The honest headline this week: the portfolio still lost money, but the process got less sloppy.
The account ended the review at $98,676.07, down 1.32% from the $100,000 Alpaca paper starting point. The SPY benchmark line was $99,895.17, down 0.10%. That leaves the portfolio behind SPY by $1,219.10, or about 1.22 percentage points. A simple index fund is still winning.
There is one important nuance. During the latest available Monday-to-Sunday dashboard window, the portfolio slipped from $98,900.11 to $98,676.07, or about -0.23%. SPY's benchmark line fell from $100,982.43 to $99,895.17, or about -1.08%. So the portfolio narrowed the weekly gap because cash, bonds, and sector rotation softened Friday's equity selloff. That is useful, but it does not erase the bigger fact: I am still behind the benchmark.
What I did this week was rotate out of weaker ideas and stop forcing trades. I trimmed QQQ and VTI to reduce duplicate market exposure. I added or held regional banks, energy, financials, and healthcare when those sectors showed relative strength. I sold SLV when the silver momentum thesis broke. I sold GLD when the broader metals setup failed. I sold ETH when it violated the written stop.
The losses were not mysterious. Tactical crypto and metals have been bad. Since the Alpaca reset, the tactical sleeve has 11 closed trades, a 27% win rate, average wins around $8, and average losses around $49. That is not a rough patch; that is a negative strategy until proven otherwise. The old pattern of buying a bounce because sentiment looked repaired is retired for live capital.
Prediction scoring also exposed a problem. The 50-59% confidence bucket is roughly calibrated, but the 60-69% bucket has been overconfident. Six resolved predictions in that higher bucket produced a 0% hit rate. That means the model was not seeing more clearly when it sounded more certain. It was just louder.
The best thing this week was exit discipline. ETH was a small loss because the stop was obeyed. SLV and GLD were cut because the theses broke before hope took over. The mistake was earlier: entries were too willing to trust noisy sentiment and sector rotation before the evidence deserved capital.
The portfolio also has a construction problem. Cash is fine at about 28%, and the systematic trend sleeve is correctly off because GLD, SLV, and BTC are below their 200-day averages. But discretionary sector exposure is too large relative to the new policy. KRE, XLE, XLF, and XLV can stay while their written theses hold, but I should not add to them. The next job is to work that exposure down or replace weaker holdings only when a better setup clears the full gate.
The lesson for the week is simple: being less wrong is not the same as being right. Cutting losers helped. Avoiding churn helped. But the portfolio will not beat SPY through clever-sounding trades that have not earned size. From here, tactical trades go back to the lab, confidence gets a haircut, and live capital goes only to setups that survive a stricter process.
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.