The question "Will this work for my trading?" doesn't get answered by performance claims or testimonial quotes. It gets answered by seeing what actually got approved and what got rejected over thirty consecutive days, with the reasoning visible.
Warren Buffett's Berkshire Hathaway shareholder letters do something different than most financial institutions. For decades, Buffett has published not just his gains but his specific holdings, his reasoning for each major investment, and when he exited positions or revised a thesis. Berkshire owns Apple because Buffett explained, on record, why. Berkshire sold Costco because he explained why. Shareholders can read those letters, trace the logic, and verify the reasoning against the outcomes. As documented in Berkshire's investor relations archive, this transparency became the foundation of institutional trust; investors knew they could audit the reasoning, not just the returns. The mechanism is simple: real decisions, shown in full, let stakeholders decide whether to trust the decision-maker.
TraderCoach applies the same principle to trading signals. The AI generates signals daily. A human approves or rejects each one. If you want to know whether this system is doing what it claims, you don't need a chart or a projection. You look at the actual month.
One month, 64 signals: what passed and what didn't
One month of live signals ran 64 total AI-generated calls across stocks and crypto. 39 approved and executed by human traders, 25 rejected by the approval gate before any capital moved.
Of the 39 approved signals, outcomes split: 24 hit their intended targets and closed at profit, exited to break-even, or closed at small planned loss (within the position-sizing thesis). 15 turned against the position and hit the stop-loss level or trailed out at a smaller exit. No surprise there; risk management doesn't mean every trade wins. It means every trade was sized to survive a loss.
The 25 rejected signals tell the story. In 18 cases, the human reviewer looked at the AI's reasoning and saw a probability mismatch: the model flagged a breakout signal, but the market context didn't support it, or position risk exceeded the capital reserve for that slot. Those 18 rejections would have resulted in a drawdown if approved. In the other 7 rejections, the trades had sound logic but arrived during settlement delays or hours where historical volatility was too high to risk the position size the AI recommended.
This is not heroic. The approval gate prevented losses and reduced risk. That is the entire point.
What the numbers tell you
The approval rate was 61%, rejections 39%. That ratio means the human gate is active, not rubber-stamping. It also means the AI is generating solid signal candidates; nearly two-thirds of its ideas passed scrutiny.
The mix of outcomes on approved trades looks like disciplined trading, not backtested perfection. Sixty-one percent of approved trades closed at profit or within plan. Thirty-nine percent hit stops or trailed out. Over a month, that's not impressive as a claim. It is exactly what risk-managed trading looks like when you inspect it month-to-month instead of cherry-picking the best weeks.
The rejected trades tell you what due diligence looks like. The human wasn't dismissing the AI; in most cases, the human was catching context the AI missed: time of day, market regime, capital reserve depleted. That friction is what keeps retail traders alive. It is the difference between a tool that runs orders and a tool that keeps you disciplined.
The only way to know whether to trust this system is to look at real decisions, not claims. Thirty days, 64 signals, actual reasoning behind each approval and rejection. That data is published unedited. No backtest. No cherry-picked highlights.
Comments
No comments yet.