of the chosen rate to several companies, it will have to justify its departure from the precedent established by this case.
C. Finally, PSCKY and the Inter- ' venor Group argue that FERC arbitrarily adopted the 50-point premium, and that *61the Commission violated due process by announcing its decision sua sponte without prior notice to the parties. We focus on the latter claim.
The Due Process Clause and the APA require that an agency setting a matter for hearing provide parties “with adequate notice of the issues that would be considered, and ultimately resolved, at that hearing.” Williston, 165 F.3d at 63; see 5 U.S.C. § 554(b)(3). This requirement ensures the parties’ right to present rebuttal evidence on all matters decided at the hearing. See Bowman Tramp., Inc. v. Arkansas-Best Freight Sys., Inc., 419 U.S. 281, 288 n. 4, 95 S.Ct. 438, 42 L.Ed.2d 447 (1974); Hatch v. FERC, 654 F.2d 825, 835 (D.C.Cir.1981); Hill v. FPC, 335 F.2d 355, 363 (5th Cir.1964). In addition, the Federal Power Act authorizes FERC to approve rate increases only “after full hearings” — a requirement that itself “unquestionably imposes the duty to give adequate notice of the subject to which the orders pertain.” 16 U.S.C. § 824d(e); City of Winnfield v. FERC, 744 F.2d 871, 876 (D.C.Cir.1984).
Here, the Commission failed to place the parties on notice that its post-hearing order would contemplate an incentive-based premium for the MISO TOs. When MISO and its TOs first filed for the subject rate increase, they sought an incentive adder of 100 basis points, but FERC declined to consider such “innovative ratemaking proposals,” limiting the subject matter of the hearing to “ROE rates that essentially provide for appropriate cost-recovery.” Order Accepting in Part and Rejecting in Part Proposed Tariff Changes and Establishing Hearing Procedures, 98 FERC ¶ 61,064, 61,165, 2002 WL 123313 (2002). The ALJ refused to consider premium-related proposals, noting “that establishing an incentive based ROE seems to be outside the scope of this proceeding.” Initial Order, 99 FERC at 65,052. As a result, the record compiled at the hearing contained no evidence on the need for — or appropriate size of — such a premium.
On appeal, FERC maintains that it gave petitioners all the process they were due by considering their requests for rehearing, which contained extensive challenges to the premium. FERC Br. at 47; see PSCKY Req. for Reh’g at 9-14; Intervenor Group Req. for Reh’g at 41-49. Considering petitioners’ arguments, however, is not the same thing as allowing them to present evidence on the issue — as at least one of the petitioners pointed out to the Commission. See PSCKY Req. for Reh’g at 10 (“the parties were not given the opportunity to present evidence, or refute the use of such innovative rate making policy”); see also Williston, 165 F.3d at 63-64 (granting petition to review based on insufficient notice and lack of substantial evidence).
Nor did FERC obviate the need for such a presentation by relying, in the order denying rehearing, on its proposed policy to reward any entity that joins an RTO with a bonus of 50 basis points. The Commission issued that proposal after its sua sponte order and after the filing deadline for rehearing requests in MISO’s case. Moreover, the proposed policy is just that — it is not a rule binding on the public, or even on the agency itself. See Proposed Pricing Policy, 102 FERC at 61,066-67 (inviting public comment and holding off implementation until the issuance of a final policy statement). As such, it plainly cannot govern the outcome of this case.
Likewise, FERC’s assertion that petitioners should have been aware that it always possesses the power to modify rate proposals to ensure that they meet statutory standards plainly proves too much; FERC’s power to take such action does not carry with it the authority to exercise such power without adequate notice of the basis for doing so.
*62In sum, FERC failed to place petitioners on notice that it would consider an incentive-based premium, and ultimately applied the adder in MISO’s case without considering any record evidence. In so doing, FERC denied petitioners — and the other parties to the proceeding, for that matter — a chance to present their side of the case. We express no view on petitioners’ substantive challenges to the incentive-based adder, but grant the petition because FERC failed to provide adequate notice that it would consider such an element in assessing the pending rate proposal. In all other respects, we affirm the Commission.
. The Intervenor Group should not be confused with the intervenors in this appeal — the MISO TOs — who appear in support of the Commission. We use the label "Intervenor Group” because that is what was used in the orders under review, and because the parties themselves continue to use that shorthand before us.