The script description explains the mechanics of the csRSI MTF: the cyclically smoothed RSI, its adaptive bands, and the red and green background windows painted by a second csRSI running on a higher timeframe. This HOW-TO covers the part that turns those mechanics into a routine: when the windows tell you to pay attention, what to do inside them, and how to be notified so you can ignore the chart the rest of the time.

The workflow in one sentence. Wait for a colored window, then trade what forms inside it, a band sequence or a divergence, and do nothing between windows.

Reading the chart. Six months of the S&P 500 on the two-hour timeframe, filtered by the daily csRSI. Four episodes cover the whole method:

(1) The green window and its buys. From mid-March into early April the daily csRSI sat below its lower band, so the background turned green: buy territory on the two-hour chart. Note the plural in the heading. The two-hour line completed more than one buy sequence inside this window while the daily decline was still running; the early crosses bought bounces that faded, and the final cross at the turn of the month caught the 6,343 low and the rally of more than 1,200 points that followed. In real time nobody knows which trigger is the final one. Therefore treat entries inside a green window as a series, and size them so that an early trigger costs patience, not substance.

(2) The first sell and the warning it left behind. In mid-May the daily filter turned overbought and painted the first red window. Inside it the two-hour line crossed back down through its upper band around May 14, and the index dropped about 150 points in three sessions, from a close near 7,501 to 7,354, before the advance resumed. A valid sequence with a modest payoff, and something more important: when price pushed on to a new high above 7,600 in early June, the two-hour line answered that higher high with a clearly lower one. The first sequence paid small and left a divergence behind.

(3) The second sell: no band event, just the divergence. Inside the next red window the two-hour line never even reached its upper band again. It stalled at a visibly lower high while price stood at a higher one above 7,600, and that failure is the signal: momentum could not confirm the new high. The turn down from that lower high around June 3 was the divergence-backed trigger the first HOW-TO in this series called the strongest confirmation, and it needed no band cross at all. From the trigger the index returned nearly 300 points in five sessions, from a close at 7,554 down to 7,267. The window said look; the divergence said now.

(4) The two months since. From mid-June to today the daily filter has painted nothing, and the workflow has accordingly demanded nothing. Two months without a window is two months without a mandatory decision. The line currently sits near its lower band with the filter neutral; the next assignment arrives when the background changes color.

Why the windows earn their keep. Chart 2 zooms out to two years of the same setup. The green windows cluster at the major lows, including April 2025 and April 2026, and the red ones gather at the tops, with a few appearing mid-rally in the autumn of 2025 that paid pullbacks of the May variety rather than reversals. The filter does not predict; it concentrates your attention on the handful of weeks where the higher timeframe says the odds are stretched.

CHART 2:

Settings. This chart runs the recommended pairing for two-hour trading: indicator timeframe "Same as chart", MTF resolution one day. The script description lists the pairings for other trading rhythms, from twenty-minute charts up to weekly investors. The optional cycle length inputs sharpen both lines further; the chart above runs the standard settings, unoptimized.

The alert that replaces screen time. The windows exist so you do not have to watch the panel. Set one alert per symbol: choose the csRSI MTF value with the "Out of Channel" condition, and select the MTF high band and MTF low band as the upper and lower limits. The alert fires exactly when a red or green window opens, which is the moment the workflow above begins. Until then, the chart can stay closed.

This is a discretionary tool, and windows guarantee nothing: they mark where signals have historically carried better odds. Size positions so a failed sequence inside a window stays an annoyance, not a problem.

Indicator: