Choose the time drawing and its projection origin
Open the Fibonacci drawing tools on a standard time-based chart. Use Fib Time Zone for a two-anchor sequence, or Trend-Based Fib Time to measure A to B and place the projected intervals from C. TradingView documents bar-number coordinates for both tools and three coordinates for the trend-based version. Its Trend-Based Fib Extension is a different drawing that projects price levels.
- Place A and B on identifiable completed bars. For the trend-based workflow, place C on the bar from which you want to count forward. Keep A before B for the examples below.
- Open the drawing settings and record the anchor coordinates and enabled levels. Do not assume that two saved drawing templates use the same ratios or origin.
- Use 0 and 1 as orientation checks: for the convention below, 0 is C and 1 is one measured A-to-B interval after C. Check the displayed lines against those coordinates before using fractional levels.
- If using Fib Time Zone, record that its sequence starts from A. A line at A + 2 × (B − A) is a different location from C + 2 × (B − A), unless A and C coincide.
Convert 20 anchor intervals into explicit windows
This is a hypothetical bar-count exercise, not a market forecast or a report of TradingView results. Let A be bar 100, B bar 120, and C bar 130. The measured distance N = B − A is 20 bar intervals, although 21 candles include both endpoints. For each chosen ratio r, use target = C + round(r × N). Round to the nearest whole bar, with halves rounded up, then inspect target minus 2 through target plus 2, inclusive. This is an explicit review convention; it does not claim that the drawing uses this rounding internally.
- Freeze the ratios and the five-bar window before checking outcomes. The window width is an assumption to test, not a Fibonacci-derived confidence interval.
- Reject A = B because the measured interval is zero. Reversed anchors require a separate, documented direction rule; do not silently take the absolute value after seeing the result.
- With short intervals, different ratios can round to the same bar and windows can overlap. Merge overlapping windows for event counting so one reversal is not reported as several successes.
- Save anchor timestamps as well as bar positions. Bar indices depend on the loaded dataset and are not permanent identifiers across different histories.
| Ratio r | Distance from C | Target bar | Review window |
|---|
| 0.618 | round(12.36) = 12 | 142 | 140 to 144 |
| 1.000 | round(20) = 20 | 150 | 148 to 152 |
| 1.618 | round(32.36) = 32 | 162 | 160 to 164 |
| 2.618 | round(52.36) = 52 | 182 | 180 to 184 |
Trading bars and calendar days give different dates
A bar-count projection advances through available chart observations. A calendar-day projection advances through dates, including dates with no session. Decide which hypothesis you are testing before converting a count into a date. The examples below use an invented Monday-to-Friday calendar with no holidays; real exchange calendars can produce different answers.
- A 24/7 market and a weekday-only market do not share a bar-to-date mapping. Even a 24/7 feed can have missing observations.
- Holidays, shortened sessions, halts, and absent feed data can change the mapping. Record whether a gap is an expected closure or a data problem; do not fabricate bars to fill it.
- For a timestamp-based experiment, specify the timezone and whether a day means a local calendar date or exactly 24 elapsed hours. These can differ around daylight saving changes.
- If a target falls where no bar exists, keep it labeled as a non-trading time or apply a next-available-bar rule chosen in advance. Record the displacement; moving it after seeing price changes the test.
| Assumption | Calculation | Result |
|---|
| C is Friday, October 2, 2026; daily bars; r = 1; N = 20 | Advance 20 trading bars after C, excluding C | Friday, October 30, 2026 |
| Same C; interval interpreted as 20 calendar days | October 2 plus 20 calendar days | Thursday, October 22, 2026 |
| Five-minute bars; last session bar opens 15:55; next session starts 09:30 | Advance two available bars after 15:55 | 09:35 in the next session, assuming both bars exist |
| Same 15:55 starting time; continuous elapsed-time interpretation | Add 2 × 5 minutes | 16:05 on the same date, when this assumed session has no bar |
Recalculate after changing interval or session
Switching from 15-minute to 60-minute bars changes the number of observations between the same two timestamps. In an uninterrupted four-hour span with aligned endpoints, there are 16 fifteen-minute intervals but only 4 hourly intervals. Keeping a hardcoded count of 16 on the hourly chart changes a one-times projection from four trading hours to sixteen. Session breaks and partial bars require checking the actual timestamps.
- Preserve the anchor timestamps and recount their distance when testing a new interval. An anchor inside a larger candle needs an explicit mapping rule or should be marked ambiguous.
- Enabling extended hours can insert observations between anchors and after C. Save a new experiment with its own session label; do not combine it with the regular-session results.
- A drawing can be hidden by its timeframe visibility settings. A missing line after switching charts is not proof that its projected time has changed.
- Changing the chart timezone changes displayed clock labels. It does not, by itself, change a Pine script time calculation. Keep chart display settings separate from explicit time rules in a script.
- Use standard time-based candles for this exercise. Range or other event-driven bars do not provide a fixed duration per bar for projecting calendar dates.
Record a window where the expected event never happens
For the hypothetical 1.618 window at bars 160 through 164, define a bearish-event test in advance: at least one completed close must be below 101, a previously recorded support price in arbitrary price units. Assume the five closes are 102.0, 102.4, 102.1, 102.8, and 103.0. None is below 101. The time window arrived, but the defined event did not. This is a failed event hypothesis, not a successful timing signal.
- A close of 101 exactly would also fail the strict below-101 rule. Decide whether equality counts before the test begins.
- A close below 101 on bar 165 is outside this window. Do not widen the window or move C retrospectively to claim success.
- If one of the five required bars is missing, label the observation incomplete rather than assuming that no event occurred.
- A swing identified only after later bars have closed was not available at the swing itself. Record the anchor confirmation time and evaluate only windows still in the future at that time.
- Keep every eligible projection, including misses. Compare against non-Fibonacci windows with the same widths and event definition on a separate evaluation period. One attractive chart does not establish predictive value.
Turn the hypothesis into a reviewable chart aid
Give Pineify the A, B, and C timestamps, symbol, interval, session, ratios, rounding rule, and event definition when requesting a chart aid. Ask for visible target windows and a separate condition for the completed-bar price event. Review the generated source and test a hit, a miss, an equality boundary, and a missing-data case before creating alerts in TradingView.
- Keep a window-start notification separate from a price-condition alert. A reminder that a window has opened is not an entry recommendation.
- In Trading Journal, save the hypothesis in a plan and the later outcome in a diary entry or review notes. Include the observation date, anchor confirmation time, and a reference to the saved chart screenshot.
- Log a no-event window without inventing a trade. Trade screenshots belong with actual trade records; a diary can retain the no-trade review and its screenshot reference.
- Pineify cannot restore unavailable market history or make a retrospectively selected anchor available earlier. Generated code needs review; a successful syntax check does not prove timing accuracy.
