TD Risk Level

Tom DeMark's protective-stop levels derived from setup extremes. Computes a quantitative stop level for trades taken on the back of a completed setup, using the bar that made the most-extreme price during the setup run and that bar's true range. Buy risk sits one true-range below the lowest low of the buy setup; sell risk one true-range above the highest high of the sell setup.

Quick reference

Item	Value
Family	DeMark
Input type	Candle
Output type	TdRiskLevelOutput { buy_risk, sell_risk }
Output range	unbounded (price-units), or NaN
Default parameters	lookback = 4, target = 9 (same as TdSetup)
Warmup period	lookback + 1
Interpretation	Quantitative stop level for setup-driven entries

Formula

For a completed buy setup:
    low_extreme_bar = bar with the lowest low among the setup's 9 bars
    tr_extreme     = true_range of that bar
                   = max(high - low,
                         |high - prev_close|,
                         |low - prev_close|)
    buy_risk = low_extreme_bar.low - tr_extreme

For a completed sell setup (mirror):
    high_extreme_bar = bar with the highest high among the setup's 9 bars
    tr_extreme       = true_range of that bar
    sell_risk = high_extreme_bar.high + tr_extreme

Either field is f64::NAN until the first setup in that direction completes. Once set, the level stays until the next setup in that direction completes. See crates/wickra-core/src/indicators/td_risk_level.rs.

Parameters

Name	Type	Default	Constraint	Description
lookback	usize	4	> 0	Setup-phase close-vs-close lookback.
target	usize	9	> 0	Setup completion target.

TdRiskLevel::new returns Error::PeriodZero for zero arguments.

Inputs / Outputs

Indicator<Input = Candle, Output = TdRiskLevelOutput> with two fields (buy_risk, sell_risk).

• Python. (n, 2) float64 array, columns [buy_risk, sell_risk] with NaN-filled warmup.
• Node. Flat number[] of length n * 2, interleaved.

Warmup

warmup_period() == lookback + 1. Until at least one setup completes in each direction, the corresponding field is NaN.

Edge cases

• No setup completed yet. Both fields are NaN.
• True range using previous close. The implementation correctly uses the prior bar's close, not the extreme bar's own close, in the true-range calculation — matching DeMark's published methodology.
• NaN comparison. Downstream code must explicitly check is_nan() before using the level.
• Reset. Clears all internal state.

Examples

Rust

use wickra::{BatchExt, Candle, Indicator, TdRiskLevel};

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let candles: Vec<Candle> = (0..30).map(|i| {
        let c = 100.0 - f64::from(i) * 0.5;
        Candle::new(c, c + 0.3, c - 0.3, c, 1.0, i as i64).unwrap()
    }).collect();
    let mut tr = TdRiskLevel::new(4, 9)?;
    if let Some(o) = tr.batch(&candles)[25] {
        println!("buy stop  = {}  (NaN if no buy setup yet)", o.buy_risk);
        println!("sell stop = {}", o.sell_risk);
    }
    Ok(())
}

Python

import numpy as np
import wickra as ta

close = 100 - np.arange(30, dtype=float) * 0.5
tr = ta.TdRiskLevel(4, 9)
print('row 25:', tr.batch(close + 0.3, close - 0.3, close)[25])

Node

const wickra = require('wickra');
const tr = new wickra.TdRiskLevel(4, 9);
const close = Array.from({ length: 30 }, (_, i) => 100 - i * 0.5);
const flat = tr.batch(close.map(c => c + 0.3), close.map(c => c - 0.3), close);
console.log('row 25 buy stop:', flat[25 * 2]);

Streaming

use wickra::{Candle, Indicator, TdRiskLevel};

let mut tr = TdRiskLevel::new(4, 9).unwrap();
let candle_stream: Vec<wickra::Candle> = Vec::new(); // your live OHLCV candle feed
for bar in candle_stream {
    if let Some(o) = tr.update(bar) {
        // If you're long on a buy setup, exit when close drops below buy_risk
        if !o.buy_risk.is_nan() && bar.close < o.buy_risk {
            // stop out
        }
    }
}

Interpretation

• Protective stop. TD Risk Level is DeMark's recommended stop for trades taken on a completed setup. The one-true-range buffer below the setup's lowest low (for longs) acknowledges that the actual extreme may need room to breathe.
• Less aggressive than TDST. Compared to TdLines' TDST levels (which mark thesis-invalidation), TD Risk Level is a hard exit stop. TDST is a "the bullish setup is wrong" signal; TD Risk Level is a "get out at this price".
• Pairs with Sequential entries. Use as stop on Countdown-13 entries; if the move resumes through the level, the exhaustion thesis was wrong.

Common pitfalls

• Forgetting NaN. Same as TdLines — fields are NaN until the first setup in that direction completes.
• Confusing with TDST. TDST is the thesis-invalidation level (highest high of buy setup); Risk Level is the protective stop (low − TR of the extreme bar). Different levels with different trade meaning.
• Wrong direction usage. buy_risk comes from buy setups (the lowest low + buffer); sell_risk from sell setups. Don't swap.

References

• Tom DeMark, DeMark Indicators (with Jason Perl, 2008) — protective-stop methodology.

See also

• TdSetup — the setup phase Risk Level is derived from.
• TdSequential — pairs with Risk Level for completed entries.
• TdLines — alternative invalidation level (TDST).
• Atr — the volatility unit used in the true-range buffer.
• Indicators-Overview — full taxonomy.