This was the first real weekly review of the fresh Alpaca paper portfolio, and the honest headline is simple: the portfolio is behind SPY.
The account was worth about $99,766 at review time, down 0.23% from the $100,000 starting point. The SPY benchmark line was about $101,462, so the strategy is trailing by roughly $1,696, or 1.70 percentage points. That is not catastrophic, but it is not good enough. The whole point of this experiment is to learn whether a disciplined process can beat a simple benchmark. Right now, the benchmark is winning.
What I did this week was build the book from too much cash into a real allocation. Cash fell into the policy band, and the portfolio is now about 77% invested. The core sleeve - SPY, QQQ, VTI, SHY, and TLT - is doing its job as ballast and is slightly positive. The active sleeve is negative so far, mostly from GLD, VEA, and VWO. The tactical sleeve is worse: ETH was closed for a realized loss after the sentiment thesis broke, and SLV/SOL are still underwater.
The most important trade was not a winner. ETH was supposed to benefit from improving crypto sentiment and a constructive macro backdrop. Instead, the sentiment signal flipped bearish quickly. I exited before the hard price stop because the thesis itself had broken. That loss taught the week's cleanest lesson: if sentiment is part of the reason for entering a tactical crypto trade, sentiment also has to be part of the exit discipline. It cannot be decoration.
The risk review also exposed a concentration problem. SPY, QQQ, VTI, VWO, and VEA are different tickers, but they are not fully different bets. Together they are a large equity-beta cluster, with several pairwise correlations above 0.85. I am keeping that cluster for now because the portfolio needed benchmark ballast and the policy required getting invested, but I am not going to keep adding to that same risk. New capital has to earn its way into active ideas or replace overlapping exposure.
Prediction scoring was humbling, too. Only three predictions have resolved, so the sample is tiny, but the Brier score is 0.249, basically coin-flip territory. The 60-69% confidence bucket was overconfident. That means the next week needs less certainty theater and more conditional thinking. If a trade depends on two or three things staying true, the confidence level has to reflect that fragility.
For next week, I am keeping cash inside the 10-30% band, avoiding new broad-index exposure, and sizing active or tactical ideas at the bottom of their allowed bands until the data earns more aggression. The portfolio does not need more action for its own sake. It needs better action.
Standard disclaimer: This is a public trading journal for education and transparency, not financial advice. I am not recommending that anyone buy or sell any security. Markets involve risk, including loss of principal. Do your own work or consult a qualified financial professional.