1、Asia Pacific Equity Research03 September 2025J P M O R G A NUtilities and EnvironmentalAlan Hon AC(852)2800-Daqi Jiao(852)2800-J.P.Morgan Securities(Asia Pacific)Limited/J.P.Morgan Broking(Hong Kong)LimitedWind farm operatorsTPShr pxUp/down-(Hkd)(Hkd)side(%)LY916 HK7.606.7612%DT RE1798 HK2.202.40-8%

2、CoTickerSource:Bloomberg Finance L.P.,J.P.Morgan estimates.Note:Priced as of 3 September 2025 market close.We apply machine learning techniques to predict China wind farms monthly power generation using weather data.Power generation for wind farms is a function of installed capacity and the amount o

3、f wind energy available(which is fairly consistent over the long term).However,monthly wind speed variance leads to share price volatility.Hence,accurate power generation forecasts can create trading opportunities ahead of generation statistic releases.Back-testing our model shows an average accurac

4、y of 95%.Our model projects Longyuans wind power generation to post 20%growth in Aug,ahead of the 10%organic capacity growth in FY24.We expect a positive share price reaction to Longyuans Aug data release,when it is published.We retain our OW rating on Longyuan on valuation groundds.Predicting month

5、ly wind farm generation using machine learning:We use wind speed data and other operational metrics as inputs(see page 2).In the long run,the amount of power a wind farm operator can generate should generally trend towards the available wind resources and capacity.However,monthly wind speed variance

6、 impacts monthly generation,and hence influences near-term share price performance.From FY19-23,annual wind power generation growth tracked reasonably close to trailing capacity growth(ranging from 2.1%to 15.5%),but monthly growth ranged from-17%to+51%.During FY24,however,a rising curtailment rate a