Appeal from an order and judgment of the Supreme Court (Sise, J.), entered June 18, 2003 in Fulton County, which, inter alia, partially granted petitioner’s applications, in four proceedings pursuant to RPTL article 7, to reduce tax assessments for the years 2000 and 2001 on certain real property owned by petitioner.
In 1999, following the deregulation of the electric utility industry in New York, petitioner purchased 72 hydroelectric power plants from Niagara Mohawk Power Corporation. Facilities comprising a portion of one such plant, including dams, pipelines and a powerhouse, are located on four parcels of land in Fulton County—one parcel in the Town of Johnstown and three parcels in the Town of Ephratah. Petitioner commenced these proceedings pursuant to RPTL article 7 challenging its 2000 and 2001 real property tax assessments on those parcels. Following a bench trial, Supreme Court rejected, as fatally flawed, petitioner’s income method of valuation, which used a discounted cash flow analysis (hereinafter DCF) based on estimates of future revenue from the sale of electrical power generated at its plant. Supreme Court instead relied upon the reproduction cost new less depreciation method (hereinafter RCNLD) also provided in the appraisal submitted by petitioner. In a thorough, well-reasoned decision, Supreme Court concluded that only two of petitioner’s four parcels had been overvalued. Arguing that Supreme Court erred in not applying its DCF method of valuation, petitioner appeals.
Initially, petitioner presented sufficient evidence to rebut the presumption that respondents’ assessments were valid (see e.g. Matter of Niagara Mohawk Power Corp. v Assessor of Town of Geddes, 92 NY2d 192, 198 [1998]). Petitioner was then required to establish by a preponderance of the evidence that respondents overvalued one or more of its parcels (see Matter of Norton Co. v Assessor of City of Watervliet, 3 AD3d 760, 760-761 [2004]; Matter of Niagara Mohawk Power Corp. v City of Cohoes Bd. of Assessors, 280 AD2d 724, 726 [2001], lv denied 96 NY2d 719 [2001]). Upon our review, we will “defer to Supreme Court’s decision ‘unless such finding is based upon [an] erroneous theory of law or [an] erroneous ruling in the admission or exclusion of evidence, or unless it appears that the court . . . has failed to *542give to conflicting evidence the relative weight which it should have and thus has arrived at a value which is excessive or inadequate’ ” (Matter of City of Troy v Town of Pittstown, 306 AD2d 718, 720 [2003], lv denied 1 NY3d 505 [2003], quoting Matter of City of New York [Newtown Cr.], 284 NY 493, 497 [1940]). Weighing the entire record, including the parties’ proof as to the appropriate methodology and the various methods of valuation contained in petitioner’s appraisal (respondents having withdrawn their own appraisal during trial), and deferring to Supreme Court’s resolution of credibility in assessing the report and testimony of petitioner’s appraiser (see Matter of Golub Corp./Price Chopper Operating Co. v Assessor of Town of Queensbury, 282 AD2d 962, 962 [2001]; Matter of Saratoga Harness Racing v Williams, 256 AD2d 663, 664-665 [1998], lv denied 93 NY2d 803 [1999]), we find that petitioner failed to show that the valuation method used by Supreme Court was improper.
At trial, petitioner presented extensive appraisal evidence employing the comparable sales, DCF and RCNLD methods of valuation. Supreme Court accepted petitioner’s argument that, following deregulation of the industry, a market had developed for power facilities and, thus, they should no longer be considered specialty properties to be valued using only the RCNLD method (see Matter of Niagara Mohawk Power Corp. v Town of Moreau Assessor, 307 AD2d 669, 671 [2003]). The court was unable, however, to utilize comparable sales, the most preferred method (see Matter of Saratoga Harness Racing v Williams, 91 NY2d 639, 643 [1998]), because the parties agreed that there was insufficient market information available. In that event, where evidence of comparable sales is unavailable, “[a]ny fair and nondiscriminating method” that produces a “fair and realistic value” is acceptable (Matter of Allied Corp. v Town of Camillus, 80 NY2d 351, 356 [1992]).
Of the two remaining methods of valuation available here, the capitalization of income method for determining the value of income-producing property is preferred. The income normally capitalized, however, is the rental income of the property (see e.g. Matter of Barnum v Srogi, 54 NY2d 896, 898 [1981]; Matter of Avis Rent A Car Sys. v Town of Rye, 131 AD2d 568, 568 [1987]). Moreover, it is a “method [that] can be effective only with thorough data, including accurate actual income and operating expenses of the subject properties” (41 Kew Gardens Rd. Assoc. v Tyburski, 70 NY2d 325, 331 [1987] [emphasis added]). Even if the income from future sales of electrical power were, like rent, inextricably tied to petitioner’s specific parcels and the best indicator of their actual values (see Matter of Niag*543ara Mohawk Power Corp. v Town of Moreau Assessor, supra at 671; compare Matter of NYCO Mins. v Town of Lewis, 296 AD2d 748, 750 [2002], lv dismissed and denied 99 NY2d 576 [2003]), an issue we need not reach, we would still agree with Supreme Court that petitioner’s application of the DCF method here did not accurately reflect the taxable value of its properties.
The first defect in petitioner’s DCF approach is the failure of its appraisers to use actual income based on two power purchase agreements (hereinafter PPA) between petitioner and the Niagara Mohawk Power Corporation. While petitioner’s own experts testified as to the difficulty of forecasting the price of a commodity such as electricity into the future, Mark Pomykacz, petitioner’s appraiser, attempted to forecast what the market price would be for such a commodity over a 10-year period. Pomykacz testified that he used the PPA in place for 2000 to establish the income for that year, but he declined to use it for any part of 2001, even though it ran until September 2001. He also declined to use a second PPA covering the period from October 2001 to September 2004. Instead, Pomykacz used market rate information accumulated from November 1999 through December 2000, which Supreme Court found to be an unreasonably narrow time frame for purposes of collecting a sample in an indisputably volatile market. Pomykacz used this limited sample to then estimate electricity prices for the years 2001 to 2010 without providing a convincing reason for ignoring actual prices set by the market or the PPAs (see Midland Cogeneration Venture v City of Midland, 2004 WL 212459, *62-63, 2004 Mich Tax LEXIS 2, *164-165 [Mich Tax Trib, Jan. 23, 2004]; compare Watson Cogeneration Co. v County of Los Angeles, 98 Cal App 4th 1066, 1071-1072, 120 Cal Rptr 2d 421, 425 [2002]; Freeport-McMoran Resource Partners v County of Lake, 12 Cal App 4th 634, 643-644, 16 Cal Rptr 2d 428, 433 [1993]).
In addition, the assumptions made by Pomykacz in using a Monte Carlo simulation to confirm his forecasts of the price of electricity were effectively challenged by respondents’ experts and discredited by Supreme Court. As Supreme Court found, Pomykacz used unreliable figures that confirmed low price estimates, and he had never before used the Monte Carlo analysis in an appraisal (see Matter of Tennessee Gas Pipeline Co. v Town of Sharon Bd. of Assessors, 298 AD2d 758, 759-760 [2002], lv denied 99 NY2d 506 [2003]). To the extent that petitioner urges that its DCF method must be adopted because purchasers regularly utilize it to determine the value of power plants, we need only note that such sales are of ongoing businesses where numerous factors beyond the value of the real property and *544generating equipment are involved. These defects, combined with Supreme Court’s credibility determination that petitioner overstated its expenses, led to a significant underestimate of the net income that petitioner’s properties would produce over the forecast years.
Inasmuch as the record supports Supreme Court’s finding that petitioner’s DCF analysis was based on unreliable price forecasts and overstated operating expenses, it was appropriate for the court to reject it and elect to use the RCNLD method. While this approach must be used with caution, since it may overvalue property if insufficient obsolescence is applied (see e.g. Matter of Niagara Mohawk Power Corp. v Assessor of Town of Geddes, 92 NY2d 192, 197 [1998], supra), Supreme Court met this concern by adopting petitioner’s own figures. Contrary to petitioner’s contention, there is nothing inherently inappropriate about this approach, as we regularly upheld it for the valuation of hydroelectric facilities before deregulation (see Matter of Niagara Mohawk Power Corp. v City of Cohoes Bd. of Assessors, 280 AD2d 724, 724 [2001], supra). Accordingly, we find no basis to disturb Supreme Court’s determination.
Petitioner’s remaining argument, that Supreme Court improperly permitted respondents to present testimony by previously-undisclosed expert witnesses, lacks merit inasmuch as these witnesses testified in rebuttal as to the accuracy of petitioner’s application of the DCF approach here and did not themselves opine as to the values of petitioner’s properties (see 22 NYCRR 202.59 [h]; Matter of John P. Burke Apts, v Swan, 137 AD2d 321, 324 [1988]).
Cardona, EJ., Crew III, Peters and Mugglin, JJ., concur. Ordered that the order and judgment is affirmed, without costs.