VWAP, Support/Resistance, and Wyckoff Guide

This guide explains how to use ta4j’s VWAP, support/resistance, and Wyckoff indicator families together for both backtesting and live trading.

Why these three families belong together

VWAP stack gives value context (fair value, stretch, mean reversion zones).
Support/resistance stack gives location context (where reactions repeatedly happen).
Wyckoff stack gives phase context (accumulation/distribution lifecycle).

Combined, they answer:

Is price cheap or expensive versus local value?
Is price near a meaningful reaction zone?
Is the market in a phase where continuation or reversal is more likely?

Indicator map

VWAP family

VWAPIndicator: rolling VWAP over barCount.
AnchoredVWAPIndicator: VWAP that resets on a fixed anchor or dynamic anchor signal.
MVWAPIndicator: moving average of VWAP.
VWAPDeviationIndicator: price - vwap.
VWAPStandardDeviationIndicator: volume-weighted standard deviation around VWAP.
VWAPBandIndicator: VWAP bands (VWAP +/- multiplier * stdDev).
VWAPZScoreIndicator: normalized deviation (deviation / stdDev).

Support/resistance family

VolumeProfileKDEIndicator: kernel density estimate of volume-at-price.
- getDensityAtPrice(index, price)
- getModePrice(index)
PriceClusterSupportIndicator / PriceClusterResistanceIndicator: weighted price clustering in a rolling lookback.
BounceCountSupportIndicator / BounceCountResistanceIndicator: bounce-frequency-driven zones.

Wyckoff family

WyckoffPhaseIndicator: phase inference engine (cycle + phase + confidence + latest event index).
WyckoffPhase: record with cycle type, phase type, confidence, and latest event index.
WyckoffEventDetector, WyckoffStructureTracker, WyckoffVolumeProfile: internal components used by the phase indicator.

Quick start: compose the stacks

BarSeries series = ...;
NumFactory num = series.numFactory();

ClosePriceIndicator close = new ClosePriceIndicator(series);
VolumeIndicator volume = new VolumeIndicator(series, 1);

// 1) Value context (VWAP)
VWAPIndicator vwap = new VWAPIndicator(close, volume, 20);
VWAPStandardDeviationIndicator vwapStd = new VWAPStandardDeviationIndicator(vwap);
VWAPBandIndicator upperBand = new VWAPBandIndicator(vwap, vwapStd, 2.0, VWAPBandIndicator.BandType.UPPER);
VWAPBandIndicator lowerBand = new VWAPBandIndicator(vwap, vwapStd, 2.0, VWAPBandIndicator.BandType.LOWER);
VWAPDeviationIndicator deviation = new VWAPDeviationIndicator(close, vwap);
VWAPZScoreIndicator zScore = new VWAPZScoreIndicator(deviation, vwapStd);

// 2) Location context (S/R)
PriceClusterSupportIndicator support = new PriceClusterSupportIndicator(close, volume, 150,
        num.numOf("0.25"), num.numOf("0.5"));
PriceClusterResistanceIndicator resistance = new PriceClusterResistanceIndicator(close, volume, 150,
        num.numOf("0.25"), num.numOf("0.5"));
VolumeProfileKDEIndicator profile = new VolumeProfileKDEIndicator(close, volume, 150, num.numOf("0.5"));

// 3) Phase context (Wyckoff)
WyckoffPhaseIndicator wyckoff = WyckoffPhaseIndicator.builder(series)
        .withSwingConfiguration(2, 2, 1)
        .withVolumeWindows(5, 20)
        .withTolerances(num.numOf("0.02"), num.numOf("0.05"))
        .withVolumeThresholds(num.numOf("1.6"), num.numOf("0.7"))
        .build();

Backtesting how-to

See Backtesting for execution options. This section focuses on feature-specific patterns.

1) Gate unstable windows first

These indicators have meaningful warmup periods. Always compute a global unstable window and set it on the strategy.

int unstableBars = Math.max(
        Math.max(zScore.getCountOfUnstableBars(), support.getCountOfUnstableBars()),
        wyckoff.getCountOfUnstableBars());

strategy.setUnstableBars(unstableBars);

2) Example entry logic (accumulation continuation)

int i = series.getEndIndex();
Num price = close.getValue(i);
WyckoffPhase phase = wyckoff.getValue(i);

boolean inAccumulationAdvance = phase.cycleType() == WyckoffCycleType.ACCUMULATION
        && (phase.phaseType() == WyckoffPhaseType.PHASE_D || phase.phaseType() == WyckoffPhaseType.PHASE_E)
        && phase.confidence() >= 0.75;

boolean aboveValue = price.isGreaterThan(vwap.getValue(i));
boolean notOverextended = zScore.getValue(i).isLessThan(num.numOf("2.0"));
boolean nearSupport = price.minus(support.getValue(i)).abs().isLessThanOrEqual(num.numOf("0.5"));

if (inAccumulationAdvance && aboveValue && notOverextended && nearSupport) {
    // candidate long condition
}

3) Use location-aware exits

Profit-taking near upperBand or resistance.
Emergency exits when phase flips to distribution with high confidence.
Risk exits when zScore exceeds your stretch threshold and momentum stalls.

4) Backtest usage tips

Run sensitivity sweeps for lookbackLength, bandwidth, tolerance, and Wyckoff thresholds.
Keep thresholds in Num to avoid precision drift when moving between DoubleNum and DecimalNum.
Add slippage and fees; these setups often trade around congested zones where fills matter.
Validate across trending and ranging regimes; this stack is regime-aware, so sample both.

Live trading how-to

See Live Trading for ingestion/runtime architecture. This section focuses on strategy behavior.

1) Prefer closed-bar decisions

For production bots, evaluate entries/exits on completed bars for deterministic behavior. If you evaluate intrabar, expect more signal churn around VWAP bands and S/R boundaries.

2) Anchored VWAP for event resets

Use anchored VWAP when your execution model depends on event regimes.

Indicator<Boolean> anchorSignal = ...; // true on session open, major news, manual resets, etc.
AnchoredVWAPIndicator anchored = new AnchoredVWAPIndicator(series, anchorSignal);

Recommended anchor sources:

Session boundaries (daily/weekly open).
Structural breaks (range breakout/breakdown).
Operator-driven resets after macro events.

3) Add confidence and confluence filters

In live routing, require at least two independent confirmations to reduce noise:

Wyckoff phase confidence threshold.
VWAP relation (price > vwap for long bias, opposite for short bias).
Price proximity to support/resistance zone.

4) Persist state needed for restart safety

Persist:

Last processed bar timestamp/index.
Last WyckoffPhase and transition index (getLastPhaseTransitionIndex).
Active anchor state if using dynamic anchored VWAP.

This avoids duplicated triggers after restart.

5) Live usage tips

Do not trade while indicators are unstable after startup or re-subscription.
Use narrower position sizing when zScore is already stretched.
Widen cooldowns during choppy ranges (high false retests around VWAP).
Track rejected signals in logs for post-trade analysis.

Suggested parameter baselines

Start here, then tune per instrument/timeframe.

Component	Intraday baseline	Swing baseline
`VWAPIndicator barCount`	20-60	50-120
`VWAPBand multiplier`	1.5-2.0	2.0-2.5
`VolumeProfileKDE lookback`	100-200	150-300
`VolumeProfileKDE bandwidth`	0.25-0.75 (price-unit dependent)	0.5-1.5
`PriceCluster tolerance`	0.1%-0.3% of price	0.2%-0.6% of price
Wyckoff volume windows	5 / 20	10 / 40

Common mistakes

Using these indicators without unstable-bar gating.
Mixing incompatible scales (absolute tolerance values across very different price levels).
Treating Wyckoff phase output as deterministic instead of probabilistic.
Ignoring volume quality for instruments with fragmented/poor feed volume.

Integration checklist

Warmup bars computed from all dependencies.
Fees/slippage configured.
Parameter sweep completed on out-of-sample data.
Live restart behavior tested.
Monitoring alerts for phase flips and stretched VWAP z-scores.