How a turn marker is decided
What a marker claims, what it does not, and how to read the two states it comes in — because a marker you cannot reason about is a marker you cannot use.
The idea: read the lean, then find where it runs out
Sentiment on foreign exchange is awkward to buy and easy to read. There is no central exchange, so there is no honest volume and no order book that covers the whole market — which is why the cleanest available measure of which way the crowd is leaning is the price itself.
The combination of fading volatility, market sentiment, and order clustering reveals vulnerable levels where, once reached, a cascade of triggered orders can propel the price further.
Confirmed and unconfirmed
Every marker carries one of two states, and the difference is worth understanding because it is the difference between the condition fired and the market immediately agreed.
| ▲ confirmed ▼ confirmed |
The very next bar traded through the marker bar's own extreme — above its high for an up marker, below its low for a down one. Not the close: any trade through the level counts. |
|---|---|
| ▲ unconfirmed |
Everything else. Two quite different situations share this state, and the
chart shows them the same yellow:
|
It is decided once, and it does not flip back
A marker can move from unconfirmed to confirmed while the next bar is still open. Once that bar closes, the state is final: a confirmed marker stays confirmed no matter what price does afterwards, and an unconfirmed one is never upgraded later by a break three bars down the road. Confirmation is a statement about one specific hour, not a running score.
Why most of the chart is yellow
About a third of markers get confirmed. The test is deliberately demanding: it asks a market that has just thrust in one direction to reverse hard enough, within a single hour, to take out the extreme of the bar that fired the signal. Most turns are not that abrupt. A high confirmation rate would mean the test was weak, not that the markers were better.
What it does and does not tell you
- It is not a quality score. An unconfirmed marker is not a failed one. The exhaustion condition fired either way; confirmation only records whether the reversal was immediate.
- It is not a forecast. Confirmation is a fact about the bar that already closed, not a claim about the next ten.
- The statistics use confirmed markers only. The numbers under the symbol selector describe what price did after confirmed markers, which is why they are measured on a smaller set than the chart shows.
What the statistics on the dashboard mean
Switching instrument shows what price did after past confirmed markers on it, measured in R — the height of the marker's own bar, which is the risk implied by a stop at its other end.
Everything on that line assumes the stop: the run is measured only until price comes back through the other end of the marker's bar, and a target counts only if it was reached before that happened.
- Best run before the stop — the average of the furthest price got in the marker's direction before the stop would have ended it.
- Reached +1 / +2 / +3 R — how often that much was available, again before the stop.
- The sample — how many confirmed markers the averages are taken over, how they split between up and down, and the years covered.
These are descriptions of history on a real broker feed, with the ambiguous bars resolved down to the minute. They are not a backtest: there is no entry rule, no exit rule, no spread and no position sizing in them, and they are not a claim about what anyone would have earned.
Times and clocks
Three clocks, each with one job. The trading session and the day cut run in the broker's clock, whose midnight is the 17:00 New York rollover — the real boundary of the FX day. Everything is stored in UTC, the only timeline that never repeats or skips an hour. Your browser renders it in your own local time.