VALYNOX
VALYNOXSTRATEGY RESEARCH TECHNOLOGY
AlphaPrimeAlphaPrime
ACADEMY / Indicators and signals

Pivot points trading: levels, rules and an ES futures test

Daily pivot points are seven price lines for today built from yesterday's high, low and close. Price reacts near them often enough to be worth testing, but the result depends on whether you fade the touch or trade the break, which session built the levels, and the cost of each trade.

By · Published · updated · Editorial policy

Contents
  1. Do daily pivot points work as support and resistance in futures?
  2. How do you calculate pivot points? The Classic formula and its levels
  3. Floor, Fibonacci and Camarilla pivots: what changes between the variants?
  4. How accurate are pivot points: bounce or breakout?
  5. Day session or full session: which high, low and close go into the pivot?
  6. What does a pivot point rule show in an ES backtest?
  7. How to test a pivot rule on your own futures data
  8. Frequently asked questions
  9. References

Do daily pivot points work as support and resistance in futures?

Pivot points trading starts from seven price lines for today, computed from yesterday's high, low and close. On ES futures, price pauses or turns near them often enough to be worth testing, but whether that pays depends on whether you fade the touch or trade the break, which session built the levels, and the cost of each trade.

A pivot point is a reference price for today calculated only from yesterday's completed bar, so it is fixed before the session opens and does not move during the day. Support is a price area where falling prices tend to pause because buyers step in; resistance is the mirror image, where rising prices tend to pause. Most indicators are drawn from a moving window of recent prices and change with every bar. Pivot levels do not. That makes them easy to test: the level is the same for every trader using the same formula and the same bar, like a road map printed the night before rather than a map that redraws itself while you drive.

The usual explanation for why price reacts at pivots is that many traders and order algorithms watch the same lines, so orders cluster there. Whether that clustering is strong enough to pay for a trade after commission and slippage is the question this article tests. A level that "holds" in a chart screenshot is not evidence; a count of holds and breaks per level on your own data is. The same standard applies to any line on a chart, from pattern necklines to channel edges, as the overview of which technical analysis tools and indicators can be tested explains. The point I keep coming back to: the pivot level itself decides nothing; the rule you attach to it does.

Yesterday's high, low and close build a ladder of seven lines for today, R3 to S3 around the pivot P, and today's price path touches S1 and R1 Seven lines from one bar Yesterday's high, low, close set today Yesterday High Close Low R3 R2 R1 P S1 S2 S3 Every level is known before the open. So: test the rule, not the line.
Today's pivot ladder is drawn entirely from yesterday's bar, so every level is known before the open.

How do you calculate pivot points? The Classic formula and its levels

The Classic pivot point is the average of yesterday's high, low and close, and the six other levels are placed above and below it using yesterday's range. The only formula you need is the one for the central pivot:

P = (H + L + C) ÷ 3

Here P is today's pivot point, H is yesterday's high, L is yesterday's low and C is yesterday's close. Yesterday's range is H minus L. The other levels are built from P and that range, as the table shows in plain words.

LevelHow it is builtWhat it marks
R1Twice P, minus yesterday's lowFirst resistance above P
S1Twice P, minus yesterday's highFirst support below P
R2P plus yesterday's rangeSecond resistance
S2P minus yesterday's rangeSecond support
R3Yesterday's high plus twice (P minus yesterday's low)Third resistance
S3Yesterday's low minus twice (yesterday's high minus P)Third support

Hypothetical example

Yesterday's ES bar: high 5,060, low 5,000, close 5,040. The range is 60 points. P = (5,060 + 5,000 + 5,040) ÷ 3 = 5,033.3. R1 = 2 × 5,033.3 − 5,000 = 5,066.7 and S1 = 2 × 5,033.3 − 5,060 = 5,006.7. R2 = 5,033.3 + 60 = 5,093.3 and S2 = 5,033.3 − 60 = 4,973.3. R3 = 5,060 + 2 × 33.3 = 5,126.7 and S3 = 5,000 − 2 × 26.7 = 4,946.7. Because yesterday closed in the upper part of its range, P sits above the middle of the range and R1 is further from P than S1 is.

Hypothetical ES bar in points from P: high +26.7, low −33.3, close +6.7, range 60. From P and the range, R1 is +33.3, R2 +60, R3 +93.3, S1 −26.7, S2 −60 and S3 −86.7, each shown with its formula Every rung grows from P Hypothetical ES bar, points from P High +26.7 Low −33.3 Close +6.7 Range 60 R3 +93.3 H + 2(P − L) R2 +60 P + range R1 +33.3 2P − L P 0 (H + L + C) ÷ 3 S1 −26.7 2P − H S2 −60 P − range S3 −86.7 L − 2(H − P) Wider range, wider ladder. So: check the range before trading.
Every Classic level grows out of P and yesterday's range, with a resistance rung for each support rung.

Two things follow from the formula. The wider yesterday's range, the wider today's ladder, so a quiet day produces tight levels and a wild day produces levels that price may never reach. And where yesterday closed inside its range decides whether P leans toward the high or the low, which shifts every rung with it.

Floor, Fibonacci and Camarilla pivots: what changes between the variants?

The variants all start from the same high, low and close, but they space the levels differently, so the same input bar gives a different ladder. Floor pivots are the Classic formula under the name used on trading floors; P, R1, S1, R2 and S2 are the same, and some charting packages compute R3 and S3 slightly differently, so check your platform's formula before comparing results. Fibonacci pivots keep the same P but place the levels at 38.2, 61.8 and 100 percent of yesterday's range above and below it. Camarilla pivots are built around yesterday's close instead of P, with four levels on each side at fractions of the range, and the inner ones sit only a few points from the close.

FeatureClassic / FloorFibonacciCamarilla
Centre of the ladderP (average of H, L, C)P (average of H, L, C)Yesterday's close
Spacing of levelsFull range, mirrored around P38.2, 61.8, 100 percent of the rangeSmall fractions of the range, four levels each side
Width of the inner levelsWideMediumTight
Typical useTargets and breakoutsTargets inside the day's expected rangeFading small moves near the close

Hypothetical example

Same bar as above: high 5,060, low 5,000, close 5,040, range 60 points. Classic gives R1 at 5,066.7 and S1 at 5,006.7. Fibonacci gives R1 at 5,033.3 + 0.382 × 60 = 5,056.3 and S1 at 5,010.4, with R3 and S3 at a full range from P (5,093.3 and 4,973.3). Camarilla's first levels sit at 5,045.5 and 5,034.5, only 5.5 points from the close, and its fourth levels at 5,073 and 5,007. A trade that fades the first Camarilla level risks a few points; a trade that fades Classic R1 waits for a move of 26.7 points above the close.

For the same hypothetical bar, Classic places its first resistance 26.7 points above the close, Fibonacci 16.3 points and Camarilla only 5.5 points, with Camarilla's levels clustered tightly around the close Same bar, three ladders Dashed line: yesterday's close Classic Fibonacci Camarilla R1 +26.7 pts R1 +16.3 pts R1 +5.5 pts Tight levels mean more, smaller trades. So: pick by the move you expect.
One input bar, three ladders: the variant decides how far from the close the first trade can happen.

The practical consequence is about costs. Tight Camarilla levels produce many touches per day and small moves per trade, so commission and slippage take a larger share of each trade. Wide Classic levels produce fewer trades with more room, but price reaches R2 or S2 on fewer days. Pick the variant by the move you expect to capture per trade, then test that choice rather than switching variants after seeing a result.

How accurate are pivot points: bounce or breakout?

Accuracy has no single answer, because a bounce rule and a breakout rule bet on opposite outcomes at the same level. A bounce rule sells a touch of R1 or buys a touch of S1 and aims for P. A breakout rule buys a close above R1 or sells a close below S1 and aims for the next rung. Every time R1 holds, the bounce rule is right and the breakout rule is wrong. So the question "are pivot points accurate" only has a meaning once you say which rule, which level and which market.

QuestionBounce rule (fade the touch)Breakout rule (go with the close beyond)
EntryShort at a touch of R1, long at a touch of S1Long on the bar after a close above R1, short after a close below S1
ExitP, or a stop beyond the levelR2 or S2, or a stop back inside the level
What counts as successThe level holdsThe level breaks and price keeps going
Kind of day it needsRange day inside yesterday's ladderTrend day that leaves the ladder
Cost sensitivityHigh: short moves, many tradesLower per trade, but more false breaks after costs
At the same R1 line, the bounce rule shorts the touch and exits at P, winning when R1 holds, while the breakout rule goes long after a close above R1 and exits at R2, winning when R1 breaks Same line, opposite bets A hold at R1 helps one rule only Bounce rule R1 P Short Exit Wins if R1 holds Breakout rule R2 R1 Long Exit Wins if R1 breaks One rule's win is the other's loss. So: count holds and breaks first.
The same R1 line, two opposite bets. The hold count that proves one rule is the failure count of the other.

The way to get a number is to count. For each level and each day, record whether price came within a set distance of the level, whether it then closed back on the near side (a hold) or beyond it (a break), and how far it travelled afterwards. The hold rate at R1 is the raw accuracy of the bounce rule there; one minus that rate is the raw accuracy of the breakout rule. Neither is a profit figure until you add how far the winners went, how far the losers went and what each trade cost. A rule that holds 60 percent of the time but loses more on the breaks than it makes on the holds is not an edge.

Pivot breakouts are close relatives of other range-based rules. The Donchian channel breakout rules trade a close beyond the highest high or lowest low of the last N days; a Classic R2 breakout trades a close beyond a level that stands a full range above yesterday's pivot. Researchers have tested rules of this family for decades: Brock, Lakonishok and LeBaron (1992) studied trading range break rules on a stock index, and the survey by Park and Irwin (2007) found that results across such studies are mixed and often shrink once transaction costs and the number of rules tried are taken into account. Bounce traders often add a filter so they only fade a touch when momentum is already stretched; the guide on how to test RSI rules shows how to test such a filter as a separate step instead of assuming it helps.

Day session or full session: which high, low and close go into the pivot?

The same calendar day produces two different pivot ladders depending on whether the bar covers the full electronic session or only the day session, so fix the session definition before any test and keep it unchanged in live trading. For ES, the CME Globex session runs from 17:00 Chicago time to 16:00 the next day with a one-hour break, while the window most charting platforms label the day session runs from 08:30 to 15:15 Chicago time, covering the US cash equity hours. Exchanges can change these hours, so check the contract page on the exchange website; the free futures market hours tool shows the current session times in your own time zone.

The full-session bar usually has a higher high or a lower low than the day-session bar, because overnight trading extends the range. Its close is also different: the last trade before 16:00 is not the last trade at 15:15, and neither is the exchange's daily settlement price. Three candidate closes mean three candidate values of P, and every rung moves with it.

ChoiceFull electronic sessionDay session only
Range usedIncludes overnight highs and lowsUS cash hours only
Close usedLast trade before 16:00 Chicago time, or settlementLast trade at 15:15 Chicago time
Typical ladder widthWiderNarrower
Who shares the levelsTraders on 24-hour chartsTraders on day-session charts
Risk in testingPlatform default may use a different close than your live chartOvernight moves can open the day already beyond R1 or S1
ES full electronic session runs 17:00 to 16:00 Chicago time with the day session 08:30 to 15:15 inside it; the full-session bar has a wider range and gives a wider pivot ladder with a different P than the day-session bar Session moves every level ES, one date, Chicago time Full session 17:00–16:00 Day session 08:30–15:15 Full R1 P S1 Day R1 P S1 Same date, two different ladders. So: test and trade one session.
One date, two bars, two ladders. The session definition moves every level before the first trade is taken.

Two further pitfalls sit here. First, a continuous contract built from several expiries will, on the roll day, compute today's pivot from a bar of the old contract while you trade the new one; decide in advance whether to skip that day or to adjust. Second, a day-session test that enters at the open may find that the overnight move already placed price beyond R1, so the "breakout" happened while no bar of your data existed. Whatever you choose, the backtest and the live chart must use the same session template and the same closing price, or the levels you tested are not the levels you trade.

What does a pivot point rule show in an ES backtest?

The run below shows one kept candidate from a search of 1,024 pivot rules on ES daily bars, and its numbers describe that selection, not what pivot points earn. AlphaPrime (Valynox) tested each candidate on ES daily bars from 2019-01-02 to 2025-06-30, long and short, one contract, with a Pivot Point rule as its only entry and exit, no stop, no target and no filters. Because the bars are daily, this is not an intraday fade of a touch; the rule reads the daily bar against the pivot levels and holds positions across days. The one candidate kept out of the 1,024 tried was chosen after the fact for its return against its drawdown, and picking the best of many after seeing the results flatters it. Bailey, Borwein, López de Prado and Zhu (2014) show that the best backtest in a large search is expected to look better than the rule behind it will on new data.

Source: AlphaPrime product engine (CPU), run 2026-10-31. ES daily bars, 2019-01-02 to 2025-06-30, long and short, 1 contract, $20 commission and $25 slippage per trade.

NumberResult
Net profit (closed trades)$119,422.5
Max drawdown$35,290
Trades297
Win rate62.63%
Profit factor1.391
Average trade$402.1
Time in market71.46%
The equity curve of the kept candidate with its drawdowns. Compare the slope of the marked out-of-sample stretch with the slope before it, and the depth of the worst drawdown with the height of the climb.
The equity curve of the kept candidate with its drawdowns. Compare the slope of the marked out-of-sample stretch with the slope before it, and the depth of the worst drawdown with the height of the climb. ES · daily · 2019-01-02 to 2025-06-30 · CPU engine backtest

Three readings matter more than the headline profit. The max drawdown of $35,290 against a net profit of $119,422.5 over six and a half years is the number a trader would have had to sit through on one contract. The profit factor of 1.391 says the winners outweighed the losers by a modest margin, which is the kind of margin that costs and selection bias eat first. The time in market of 71.46 percent confirms that the rule is a position rule on daily bars, not a day trade. The average trade of $402.1 is measured after the $20 commission and $25 slippage charged on every trade, which is the only way the ranking of the 1,024 candidates could reflect what was tradeable; because a pivot rule fires only when a level is reached, AlphaPrime's backtesting charged those costs inside each of the 1,024 evaluations rather than subtracting them from the finalists afterwards.

Five steps from pivot formula to exported strategy: choose the variant and session, write the entry and exit, set costs per trade, test out of sample, and export to the platform. A rule that fails the out-of-sample test goes back to step 2 as a new candidate, not a re-tune. Five steps, no re-tuning A failed test means a new candidate 1 Variant and session 2 Entry and exit 3 Costs per trade 4 Test out of sample 5 Export to platform If it fails: new candidate
The summary of the kept candidate in the report. Read the average trade next to the commission and slippage charged per trade, and the trade count next to the length of the period. ES · daily · 2019-01-02 to 2025-06-30 · CPU engine backtest

What the run does not show is also useful. It did not test a bounce rule on intraday bars, it did not test the Fibonacci or Camarilla variants, and it did not test a day-session pivot. Each of those is a different experiment with its own count of holds and breaks, and each would need its own search and its own out-of-sample judgment.

How to test a pivot rule on your own futures data

Write every definition and every pass threshold before the first run, charge costs from the start, and judge the result on data the rule never saw. A pivot test is easy to bend after the fact: a different close, a different variant or a looser touch distance each gives a new ladder and a new result, and the one you keep will be the one that happened to fit. Fixing the order of decisions removes that temptation.

  1. Pick the session and the closing price. Full session or day session, last trade or settlement, and how to handle the roll day. Write it down.
  2. Pick the variant and the levels you trade. Classic, Fibonacci or Camarilla, and which rungs (R1 and S1 only, or out to R2 and S2).
  3. Write the entry and exit rule, bounce or breakout, with its thresholds. Define a touch in ticks, define a break as a close beyond the level, and set the minimum average trade after costs and the minimum trade count you will accept.
  4. Set commission and slippage before the first run. Use your broker's round-trip cost and at least one tick of slippage per side for a liquid index future, more for stop entries at the level.
  5. Run in-sample and count trades per level. A level that produced a dozen trades in five years cannot be judged; drop it or merge it with its neighbour before looking at profit.
  6. Judge unchanged on out-of-sample data against the thresholds from step 3. No re-tuning after the look. If it fails, the next idea is a new candidate.
Six steps in order: pick the session and closing price, pick the variant and levels, write the rule with its pass bar, set costs up front, run in-sample and count trades per level, then judge unchanged on out-of-sample data Decide first, look last Six steps, in this order 1 2 3 4 5 6 Session Range and close Variant Which levels Rule Plus pass bar Costs Set up front In-sample Count per level Out-of-sample Judge unchanged
Definitions first, costs from the start, unseen data last. Each box must be closed before the next opens.

Step 6 is where most pivot tests go wrong, because the ladder offers so many knobs. The guide on how to split in-sample and out-of-sample data and how much is enough covers how many out-of-sample trades a per-trade figure needs, and the same discipline applies to any chart level, including the exact rules for a double bottom pattern, where the neckline plays the role R1 plays here. If you run the search in AlphaPrime, set the thresholds from step 3 in strategy validation before the search starts, so the out-of-sample and walk-forward decisions are made by rules you wrote in advance.

A concrete first test: take two years of ES daily bars built from the full session, compute Classic P, R1 and S1 for each day, and count how many days price touched R1 and closed below it versus closed above it. That one count tells you whether a bounce rule or a breakout rule is the one worth writing in step 3.

Frequently asked questions

How do you calculate daily pivot points?

Take yesterday's high, low and close and average them: P = (H + L + C) ÷ 3. R1 is twice P minus yesterday's low, S1 is twice P minus yesterday's high, R2 is P plus yesterday's range and S2 is P minus the range. The levels are fixed for the whole of today because they use only yesterday's completed bar.

Are pivot points accurate?

Accuracy depends on the rule, not the level. A bounce rule counts a hold at R1 as a success; a breakout rule counts the same hold as a failure. Count holds and breaks per level on your own data, add how far price travelled after each, and subtract commission and slippage. Only then do you have a number for one rule on one market.

Should pivot points use the day session or the full session?

Either works as long as the backtest and the live chart use the same one. The full session includes overnight highs and lows and usually gives a wider ladder; the day session gives a narrower ladder and a different close. Decide before testing, write down which closing price you use, and check how your continuous contract handles the roll day.

Which pivot variant is best for futures?

No variant is best in general. Classic and Floor place wide levels suited to targets and breakouts, Fibonacci places levels inside the expected range, and Camarilla places tight levels near the close for fading small moves. Pick by the size of move you expect per trade, because tight levels mean more trades and a bigger share lost to costs, then test that one choice.

Do pivot points work for day trading ES futures?

Pivot levels are often used intraday because they are known before the open and shared by many traders. Whether an intraday rule pays depends on the average move per trade after costs, which is small for inner levels. Test the rule with realistic commission and slippage on intraday bars, and compare the result with the same rule at double the slippage before trusting it.

References

  • Bailey, D. H., Borwein, J., López de Prado, M., & Zhu, Q. J. (2014). Pseudo-Mathematics and Financial Charlatanism: The Effects of Backtest Overfitting on Out-of-Sample Performance. Notices of the American Mathematical Society, 61(5), 458–471.
  • Brock, W., Lakonishok, J., & LeBaron, B. (1992). Simple Technical Trading Rules and the Stochastic Properties of Stock Returns. The Journal of Finance, 47(5), 1731–1764.
  • Park, C.-H., & Irwin, S. H. (2007). What Do We Know About the Profitability of Technical Analysis? Journal of Economic Surveys, 21(4), 786–826.