Weekly review: +2.29R but -0.71% on capital - sizing, not direction, decided the week

This week (19-23 Oct) we published 4 ideas: MSFT long, US 30 long, US Tech 100 short and US 500 short. The last one is still open. 4 ideas closed this week, counting the US 500 short published on 16 Oct: two hit target 1 (+2R each), one hit its stop (-1R) and one was closed manually (-0.71R). In R that is +2.29. In capital it is -0.71%, because both winners were sized at 0.25% risk and both losers at the full 1%.

What worked: the two index ideas that followed the short-term move reached target within two sessions.
What did not: MSFT long was stopped out in a falling tech tape, and position sizing is inconsistent. The returns depend more on how much was risked than on whether the idea was right.

Since launch: 19 trades with an outcome, win rate 36.8%, average +0.01R, profit factor 0.86, return -1.4%. That sample is far too small to show an edge, or the lack of one, and every pattern in it is still a hypothesis. Every idea so far was scored conviction 3, so the score has not yet told us anything.

Next week: risk per idea should be uniform unless the idea itself states a reason, and the conviction score must actually discriminate. No new rules this week, because no error repeated often enough to justify one.