TD Differential

Tom DeMark's three-close momentum-divergence reversal pattern. Flags an exhaustion-and-reversal candle whose buying or selling pressure has shifted from the prior bar. Combines a direction filter (two consecutive lower / higher closes), a buying-pressure check (close − TrueLow), and a selling-pressure check (TrueHigh − close), both measured against DeMark's true range.

Quick reference

Item	Value
Family	DeMark
Input type	Candle (uses high, low, close)
Output type	f64 — +1.0 buy, -1.0 sell, 0.0 no signal
Output range	{-1.0, 0.0, +1.0}
Default parameters	none — TdDifferential::new()
Warmup period	3
Interpretation	Single-bar reversal signal; momentum shift detector

Formula

Pressure is measured against DeMark's true range, which folds in the previous close (Jason Perl, DeMark Indicators, 2008):

TrueLow[i]   = min(low[i],  close[i-1])
TrueHigh[i]  = max(high[i], close[i-1])
buying[i]    = close[i] - TrueLow[i]
selling[i]   = TrueHigh[i] - close[i]

Buy signal (+1.0) on bar i when:

1. close[i] < close[i - 1]  and  close[i - 1] < close[i - 2]   (two lower closes)
2. buying[i]  >  buying[i - 1]              (buying pressure rises)
3. selling[i] <  selling[i - 1]             (selling pressure falls)

Sell signal (-1.0) on bar i when:

1. close[i] > close[i - 1]  and  close[i - 1] > close[i - 2]   (two higher closes)
2. selling[i] >  selling[i - 1]             (selling pressure rises)
3. buying[i]  <  buying[i - 1]              (buying pressure falls)

Otherwise the output is 0.0. See crates/wickra-core/src/indicators/td_differential.rs.

Parameters

None — TdDifferential::new() takes no arguments.

Inputs / Outputs

Indicator<Input = Candle, Output = f64>. Python: TdDifferential().batch(high, low, close) returns a 1-D array.array('d') (first two bars are NaN). Node: batch(high, low, close) returns number[] (first two bars NaN); update(high, low, close) returns number | null (first two bars null).

Warmup

warmup_period() == 3. The pressure of bar i − 1 needs the close of bar i − 2, so the indicator emits its first value on the third input candle.

Edge cases

• No signal. Most bars emit 0.0 — both buy and sell conditions are strict (four inequalities each). Identical bars emit 0.0 (no_signal_on_neutral_bars).
• Single lower close. One lower close is not enough: the prior bar must also have closed below its predecessor (single_lower_close_is_not_enough).
• Reference fixtures. Closes 10 → 9 → 8.5 with highs 11, 10, 9 / lows 9, 8, 7 give buying 1 → 1.5 and selling 1 → 0.5, so +1.0 (buy_signal_after_two_lower_closes_with_shifting_pressure); closes 8 → 9 → 9.8 with highs 9, 10, 11.5 / lows 7, 8, 9.5 give -1.0 (sell_signal_after_two_higher_closes_with_shifting_pressure).
• Direction conflict. A bar that's both up and down at the same time is impossible; the strict close comparison filters noise cleanly.
• NaN / infinity. Rejected by Candle::new upstream.
• Reset. Clears the two-bar cache (reset_clears_state).

Examples

Rust

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

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let mut td = TdDifferential::new();
    // Two lower closes (10 -> 9 -> 8.5) while buying pressure rises
    // (1 -> 1.5) and selling pressure falls (1 -> 0.5).
    let c1 = Candle::new(10.0, 11.0, 9.0, 10.0, 1.0, 0)?;
    let c2 = Candle::new(9.0, 10.0, 8.0, 9.0, 1.0, 1)?;
    let c3 = Candle::new(8.5, 9.0, 7.0, 8.5, 1.0, 2)?;
    let _ = td.update(c1);
    let _ = td.update(c2);
    let v = td.update(c3);
    println!("signal = {v:?}"); // signal = Some(1.0)
    Ok(())
}

Python

import numpy as np
import wickra as ta

h = np.array([11.0, 10.0, 9.0])
l = np.array([ 9.0,  8.0, 7.0])
c = np.array([10.0,  9.0, 8.5])

td = ta.TDDifferential()
print(list(td.batch(h, l, c)))  # [nan, nan, 1.0]

Node

const wickra = require('wickra');
const td = new wickra.TDDifferential();
console.log(td.batch([11, 10, 9], [9, 8, 7], [10, 9, 8.5])); // [ NaN, NaN, 1 ]

Streaming

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

let mut td = TdDifferential::new();
let candle_stream: Vec<wickra::Candle> = Vec::new(); // your live OHLCV candle feed
for bar in candle_stream {
    if let Some(v) = td.update(bar) {
        if v > 0.0 { /* buy reversal pattern */ }
        if v < 0.0 { /* sell reversal pattern */ }
    }
}

Interpretation

• Single-bar exhaustion. TD Differential flags one bar where price has closed lower (higher) two bars running but the pressure distribution has already shifted — classic exhaustion-bar pattern.
• Pair with Setup / Sequential. TD Differential signals at bar level; pair with Setup or Countdown progress to filter noise and only act when both layers agree.
• Discrete output. The trinary {-1, 0, +1} output is easy to feed into rule-based systems; no thresholds to tune.

Common pitfalls

• Treating every signal as actionable. TD Differential signals appear on individual bars and many are false positives. Use as a filter, not a primary trigger.
• Forgetting the direction filter. The first rule (two consecutive lower closes for buys) is the direction filter; without it the pressure tests would fire on any momentum-shift bar.
• Using the plain bar range. Pressure is measured against the true low / high (folding in the previous close), not close − low / high − close; a gap bar changes the pressures.
• Ignoring DeMark's variants. DeMark also published TD Reverse Differential and TD Anti-Differential with different close patterns; this is the base TD Differential.

References

• Tom DeMark, The New Science of Technical Analysis (1994) — TD Differential.
• Jason Perl, DeMark Indicators (Bloomberg Press, 2008) — the three-close rule and true-range pressure definitions.

See also

• TdSetup — wider momentum-exhaustion signal.
• TdSequential — full Setup + Countdown.
• TdOpen — gap-and-fade reversal sibling.
• TdPressure — volume-weighted pressure oscillator.
• Indicators-Overview — full taxonomy.