+9.64%
Project Midas AI 公开组合收益Project Midas AI Public Portfolio Return
+9.64%
vs S&P 500
+12.72%
vs NASDAQ 100
+-0.0%
S&P 500
-3.08%
NASDAQ 100
持仓Holdings
15/18
现金Cash
7.1%
市场研判Regime
NEUTRAL
决策评分Avg Score
54.8/100
风险调整指标Risk-Adjusted Metrics
夏普比率Sharpe
每承受 1 单位总波动，组合获得了多少超额收益；越高越好。
Shows how much excess return the portfolio earns for each unit of total volatility. Higher is better.
1.37
优Strong
索提诺比率Sortino
和夏普类似，但只惩罚下行波动，更适合看回撤敏感度。
Similar to Sharpe, but only penalizes downside volatility, making it more sensitive to drawdown risk.
1.38
中Avg
最大回撤Max Drawdown
从阶段高点跌到低点的最大幅度；绝对值越小，过程越稳。
The largest peak-to-trough drop. A smaller magnitude means the ride was steadier.
-5.7%
弱Weak
超额阿尔法Alpha
扣除市场整体波动和参考无风险收益后，策略额外创造的回报；正值更好。
The return generated beyond what market exposure and the reference risk-free yield would imply. Positive is better.
+20.4%
优Strong
贝塔系数Beta
相对基准的波动敏感度；1 接近大盘，低于 1 更稳，高于 1 更激进。
Measures sensitivity to benchmark moves. Around 1 tracks the market, below 1 is steadier, above 1 is more aggressive.
0.54
低波Low Vol
卡玛比率Calmar
年化收益相对于最大回撤的效率；越高说明同等回撤换来的收益更多。
Shows how efficiently annualized return was earned relative to max drawdown. Higher means better return for the pain taken.
3.92
优Strong
盈亏比Profit Factor
盈利交易总利润除以亏损交易总亏损；大于 1 说明整体是赚钱的。
Gross profits divided by gross losses. Above 1 means winning trades outweigh losing ones overall.
0.74
弱Weak
平均盈亏比Avg W/L
平均盈利单笔除以平均亏损单笔；越高说明赚时更敢赚、亏时更能控损。
Average winning trade divided by average losing trade. Higher means gains are meaningfully larger than losses.
2.03
优Strong
基于 115 个交易日快照，这些风险收益指标按 4.3% 的现金/短债参考收益率计算 Based on 115 trading days, these risk-adjusted metrics assume a 4.3% cash/short-term Treasury reference yield
绩效对比Performance vs Benchmarks
AI 组合AI Portfolio
S&P 500
NASDAQ 100
数据来自第三方（含 Yahoo Finance），可能存在延迟Data from third-party providers incl. Yahoo Finance, may be delayed
当前持仓Current Holdings
| 代码Symbol | 名称Name | 成本Avg Cost | 现价Price | 盈亏P&L | 权重Weight |
|---|---|---|---|---|---|
| NEE | NextEra Energy | 91.93 | 94.08 | +2.3% | 10.46% |
| WMT | Walmart Inc. | 128.77 | 126.79 | -1.5% | 9.29% |
| SEI | Solaris Energy Infrastructure | 49.45 | 62.38 | +26.1% | 10.74% |
| PFE | Pfizer Inc. | 27.30 | 26.91 | -1.4% | 4.82% |
| MRK | Merck & Co., Inc. | 115.91 | 121.42 | +4.8% | 5.13% |
| JPM | JPMorgan Chase & Co. | 283.44 | 309.92 | +9.3% | 5.35% |
| BIP | Brookfield Infrastructure Partners L.P. | 37.92 | 36.56 | -3.6% | 4.72% |
| ABT | Abbott Laboratories | 105.46 | 100.28 | -4.9% | 4.65% |
| KO | The Coca-Cola Company | 74.75 | 77.47 | +3.6% | 5.07% |
| WEC | WEC Energy Group, Inc. | 112.18 | 117.52 | +4.8% | 5.13% |
| BMY | Bristol-Myers Squibb Company | 58.54 | 58.62 | +0.1% | 4.9% |
| GLD | SPDR Gold Shares | 431.81 | 437.18 | +1.2% | 5.5% |
| GDX | VanEck Gold Miners ETF | 94.97 | 99.43 | +4.7% | 5.69% |
| SLV | iShares Silver Trust | 68.59 | 69.10 | +0.7% | 5.48% |
| IBIT | iShares Bitcoin Trust ETF | 41.41 | 41.56 | +0.4% | 5.46% |
当前研判Current Regime
— 2026-04-09
研判日期Decision Date
2026-04-09
NEUTRAL
今日执行了 1 项调仓1 action taken today
Portfolio remains defensively positioned with core holdings in healthcare (MRK, ABT, BMY), utilities (NEE, BIP, WEC), and precious metals (GLD, SLV). All holdings show intact theses; no sells warranted. Today's decision adds Bitcoin ETF (IBIT) to capture digital‑asset diversification and institutional adoption tailwinds, using available cash while staying within risk limits. Forward outlook: monitor inflation data and ceasefire durability; watch overbought technicals in utilities and banks for potential trim opportunities.
投资组合保持防御性配置，核心持仓集中在医疗保健（MRK、ABT、BMY）、公用事业（NEE、BIP、WEC）和贵金属（GLD、SLV）。所有持仓的投资逻辑完好，无需卖出。今日决策新增比特币ETF（IBIT），以捕捉数字资产多元化和机构采用带来的顺风，使用可用现金同时保持在风险限制内。前瞻展望：关注通胀数据和停火协议的可持续性；留意公用事业和银行股的技术超买情况，寻找潜在减持机会。
买入
Buy
IBIT
iShares Bitcoin Trust ETF
IBIT provides diversified crypto exposure with strong structural quality (AUM $53B, iShares). Macro fit: Bitcoin acts as alternative asset and inflation hedge; ceasefire has boosted risk sentiment. Catalyst: Morgan Stanley ETF launch, recent price rally. Technicals show improvement (MA5 above MA20, RSI 57.9). Composite score 3.8 meets BUY threshold, and portfolio has zero crypto allocation (well below 15% limit).
最近变更Recent Changes
这里只保留最近 30 天内真正发生过持仓调整的记录；无变更复核默认不展示。 Only real holdings changes from the last 30 days are shown here; no-change reviews are intentionally hidden.
决策追踪记录Track Record
胜率Win Rate
得分达到 60 分及以上的已评估决策占比；要和盈亏比一起看。
The share of evaluated decisions scoring 60 or above. Best read together with payoff metrics, not alone.
33.3%
平均分Avg Score
所有已评估决策的平均得分，用来观察模型整体执行质量。
The average score across all evaluated decisions, giving a quick view of overall execution quality.
54.8/100
已评估Evaluated
已经完成 T+5 自动评分的决策数量；样本越多，参考性越强。
The number of decisions already scored by the T+5 evaluation loop. More samples make the stats more reliable.
6
最佳Best
/
最差Worst
历史已评估决策中的最高分和最低分，用来看上限与失误幅度。
The highest and lowest evaluated decision scores, showing both upside and the size of mistakes.
90.0 / 15.0
最佳Best
90.0
Removed XOM
The current market environment is assessed as RISK-OFF, characterized by defensive sector leadership and elevated cash allocation. Our portfolio refl...
2026-03-27
教训Lesson
15.0
Removed GEV
Market regime is clearly RISK-OFF with elevated VIX (23.51) and defensive sector rotation. Energy, Utilities, Consumer Staples are outperforming whil...
2026-03-16
以下仅展示最近 30 天内已完成评估的决策；最新记录在表格顶部。若列表超出可视区域，可向下滚动查看。
Evaluated decisions from the last 30 days only; newest rows are on top. Scroll down if the list is taller than the viewport.
| 日期Date | 市况Regime | 得分Score | 操作Actions |
|---|---|---|---|
| 2026-04-02 | UNKNOWN | 56.2 | HOLD |
| 2026-03-27 | RISK-OFF | 90.0 | SELL XOM -5.97% BUY BMY +5.45% |
| 2026-03-20 | RISK-OFF | 55.0 | SELL COST +0.26% BUY ABT -0.6% BUY KO +0.67% BUY WEC +0.48% |
| 2026-03-16 | RISK-OFF | 15.0 | SELL GEV +6.74% BUY BIP -6.22% |
| 2026-03-13 | RISK-OFF | 70.0 | SELL LMT -0.58% BUY JPM +1.52% |
| 2026-03-12 | RISK-OFF | 42.5 | SELL TSM +3.04% BUY MRK +0.71% |
评估方法Evaluation Method
每条决策在 T+5 个交易日后自动评估。买入/加仓：股价上涨越多得分越高。卖出/减仓：股价下跌越多（验证了退出决定）得分越高。≥60 视为"胜"。结果自动反馈给 AI 校准未来决策。 Each decision is auto-evaluated T+5 trading days later. BUY/INCREASE: higher score for larger gains. SELL/DECREASE: higher score for larger declines (validating the exit). Score ≥60 is a "win". Results feed back to the AI for self-calibration.
一个独立开发者的公开实验：让 AI 管钱，所有决策可追溯、可验证。
An indie developer's public experiment: let AI manage money, every decision traceable and verifiable.
Project Midas 不是又一个 AI 概念演示。这是一个用真实市场数据运行的虚拟组合，每一笔买卖都记录在案，每一次判断都接受时间检验。我们想知道的只有一件事：AI 到底能不能做出好的投资决策？ Project Midas is not another AI concept demo. It's a virtual portfolio running on real market data, where every trade is logged and every judgment is tested by time. We're asking one question: can AI actually make good investment decisions?
运作机制How It Works
每个交易日，系统自动执行四阶段决策流水线。没有人工干预，没有事后修改。AI 独立判断，系统忠实执行。
Every trading day, the system runs a 4-phase decision pipeline automatically. No manual override, no after-the-fact edits. The AI decides, the system executes.
市场研判Regime
→
持仓审查Review
→
机会扫描Scan
→
风控执行Execute
数据来源Data Source
全部行情数据来自 Yahoo Finance（通过 yfinance），包括实时价格、历史 K 线、财务报表和技术指标。无私有数据，无信息优势——和散户拿到的数据完全一致。
All market data comes from Yahoo Finance (via yfinance): real-time prices, historical candles, financial statements, and technical indicators. No proprietary data, no edge — the same data any retail investor can access.
为什么公开Why Public
AI 投资有太多"回测神话"和事后诸葛。这个页面存在的意义是前瞻性记录：每个决策在执行前公布，准确度在事后自动评分。对就是对，错就是错，没有修改空间。
AI investing is full of backtesting myths and hindsight narratives. This page exists as a forward-looking record: decisions are published before outcomes, accuracy is scored automatically after the fact. Right is right, wrong is wrong — no room for revision.
评估体系Scoring
决策发布后，系统会在积累足够市场数据后自动计算准确度评分，衡量 AI 判断与实际走势的吻合程度。高分不被美化，低分不被隐藏。
After decisions are published, the system automatically calculates an accuracy score once sufficient market data accumulates, measuring how well the AI's judgment aligned with actual outcomes. High scores aren't glorified, low scores aren't hidden.
⚠ Project Midas 是一个虚拟教育实验项目，不构成任何投资建议。所有交易均为模拟操作，不涉及真实资金。过往表现不代表未来收益。投资有风险，请自行尽调。
⚠ Project Midas is a virtual educational experiment and does not constitute investment advice. All trades are simulated with no real capital involved. Past performance does not guarantee future results. Always do your own due diligence.