Weekday-Hour Fade Grid
A fade at 10am on a Tuesday and the same fade at 2pm on a Friday are not the same trade, and your instrument has already logged the difference. Paste dated intraday bars and this analyzer crosses the day of the week with the hour of day, tags every test of the day's running high or low, and grades the next 3 bars as a fade back, a break through, or chop. The result is a grid of day-and-hour boxes: which ones pay for fading the edge and which ones punish it. The test and outcome rules are the same convention as the Fade Hour Analyzer, extended across both axes, stated below so you can check the math.
Rows as date time,open,high,low,close. The date is required so the day of week is known: use 2026-04-06 or 04/06/2026 (month/day/year), with a 24-hour or AM/PM clock. 8 or more weeks of 15, 30, or 60-minute bars gives the grid enough tests per box. Volume is ignored.
No grid yet
Paste the bars and hit Build. The day and hour lists fill in from your data, and the verdict follows the box you pick.
How it grids. Each day is grouped by its pasted date, and the day of week comes from that date, not from the order bars arrive in. The running high and low are the day's high and low so far. A test is a bar, at least 4 bars into the day, whose high comes within 10% of the running range of the running high (or whose low comes within 10% of the running low), with the running range at least 40% of the median full-day range in the sample, so the first tiny wiggle of the morning doesn't count. Outcome, over the next 3 bars: a close beyond the edge by more than 10% of the running range is a break; a close at least 30% of the running range back toward the midpoint is a fade; anything else is chop. Each test lands in the box for its day of week and its hour. A test can't repeat on consecutive bars at the same edge; the edge has to be left (more than 10% away) and revisited. Verdict per box: fade rate 60% or higher with 8 or more tests is a fade window; break rate 40% or higher is a breakout hour, don't fade; fewer than 8 tests is thin data, read the neighboring boxes instead. Bars with no parseable date are dropped from the grid since the day of week can't be known. Keyboard: Enter builds the grid, 1 loads the example, click any box to select it.
Educational tool only, not financial advice. A box needs several weeks of tests before its rate means anything, and every rate drifts once the regime changes.
Built by Jason Parker, founder of Trading Ranges.
Built by Jason Parker, founder of Trading Ranges.