If the small timeframe looks right and the large one disagrees, pass

Closed bars, three frames, and a skip when they fight.

If the small timeframe looks right and the large one disagrees, I pass. On closed bars I read M15, H1, and H4: EMA 21 versus 55, and RSI(14) above 55 or below 45. At least two frames must agree. A clean M15 against a fighting H4 is a skip, not a smaller ticket.

Where the rule comes from

RSI and EMA only know the window you opened. An M15 bounce inside an H4 sell is not the same event as the same bounce when H1 and H4 already agree.

The extra layer I actually use is small:

1. Three frames, same symbol: M15, H1, H4. 2. On each closed bar, fast EMA 21 against slow EMA 55. Fast above slow is bull structure. Fast below slow is bear. 3. RSI(14) on that same closed bar. Above 55 is a long vote. Below 45 is a short vote. Between 45 and 55, that frame does not vote. 4. Structure and RSI both have to be present. One without the other is not a vote. 5. I need at least two of the three frames in the same direction. 6. Higher-frame bars map by time. I do not treat an H4 candle as if it were the same bar as the M15 I just closed.

I log the skips. One line is enough: M15 long idea — H4 bear — passed.

The first question after a close — did the bar lengthen or shorten — is a sibling note, Close-bar momentum. This page is whether the frames agree.

How I judge a frame

Three checks, in this order.

1. Closed bars only

An H1 wick that reverses into the hour close is not a vote. If H4 is open, I use the last closed H4.

2. A vote is EMA plus RSI

Fast 21 against slow 55 names structure. RSI(14) has to clear 55 or 45. An RSI of 53.37 on a tidy M15 is not a long vote. Do not round the middle into a side.

3. Disagreement is a pass, not a smaller size

If M15 looks right and H4 fights, I do not "take half." I pass. Two-of-three is the floor. Three-of-three is not required.

What the figure is, and is not

The red arrows are alignment flags. They are not orders. I did not take them as entries. Alerts were off.

On the locked tape at 2026-10-23 04:00 (server time) I will not write 3-of-3 aligned. Measured on that timestamp:

M15: EMA21 1.12199 sits above EMA55 1.12191, but RSI 53.37 sits between 45 and 55, so that frame has no vote. H4: EMA21 1.12225 sits under EMA55 1.12231, RSI(14) at 45.00 — soft bearish, on the 45 line rather than below it. I did not log an H1 print for this note, so I will not invent one.

Nearby red arrows on 22–23 October are the indicator's min-2-of-3 marks. That is not the same claim as three frames aligned on the 04:00 bar.

The failure I care about: waiting for alignment

Passing when frames disagree is the rule. Waiting until they all agree is how the rule goes stale.

By the time M15, H1, and H4 print the same vote, the move that made the idea cheap is often already on the chart. The stop sits farther away. The same invalidation now costs more R.

I would rather pass the M15 that fights H4 than wait for H4 to join and then chase. A skip is cheap. A late aligned entry is a worse ticket.

I do not treat "confluence" as a win-rate. I have no sample in this note that would let me write a percentage. Two frames in the same direction is a permission filter, not a forecast.

When the observation does not count

• RSI in the middle. 45 to 55 is not a vote. 53.37 is not a long.
• Open bars. A forming H4 is not today's H4 vote.
• Shopping for a frame. Opening M5 until it agrees is not this rule. The three frames are fixed before the sitting.
• Arrows as instructions. If a mark appears and I have not done the closed-bar checklist, I have not followed the rule.
• Taking the M15 "just this once." That is a rule break. Write it down. The public template is Rule-break review.

Narrow ranges flip votes. I pass more those days. Some skips would have worked. That is what a filter costs.

What the training desk can actually run

The original stack is M15, H1, and H4. On BarionQuant Desktop that needs Standard — the seat that adds M5, M15, and H4 next to H1 and daily. I will not write "open the free seat and follow M15."

Free can run a two-timeframe version: H1 against daily, same EMA 21/55 and RSI(14) 55/45, both closed. That is a simplification, not the original rule. Two frames are not three. Daily is not H4. If you only have Free, treat the H1/daily read as practice in passing. Do not call it the M15/H1/H4 stack.

There is no session file on this page. I am not going to tell you to download this tape and walk it.

Windows seats start at download the Windows build. What Standard adds, including M15 and H4, is on pricing. Public USD card frozen 2026-09-14: Standard $15/mo or $150/yr. Live pricing checked 2026-10-27 still lists Free as H1 and D1.

Who should skip this page

• Anyone who wants the arrows to be buy and sell orders.
• Anyone who needs a win-rate, a "high-probability" confluence, or a call for today.
• Anyone who only has Free and expected to run M15 and H4 on it.
• Anyone looking for a broker, a live signal, or a script that sends orders. This desk does not.

FAQ

Do I wait until all three frames agree? No. Two is the floor. Waiting for the third is how the entry gets late and the R gets worse.

Can I do this on Free? Only the H1-versus-daily simplification. Say so in the journal. Do not call it the M15/H1/H4 rule.

Are the red arrows entries? No. They are min-2-of-3 alignment flags. Alerts were off. I still do the closed-bar checklist.

Is the MetaTrader file listed on barionquant.com? No. This page is an essay, not a catalog row for that Market file. The Market file is a different surface. The training desk does not ship it.

Next step

If you took an M15 that fought H4, use the rule-break review. For the first question after a close, read close-bar momentum. For a seat that can hold M15 and H4, download the Windows build and read pricing before you assume Free can.

A free MetaTrader 5 file that can paint min-2-of-3 marks is Market product 181820 (https://www.mql5.com/en/market/product/181820) (checked 2026-10-27: FREE, version 1.36, published 17 June 2026). Alignment flags, not orders. The listing's buy/sell arrow copy is not this essay. The training desk does not ship that file.

For study and simulated practice only. This is not investment advice.

Educational content. BarionQuant is not a broker and does not provide investment advice or trade signals. Past practice results do not predict future outcomes.