MintPop 为企业战略部门、投资机构与行业研究者提供定制化的数据与行业研究服务，涵盖数据建模、市场测算与竞争格局分析，以明确结论支持决策。MintPop provides custom data and industry research for corporate strategy teams, investors and industry researchers, spanning data modeling, market sizing and competitive analysis, with clear conclusions that support decisions.
产业链结构、竞争格局、商业模式与关键驱动变量，明确价值分布与竞争优势来源。Value-chain structure, competitive dynamics, business models and key drivers, identifying where value accrues and what sustains advantage.
市场规模、价格与需求预测，完整呈现预测区间、关键假设与误差水平。Market size, price and demand forecasts, presented with ranges, key assumptions and error levels.
整合公开与行业数据，交付可复用的模型、底稿与图表，每项结论均可追溯至数据来源。Public and industry data integrated into reusable models, workpapers and charts, with every conclusion traceable to its source.
以下研究由 MintPop 创始团队在公司成立前完成，用于展示我们的研究方法与交付形态。The following studies were completed by the MintPop founding team prior to the company’s formation and illustrate our research methods and deliverables.
PJM 批发电价七年上涨 151%。其中天然气贡献 28%，其余主要来自数据中心需求增长与电网紧张度上升。PJM wholesale power prices rose 151% over seven years. Natural gas accounts for 28% of the increase; the balance is driven mainly by data-center demand and rising grid scarcity.
研究服务于一家分布式光伏公司的自有资金决策，核心问题是：AI 用电需求推高电价这一判断在多大程度上成立，相关投资的价值与风险如何把控。PJM 是美国最大的区域电网，覆盖 13 个州和华盛顿特区，服务人口约 6,500 万。Conducted for a distributed-solar company’s proprietary capital. The core question: to what extent AI power demand is raising electricity prices, and how to assess the value and risk of related investments. PJM is the largest US regional grid, covering 13 states and DC and serving about 65 million people.
PJM 批发电价：2019 年均 $33.5/MWh，最近 12 个月均价 $84.3/MWhPJM wholesale power: $33.5/MWh average in 2019, $84.3/MWh over the last 12 months
全网负荷净增量中，数据中心最集中的北弗吉尼亚分区占 74%Northern Virginia, the zone with the highest data-center concentration, accounts for 74% of net load growth
容量拍卖价三年内由 $28.92 升至 $333.44 /MW-天，连续两年触及监管上限Capacity auction prices rose from $28.92 to $333.44 /MW-day in three years, reaching the regulatory cap two years running
2022 年电价与气价同步走高，主要由气价驱动；2024 年起电价涨幅明显快于气价。按年均价计算，2026 年（1–7 月）气价较 2019 年上涨约 45%（$2.57 至 $3.73），电价达到 2019 年的 3 倍（$33.5 至 $100.7）。In 2022 power and gas rose together, driven mainly by gas. From 2024 power prices outpaced gas. On annual averages, 2026 gas (Jan–Jul) is about 45% above 2019 ($2.57 to $3.73), while power reached 3× its 2019 level ($33.5 to $100.7).
将 2019 年以来的涨幅拆分为四项，合计 100%。天气项为负值，反映最近一年气温略低于 2019 年。The increase since 2019 decomposed into four components summing to 100%. Weather is negative, reflecting a slightly cooler latest year versus 2019.
为一家大型主题公园与度假村运营商构建 12 个月天然气价格预测模型，平均误差约 10%。模型提前识别出期货定价偏高，并测算了调整锁价比例带来的成本节省空间。A 12-month natural-gas price model built for a large theme-park and resort operator, with mean error around 10%. The model identified futures as overpriced in advance and quantified the savings from adjusting the hedge ratio.
该公司每年在财年开始前通过期货锁定约 80% 的用气量，此前的采购决策主要依据交易员经验与期货价格。Each year ahead of the fiscal year, the company hedged about 80% of its gas volume with futures; procurement decisions previously relied on trader judgment and the futures curve.
12 个月预测的平均误差（样本外测试期：2023 年 11 月至 2024 年 10 月）Mean error of the 12-month forecast (out-of-sample test: Nov 2023 – Oct 2024)
同期期货锁定价较实际现货价平均高出 55%，12 个月中每月均高于现货Over the same period futures averaged 55% above actual spot, exceeding spot in each of the 12 months
采购量中每将 10% 由期货转为现货采购，年均可节省约 22.9 万美元Shifting each 10% of volume from futures to spot could save about $229K per year
测试期 12 个月中，期货锁定价每月均高于实际现货价，模型预测值与实际价格高度贴合。Across the 12-month test window, the futures strip exceeded actual spot every month, while the model forecast closely tracked actual prices.
方法：从 20 多个候选变量中筛选出 6 个驱动因子，整合 CME 逾 100 万条期货日度记录与 EIA 月度基本面数据，时间跨度 12 年。误差结果基于上述测试期。Method: 6 drivers selected from 20+ candidates, integrating over 1 million CME daily futures records with 12 years of EIA monthly fundamentals. Error results are based on the test window above.
聚焦单一研究问题
核心数据与图表
结论摘要One focused research question
Key data and charts
Summary of conclusions
定制研究框架
数据模型与底稿交付
结论汇报会Custom research framework
Data model and workpapers
Final readout session
全年研究额度
专属分析师团队
季度行业简报，48 小时响应Annual research allowance
Dedicated analyst team
Quarterly briefings, 48h response