Blog
How the system works, what we change, and why — including the corrections.
Here is every signal we got wrong, with the levels we published
Our track record has always counted the losses. It never showed you the setup behind them. From today a closed signal publishes its entry, stop and target, and every one that stopped out is listed on a single page.
Read →We added 402 instruments, and here is exactly what that is worth
The scanner now reads 2,175 instruments across 26 markets — Japan, Hong Kong and Korea went from a 26-instrument pilot to 428. We also measured what an expansion actually buys, and it is smaller than it sounds.
Read →New strategies are now in forward test
Our live signals are performing well — 64% of resolved signals have reached their target. We also backtested a wider set of strategies, and several performed better still — 9% to 39% annualised in simulation. Those are now running in forward test, and we plan to open them to subscribers in a few months.
Read →Six of fifteen signals were gone by the close
We used to scan an hour before each market closed. Some of the setups we published had stopped existing by the closing bell, and other real ones appeared after we had stopped looking. Here is what we measured and what we changed.
Read →We changed the scoring rule again, and this time the number went up
Three days ago we lowered our win rate from 68% to 60% and said we would not change the rule again unless there was a structural reason, stated here, with the numbers. Here is that reason, and those numbers. The rate is now 66%.
Read →We lowered our own win rate from 68% to 60%
We changed how we score every signal on our track record. The headline number went down. Here is the data that made us do it, and why the old number was the wrong one.
Read →