We added 402 instruments, and here is exactly what that is worth
Our scanner now reads 2,175 instruments across 26 markets, up from 1,773.
Almost all of the new ground is the Far East. Japan, Hong Kong and South Korea were a 26-instrument pilot from July; they are now 428. Japan went 8 → 217, Hong Kong 8 → 75, South Korea 6 → 132. Taiwan stays at 4 for now — we could not source a constituent list we trusted, and we would rather add nothing than add names we have not checked.
If you subscribe, these arrive by default. Nothing is being taken away, and /manage lets you untick any market in one click.
What we rejected, and why that is the interesting part
We probed 457 candidates and kept 423. Every one had to clear the same gates:
- Two years of daily bars and ten years of weekly bars, both with at least 300 closed bars. That is not arbitrary: our trend filter uses a 100-period average, and below 300 bars its warm-up error is large enough to matter.
- Median six-month turnover of at least $1M a day, converted to dollars. A signal you cannot get filled on is not a signal.
- High/Low integrity. This one exists because of a mistake we nearly made. Two European instruments passed every check we had in a previous round — full price history, real volume — and then turned out to have daily highs and lows reported equal to the closing price. Our stops come from recent swing highs and lows, so both would have published stop levels with no relationship to the actual price range. Row count is not a data check. We measure the width of every candle now.
34 candidates failed: 31 for short weekly history, 3 for being below the turnover floor. Three of those three were within a rounding error of the line. We dropped them anyway, because a floor you bend is not a floor.
What an expansion is actually worth
Here is the part most coverage announcements leave out.
We measured our own throughput this week: 1,773 instruments produce roughly 0.064 candidate setups per instrument per month. So 402 more instruments is about 26 more candidates, and about 10 more published signals a month in the current market.
That is worth having. It is not a transformation, and it does not make any individual signal better. Adding instruments is a volume lever, not a quality one, and we would rather say that plainly than let a bigger number imply something it does not.
While we are being plain: our volume fell this month
August produced 45 published signals against July's 154. We went looking for why, and the answer was not what we expected.
Roughly a third of the drop is simply fewer setups forming. The rest is our own filter. We drop any signal whose first target sits less than 1% from the entry — too small to be worth acting on. In July that filter rejected 9% of candidates. In August it rejected 60%, in every market, not one.
The obvious move would be to loosen the filter and get the volume back. We checked what that would have bought, because every rejected candidate is tracked in the background exactly so this question can be answered with data rather than instinct. Those held-back signals resolved at a 67.7% win rate and a loss of 0.25R per trade — a near-identical win rate to what we publish, and far worse money, because a target that close is easy to touch and worth nothing when you touch it.
So we are not loosening it. The volume it would buy is made of losing signals, and we would rather send you five signals a week than fifteen that are quietly negative.
That is also why 402 new instruments matter: expanding the universe is the one way to add signal volume that does not come at the cost of signal quality.
Talk to us
Questions, comments or suggestions are welcome — send them to info@algosignals.finance. If there is a market you want covered, say so; it goes on the list for the next round.
Not financial advice.
⚠️ Not financial advice.