A multi-timeframe protocol assigns different questions to related views of the same data. Define a context filter, a setup condition and a trigger, then check when each becomes available. Passing the signal is separate from obtaining an acceptable executable price and calculating volume. The examples below are our own hypothetical controls, not personal trading results or a reconstruction of an author’s original system.

Define what each timeframe contributes

A longer candle aggregates shorter intervals when source, boundaries and price type match. It keeps selected endpoints and extrema rather than the full path. Different candles can describe opposite close-to-open directions without contradiction; they are views of related inputs, not independent forecasts.

Our H4/H1/M15 teaching protocol is defined below. It does not establish a win-rate improvement, a universal hierarchy or the original rules of Elder’s Triple Screen. The setup condition can require a pullback against the context direction, so the three roles need not show identical price directions.

Three timeframes, three distinct jobs

At the heart of multi-timeframe analysis is a clean division of labour between three charts. Each one answers a different question and never tries to do the job assigned to the other two.

The three timeframes and their roles
Higher timeframeEstablishes trend direction, acting as a directional filter rather than an entry tool
Intermediate timeframeIdentifies the specific setup, such as a support, a resistance, a pattern, or a confluence of factors
Entry timeframeSupplies a defined trigger; a separate execution check follows
Ratio between timeframesA chosen parameter; not a universal four-to-six requirement
All required conditions passThe signal is complete; this is not three identical directions or a guarantee of execution

For a synthetic Bid EUR/USD day, use UTC half-open intervals with H4 beginning at 00:00, 04:00 and 08:00. After data receipt, H4 closes at 04:00 and 08:00 are 1.0900 and 1.0920: the context filter permits a buy idea. H1 closes at 08:00 and 09:00 are 1.0920 and 1.0904: the chosen setup is a fall that stays above the predefined 1.0880 boundary. M15 ending 09:00 has H = 1.0908; the next M15 ends 09:15 with C = 1.0909, above that high. All three conditions pass after receipt of the completed bars, although H1 falls.

Four combinations tailored to trading styles

The sets in the table are possible role assignments, not mandatory or demonstrated most-common combinations. Choose data and observation times for a defined rule. Longer holding periods can use broader context; see the distinction between styles in position trader.

Standard timeframe sets by trading style
Position traderWeekly, daily and H4 — positions held for weeks or months
Swing traderDaily, H4 and H1 — positions held from a few days to a few weeks
Day traderH4, H1 and M15 — positions closed before the daily session ends
ScalperH1, M15 and M5 — positions held from minutes to about an hour
Check boundaries, not just labels24 hours / 5 minutes = 288 for a full assumed day, but that ratio alone does not invalidate a defined method

After 09:15 in the synthetic day, H4 08:00–12:00 remains incomplete. Use the H4 ending 08:00 for this completed-bar rule, not its later successor. A chart-display timezone change does not by itself change script calculations or session construction. Preserve source, time convention and provider availability alongside the candle label.

Rules of alignment and when to stand aside

The protocol requires all three different conditions. Evaluate once after the M15 ending 09:15 is supplied, then apply current-price and execution checks. Missing data or a failed condition ends that assessment without an order; a valid signal is not indefinite permission to buy later.

  1. Complete signal. The H4 comparison, H1 pullback and M15 event all meet their definitions. Proceed to the separate current-price check; identical directions and a highest-probability claim are not required.
  2. Missing or failed condition. No order under this protocol. Reduced volume cannot supply a missing observation or make a false condition true.
  3. Execution rejected. Even a completed signal can fail the later price, validity or provider check. Record that reason separately from a failed signal condition.

An additional synthetic timing variant receives H4 at 08:00:02, H1 at 09:15:03 and M15 at 09:15:00.500. The complete set becomes available at their latest receipt time, 09:15:03. An assessment at 09:15:01 lacks H1 even though that candle ended at 09:00. If the trigger expires two seconds after the M15 end, no moment meets both completeness and validity. These delays are chosen controls, not measured provider latency.

A valid signal may fail the actual-price check

After collecting three readings, ask a different question: is current price within this version’s pre-recorded boundary?

Scroll sideways to see the full diagram

Price after a signal can reject the decisionM15 closes at Bid 1.0909. Later Ask 1.0914 exceeds the assumed 1.0912 threshold, so the model sends no order.THREE SIGNALS DO NOT REPLACE PRICE1.09091.09121.0914M15: BidLater: AskC = 1.0909Ask = 1.0914Decision boundary: Ask no higher than 1.0912.1.0914 exceeds that boundary.This version: no order sent.Signal and execution conditions are separate gates.

Hypothetical protocol: all H4/H1/M15 signal conditions pass after data arrive. M15 closing Bid = 1.0909. A separate pre-order check requires current Ask≤1.0912; later observed Ask = 1.0914. The threshold was set beforehand and is not a recommendation. No simultaneous Bid is shown, so the difference from earlier Close is not a spread measurement. MetaTrader 5 — Basic Principles distinguishes order instructions from executed deals and uses Ask for buying in the described trading model.

The threshold is a check before sending an order. It neither guarantees a later fill price nor represents a submitted Buy Limit.

Record “rejected — current Ask outside threshold”. This extra price condition preserves the earlier signal calculation while blocking this version’s execution. Changing the boundary later creates a different protocol. Passing the gate still does not replace confirmation of actual fill price and quantity.

EUR/USD step by step (illustrative example)

The next table preserves a separate hypothetical swing trade. Its indicator values and candle labels are assumed, not calculated from a complete supplied history. A price structure condition still needs pivot-selection and availability rules when implemented.

Top-down analysis — EUR/USD on the daily, H4, H1 sequence (illustration)
Step 1 — daily chart200-period EMA rising, swing high at 1.0950, higher low at 1.0810, indicating a clean uptrend
Step 2 — H4 chartPullback to 1.0880, where the 50 EMA, the 50 percent Fibonacci retracement of the 1.0810–1.0950 leg, and a pin bar all coincide
Step 3 — H1 chartBullish engulfing closing at 1.0890 confirms the H4 pin bar, providing the entry signal
EntryAssumed actual buy fill at Ask 1.0895 after the signal close of 1.0890; the difference is not a measured spread
Stop lossRequested Bid 1.0865, 30 pips below the assumed Ask entry, not a stated 30-pip distance below an unknown candle low
Take profitAssumed exit Bid 1.1000: 105 pips gross reward versus 30 pips planned risk; reward/risk = 3.5
Analytical outcomeThe assumed conditions meet this separate example’s signal rule; no future fill or profit is established

Now compare two independent volume models after the first protocol signal. Assume buy Ask 1.0911, a USD account, 100,000 EUR per lot and pip = 0.0001 USD per EUR, giving USD 10 per pip per lot. Choose a USD 50 budget, 0.01-lot increment and USD 7 round-trip commission per lot, omitting financing and other fees. Ordinary stop levels at Bid 1.0881 and 1.0905 are 30 and 6 pips from the entry. Do not add spread again to these executable-side differences. These figures are hypothetical, not a provider tariff.

Practical takeaway: Context, signal and executable price are separate conditions; passing one does not replace the others.

Mistakes that keep coming back

The planned scenario cost per lot is USD 307 or USD 67. Divide the USD 50 budget by each and round down to the 0.01 increment: 0.16 or 0.74 lot. Loss including commission at the requested stop is USD 49.12 or USD 49.58. Five pips of worse closing fills gives USD 57.12 or USD 86.58. The narrower stop admits 4.625 times the volume and adds USD 37 rather than USD 8 from the same five-pip slippage. Five pips is not a maximum; this compares sensitivity, not which stop is profitable.

  • Changing conditions after seeing the outcome. Freeze the version, data and observation times used for the assessment. Do not change the 1.0880 boundary afterward to rescue a rejected idea.
  • Using votes instead of the defined conditions. The example explicitly needs a falling H1 pullback inside its allowed boundary. Different roles can all pass without three upward candles.
  • Using later final values early. H4 08:00–12:00 is not complete at 09:15. Do not backfill an earlier decision with its eventual close.
  • Ignoring incomplete data. Record which required input is missing. A previous completed H1 can be substituted only if the rule explicitly permits it, and its setup condition may differ.
  • Inferring execution quality from chart agreement. Measure actual spread, fills and costs for the product and provider. The shared diagram’s later Ask check can reject an otherwise valid signal. A label for the current market regime does not supply those missing execution data.

What to do tomorrow

Practice reconstructing an assessment from its recorded inputs. The tasks below test chronology, signal validity and monetary sizing; a fixed week, three weeks or twenty trades is not a universal evidence threshold for a profitable method.

  1. Write the context, setup and trigger definitions, including source, timeframe boundaries and when each value becomes available. Keep the chosen version with the assessment.
  2. For the synthetic first model, verify 1.0920 > 1.0900, 1.0880 < 1.0904 < 1.0920 and 1.0909 > 1.0908. State which closed H4, H1 and M15 bars supply these conditions.
  3. Record a rejection separately for missing data, failed signal, expired trigger or later Ask outside the price boundary. Do not convert a rejected case into a smaller trade.
  4. Reduce the narrow-stop model’s budget to USD 30: 30/67 gives approximately 0.44776 lot, rounded down to 0.44. The planned loss is USD 29.48, and five pips of worse fill raises it to USD 51.48. A tighter price stop does not fix the monetary loss at the budget.
  5. Keep decision time, rule version, used candle identifiers and values, expiry, planned and actual prices, volume, costs, rejection reason and final outcome. Compare the complete protocol with a predefined alternative on a separate evaluation sample, including rejected signals.

This protocol separates data, conditions, execution and risk; it is not presented as Elder’s original specification. Related methods can be explored in trend-following systems, and the technical analysis section at ForexMechanics.