Williams VIX Fix: Formula, Bands and Pineify Setup
The Williams VIX Fix measures the distance between the current low and the highest recent close, as a percentage of that highest close. Chris Moody's CM_Williams_Vix_Fix publication attributes the idea to Larry Williams and adds threshold-based highlighting.
A large reading identifies a large downward price excursion relative to the selected reference. It does not establish that selling is exhausted or that the next trade will be profitable.
Calculate the raw WVF correctly
Let H be the highest closing price over the lookback, including the current bar, and L be the current low:
WVF = 100 × (H − L) / H
This is not highest high versus current close. For an illustrative highest close of 100 and current low of 94, WVF is 6. If the current low falls to 90 while H stays 100, WVF becomes 10. The change says nothing about the subsequent return.
The denominator must be meaningful and nonzero. Ordinary positive-price interpretation should not be applied mechanically to zero or negative-priced series. A 22-bar lookback means 22 daily bars on a daily chart, but only 22 hourly bars on an hourly chart.
Why this is different from the VIX
Cboe's VIX methodology uses SPX options to estimate 30-day expected volatility. WVF uses the chart's close and low. It neither recovers that options calculation nor measures the standard deviation of realized returns.
"Fear gauge" is an informal interpretation, not a direct observation of investor sentiment. A WVF value of 20 and a VIX value of 20 are not interchangeable measurements. The ability to calculate WVF without options data also does not prove it detects bottoms in every asset class.
Bands and range thresholds
Pineify's current component calculates an upper band as the moving average of WVF plus a multiplier times its standard deviation. Both the average and standard deviation use the Bollinger Band Length. The band is applied to WVF, not to the price chart.
The component also calculates:
- Range High = highest WVF over the range lookback ×
ph. - Range Low = lowest WVF over that lookback ×
pl.
Despite the input labels containing "Percentile," these are scaled extrema. A factor of 0.85 means 85% of the highest value, not an empirical 85th percentile and not an 85% chance of a reversal.
For example, if the highest WVF in the window is 10, Range High at 0.85 is 8.5. How many observations exceed 8.5 depends on the data; the formula does not force 15% to do so.
The default histogram is highlighted lime when WVF is at least the upper band OR at least Range High. Otherwise it uses gray. A hidden band can still participate in this color condition. A lime bar therefore need not mean both thresholds were exceeded.
Configure the Pineify component
Search for Williams VIX Fix in Pineify. The current catalog marks it as a paid feature. Check your account entitlement before assuming free access.
| Control | Current default | Calculation affected |
|---|---|---|
Lookback period (pd) | 22 | Highest close used in raw WVF |
Bollinger length (bbl) | 20 | WVF mean and standard deviation |
Band multiplier (mult) | 2.0 | Distance of the upper band from its mean |
Range lookback (lb) | 50 | WVF maximum and minimum |
Highest factor (ph) | 0.85 | Scaled maximum |
Lowest factor (pl) | 1.01 | Scaled minimum |
The display can show the range lines and upper standard-deviation band; both are hidden by default. The outputs expose WVF, Upper Band, Range High and Range Low. There is no "Highs Not Lows" inversion switch in this component.
One implementation limitation needs attention: the UI presents the band multiplier as a decimal, but the current calculation function declares it as an integer parameter. This page does not recommend fractional multiplier settings as validated configurations. Inspect and compile the generated script before relying on such a change.
Export the script and use TradingView's Pine Editor to save and add it to a chart. Code generation does not guarantee a successful compilation or a profitable strategy. Compare a few raw values and highlighted bars with the formulas above.
Turn a visual observation into a test
Choose the event first. A value above a threshold, a first crossing above it, and a return below it are three different conditions. Several consecutive lime bars can belong to one prolonged price decline; treating them as independent bottom predictions can distort an evaluation.
For a hypothetical study of first threshold crossings, record each qualifying timestamp and the price changes over predefined horizons. Include events followed by further declines. If you then turn the study into a strategy, specify actual order timing, exits, sizing and transaction costs separately.
A lower WVF peak during a later price low can arise from changes in the rolling highest close as well as the current low. Check the denominator before interpreting divergence as reduced selling pressure.
Keep the unfiltered rule as a comparison when adding support levels, RSI or another timeframe. Those additions do not establish a higher success probability without an evaluation using the same data and execution assumptions. TradingView's strategy documentation explains the simulated-fill model.
Live-bar behavior
The current low can fall while a candle is open, and the current close can change the highest-close reference. WVF and its thresholds can therefore change intrabar. Follow TradingView's repainting guidance when deciding whether your rule uses live or completed bars.
A close-based rule changes timing; it does not prove that a bottom has formed. This page provides no independently supported 62% hit rate, SPY or QQQ trade result, or best daily/4-hour setting.
Frequently asked questions
Does WVF use the highest high and current close?
No. The calculation described here uses the highest close over the lookback and the current low.
Does a lime bar mean the bottom is confirmed?
It means WVF met at least one of the configured upper thresholds. Price can continue falling afterward.
Is the Highest Percentile input an actual percentile?
In this implementation it multiplies the rolling maximum. It does not rank observations or compute an empirical quantile.
What settings are best for crypto?
No verified comparison here establishes a best preset. Record the feed, bar interval, lookbacks and evaluation rule before comparing configurations.