Pivot points: formulas, variants and strategy tests

In a hypothetical session, a trader can mark pivot levels calculated from earlier prices. A subsequent reaction does not by itself establish predictive accuracy or a profitable trade after costs.

Pivot inputs: formula, period and data

Pivot points are reference levels calculated from a specified set of prices. Record the formula, instrument, price series, timezone and session boundaries. A daily period can cover different hours at two providers; daily and intraday feeds can also have different OHLC. In futures, settlement is not necessarily the last trade.

The Traditional formulas used below take the preceding period’s High H, Low L and Close C. Their central mean P is a reference, not a forecast of where the next session will trade. R1 − S1 equals H − L, but equal width does not make the R1–S1 interval identical to the earlier Low–High interval.

Two traders need not see identical pivot levels. Compare their formula, session, feed, settlement/last-trade choice and missing-data rules before interpreting a discrepancy. Knowing a calculated level does not reveal who has orders there or demonstrate that greater popularity makes a reaction more reliable.

Traditional: P = (H + L + C)/3

For this Traditional variant, P = (H + L + C) / 3. Use complete preceding-period observations and retain precision through all derived calculations. A price above P satisfies an above-P condition; the formula does not establish a bullish outcome.

Subtracting S1 from R1 gives H − L exactly. With the model below, both widths are 80 pips, yet S1 is above the earlier Low and R1 above the earlier High. The gap R2 − R1 is H − P; S1 − S2 is P − L, so those gaps depend on Close. No formula constrains the next period to R3/S3 or states how often either will be reached.

Calculating pivot points by hand — a concrete example

Under the Traditional formulas used here, for H = 1.0890, L = 1.0810 and C = 1.0860, PP = 1.085333... . Keep full precision for intermediate calculations, then round outputs: R1 = 1.0897, S1 = 1.0817, R2 = 1.0933, S2 = 1.0773, R3 = 1.0977 and S3 = 1.0737 to four decimals. Rounding PP first changes the derived levels.

For this invented EUR/USD Bid example, the input period is 22 September 2026 00:00 UTC to 23 September 00:00 UTC, excluding the right endpoint. These are not historical quotes. Use its completed H/L/C for the next analogous period after the data arrive. Do not assume every broker uses New York close. Fixed input data yield fixed levels; a feed revision can change them without a new observation-period move.

Observation rules are separate from trade rules

A bounce, breakout and reversal are different research conditions. Define them before evaluation and do not infer a win rate or reward/risk ratio simply from the location of a line.

1. Return above S1. Our original completed-M15 filter requires Open > S1, Low ≤ S1 and Close > S1. With exact S1 = 1.081666666… , a fictional 23 September bar 09:00–09:15 UTC has O/H/L/C = 1.08190/1.08210/1.08140/1.08180 and passes all three. A temporary return at 09:07 cannot establish the eventual Close.
2. Crossing a chosen level. Specify whether the condition concerns a trade, a quote or a completed Close, and when any order becomes possible. A move above P does not prove that the rest of the day will rise, nor does it establish a lower win rate than a bounce rule.
3. Data-revision counterexample. Keep H/L fixed but increase prior C by 3 pips from 1.0860 to 1.0863. P, R2 and S2 rise 1 pip; R1, S1, R3 and S3 rise 2 pips. New S1 = 1.081866666… is above the unchanged M15 Close 1.08180, so the return-above filter fails. That change comes from revised input data, not a new bar path.

Camarilla: selected levels around Close

The Camarilla levels shown here use the prior Close plus or minus fractions of the preceding range. We show the third and fourth pairs; TradingView also documents further levels. These names identify arithmetic, not a known maximum daily range or mandatory signals.

A method can test fading or breaking a chosen Camarilla level, but it must state entry, invalidation and execution conditions independently. Tighter numerical spacing does not establish scalping performance, typical touch frequency or a predictable breakout range.

Woodie: check which input receives double weight

TradingView Woodie uses the current period Open: P = (H + L + 2O)/4. With the earlier H/L and invented new Open O = 1.0870, P is 1.08600. Current O must be received before this value can be used; its line cannot be treated as known before that opening observation.

A close-weighted alternative P = (H + L + 2C)/4 instead gives 1.08550. That is a different input formula, not rounding error and not TradingView’s Woodie setting. State the exact formula rather than relying on a platform label.

TradingView Traditional and Classic share P, R1/S1 and R2/S2, but Classic has R3/S3 = P ± 2(H − L), here 1.10133/1.06933 to five decimals. Record the pivot period explicitly: TradingView Auto can change it when the chart timeframe changes. Sierra also documents skipping insufficient-data periods; missing observations can change which period supplies the calculation.

Fibonacci: range multipliers, not accuracy rankings

Fibonacci pivots keep the classical pivot calculation but apply different multipliers to the support and resistance levels. Instead of the full daily range, they use Fibonacci proportions: 0.382 for the first level, 0.618 for the second, and 1.000 for the third. R1 becomes Pivot + 0.382 × (H − L), R2 is Pivot + 0.618 × (H − L), and R3 is Pivot + 1.000 × (H − L). The supports mirror these down from the pivot.

For an invented 40-pip range, the first two Fibonacci offsets are 0.382 × 40 = 15.28 and 0.618 × 40 = 24.72 pips; Traditional R2 is 40 pips above P. For the earlier 80-pip EUR/USD model, Fibonacci R1/S1 are 1.08839/1.08228 to five decimals. Narrower spacing is an algebraic difference, not evidence of better EUR/CHF entries or a typical range.

Compare variants on identical data and costs with a separate evaluation period. Do not select whichever line happened to match a successful reaction. The formulas alone do not rank GBP/USD, XAU/USD or quiet crosses, or predict that a stop will be taken out by the next swing.

Pivot levels require an exact formula, session boundary and data source; familiar levels do not guarantee reactions.

Common mistakes and what is really worth taking away

A zone width is another test parameter. Define its distance, quote side and equality rule before observing outcomes. No universal 10–15-pip EUR/USD band or four-pip stop-failure rate follows from these formulas.

A touch is an observation, not a complete strategy. Entry, size, stop/target, costs, missing-data and publication rules still need definitions. Adding a candle or RSI condition changes the test; it does not itself validate an advantage.

Macro releases can change volatility, liquidity and execution costs. Specify whether the strategy trades such events and how position size, stop placement and gap risk are handled. Doubling a stop without reducing volume increases planned loss and is not a universal protective rule.

Keep calculation and trading outcome separate. In an ideal two-outcome model with loss 1 unit and gain 1.5 or 2 units, the no-cost breakeven win rates are 40% and 33.33% respectively; actual costs and variable outcomes change them. Neither those ratios nor a familiar pivot establishes the actual win rate or net expectancy of a rule.

Related reading: how to draw support and resistance — without them, pivot points lack full context; Fibonacci retracement — a natural complement to pivot levels for in-trend pullbacks; the complete breakout procedure — an alternative approach to breakouts that fits the logic of the Camarilla H4 and L4 levels.