Features — Sessions Alpha

Detection

What it watches.

Six session ranges

Asia, pre-London, the New York morning, plus a spare slot. Each records its own high and low and watches for a sweep of either edge. Turn off the ones you do not trade.

Session highs and lows

Drawn and kept live until something takes them out. Any of them can arm a setup.

Hourly liquidity pivots

Swing highs and lows read from the hourly chart regardless of the timeframe you are on, so the levels do not change when you zoom.

Previous day and week

Highs, lows and opens, with lines that stop drawing once the level has been cleared.

Daily and weekly midpoints

Reference levels, and the basis of the optional premium / discount filter.

Filters

What it refuses.

Minimum gap size

In points, or as a multiple of ATR so it scales with volatility. Gaps under the threshold still draw as dimmed markers with the shortfall printed, so you see what was excluded.

Gap must form near the sweep

A gap that forms long after the sweep belongs to a different move. Without this bound a setup can be claimed by an unrelated gap far away, leaving an entry stranded from its own stop.

Max sweep depth

How far beyond the range price may travel and still count as a sweep rather than a breakout.

Minimum R:R

Rejects a setup where the stop is too far relative to the room left inside the range.

Turtle soup confirmation

Optional. Requires a prior 5-minute extreme to have been taken and closed back through.

Inverted FVG confirmation

Optional. Requires a gap from the sweep leg that price has closed back through.

Attempts per sweep

Whether a failed setup gets a second try, and on the Sweep model the next only arms once the previous trade has closed.

Session and no-trade windows

Restrict when entries may fire without restricting when the sweep may happen, and skip the minutes around the New York open where a one-minute candle can mean nothing.

The settings panel — Part of the settings panel. Everything is grouped by what it does to your trading rather than by where it sits in the code.

Risk

Where the stop goes.

Three stop modes and two target modes. None of them changes how many signals you get — they change what happens on the signals you were already getting.

Stop on the far edge of the gap — Far edge of the inverted gap — the tightest of the three, and logically the right level. If price closes back through the gap the other way, the inversion failed and the idea is dead.

Stop at the sweep extreme — The identical signal with the stop at the sweep extreme. Wider, survives a retest of the low, and costs several times the risk to earn the same R. Neither is safer — they move the same risk between frequency and size, and which you want depends on your drawdown rules far more than on the market.

ATR stop

Sizes from volatility instead. Use knowingly — it ignores the setup, so on a deep sweep it can land inside the sweep leg.

Fixed R target

Predictable and easy to size against.

Nearest liquidity target

Aims at the closest untouched level on the profit side. Often pays more than a fixed target, but you cannot know your R before you take the trade.

Output

What it gives you.

Captioned signals

What was swept, how many bars ago, how far price has travelled since, and the size of the gap that inverted. CRT signals also name the range and list which confirmations fired.

Tracked stop and target

Drawn at entry, closed at the exit, with the R outcome printed beside them.

Alerts with live prices

Plain text or JSON for a webhook, carrying entry, stop, target and R, plus an optional account tag so a receiver can route the order.

Sweep alerts

Separate from the model alerts. Fires when a level is traded through — a heads-up that a setup may be forming.

No repainting

Every model decision is made on a closed bar. A wick that pokes through a level and recovers cannot consume a setup or fire an alert.

Timeframe notice

The CRT model runs on 1-minute charts only. Anywhere else it switches off and says so at the bottom of the chart, rather than silently doing nothing.