@risk_first

Add: write the rejection criteria before you open the report. If you negotiate with the chart after seeing OOS, you already overfit your process.

Used Portfolio Doctor by Zoe on a 12-position mess. The useful part was not the letter-grade — it was the recovery-tree forcing me to write down "if USD spikes, which three legs die together?". Cons: OCR of my broker screenshot misread one lot size, and the narrative tone can…

I still gate every sleeve through forward survival thinking — WorldBotRank as one input, my own demo log as the real one. Agentic is orchestration sugar; risk ownership is not optional.

Daily PnL correlation matrix in a spreadsheet first. Anything >0.6 over 60 trading days shares a risk budget. My EA Farm is handy for seeing what is running; QuantMogul MC is where I decide if the farm's joint drawdown is survivable.…

The 30% in-sample to out-of-sample haircut matches my rule of thumb — I assume 25-40% degradation and I am pleasantly surprised maybe once a year. And thank you for the honest "not fundable, it is a building block" framing. That sentence is what most WFO posts carefully omit.

Flat is a win in a transition year — most single-strategy accounts I track were down 20-40%. What was the return correlation between your trend and mean-reversion sleeves? My similar pair ran about -0.3 in 2025, which is the only reason my year looked like yours.

My logging list (English, hope that's ok). Per trade: requested vs filled price, spread at entry, duration, and P&L vs the same-period backtest trade. Weekly: live equity curve plotted against the backtest's percentile band. Pass criteria I actually use: live PF within 20% of backtest PF, average…

The asymmetry is the tell. Symmetric slippage is honest market impact; consistently asymmetric slippage is the broker's business model. Question: did B's asymmetry persist after you excluded the news windows? That separates "worse during chaos" from "worse by design".

Same rule on PortfolioVerse: process theft only. I rewrite the idea in my own words, WFO it in QuantMogul, then demo. If I cannot explain the edge without the leaderboard rank, I discard it.

Monotonic vs jump is exactly the right framing — mine were both, one per EA, and I never separated them until now. Adding divergence-shape tracking to the monthly log. This thread is why I post data here.