When the lowest low and the highest high happen: price on x, minutes after the scheduled start on y, inside a window the owner sets. Nobody on the AI side analyses.
history3.php · detailed.php · detail.php · when the lows and highs happen
Reading published files.
This page has six numbered sections. 1 chooses the matches. 2 shows where and when their two extremes happened. 3, 4 and 5 replay buy-and-sell rules over the recorded tapes, on a population declared in 4 and never on the selection made in 1. 6 is the one record per match behind every view in 2.
These controls pick the matches. Section 2 and section 6 describe exactly this selection. The replays in sections 3, 4 and 5 do NOT: they run on the calculator's own declared population, set in section 4.
Click any mark, any cell or any table row to pin that match's whole record: the popup carries close, copy, and a link to the recorded tape on odds.php. Hovering never opens a popup; hovering a (?) or a header shows its words only.
Every view below draws the same matches chosen in section 1, read four different ways. Green means the match was won, red means it was lost, and blue marks whatever you have selected.
How do low and high prices and times differ between winners and losers, and between the two event orders? Drawn from the recorded tapes of the matches selected in section 1: one pair of marks, joined by one line, per match.
x axis: price in cents, 1c at the left to $1 at the right · y axis: minutes after the scheduled start, the window's lower handle at the bottom and its upper handle at the top
At the same prices, how do the timing distributions of lows and highs differ between winners and losers? The same matches and the same axes as chart A in 2.1, split into four panels with the joining lines removed so marks stacked on top of each other stay visible.
x axis: price in cents, 1c at the left to $1 at the right, the same scale as chart A · y axis: minutes after the scheduled start, the same range as chart A
Which price-time bands hold more lows than highs? How does the share of winners differ across price-time bands? The same matches as chart A in 2.1, counted into two tables of fixed price and minute bands: one table for the lows, one for the highs.
x axis: price bands of 5c from 1c to $1, with the sub-1c band owned separately and the $1 band owned by the settlement · y axis: minute bands after the scheduled start
Within the same low-price band, how do price ranges and time gaps differ between low-first and high-first matches? The same matches as chart A in 2.1, one mark each, sorted into panels by how cheap the lowest low was.
x axis: the time gap in minutes, the high's minute minus the low's minute, negative when the high came first · y axis: the price range in cents, never negative
Both grids here are what-if replays of a buy-and-sell rule over the recorded tapes, at $1 per match, before fees. Grids 3.1 and 3.2 take the FIRST rule of the calculator in section 4, vary two of its settings, and hold every other setting fixed. The matches replayed are the population declared inside that calculator — the filters in section 1 never enter it, and neither do they enter section 4 or section 5.
With the same clocks and population, how does total replay net differ across buy and sell thresholds? Every cell replays the calculator's first rule (section 4) over the calculator's declared population, changing only the buy threshold and the sell target.
x axis: the buy threshold in cents · y axis: the sell target in cents · every cell prints the total replay net in dollars and the bought / evaluable counts
With the same price rule and population, how does total replay net differ across entry and exit deadlines? Every cell replays the calculator's first rule (section 4) over the calculator's declared population, changing only the last minute it could have bought and the last minute it could have sold.
x axis: the entry deadline in minutes after the scheduled start · y axis: the exit deadline in minutes after the scheduled start · every cell prints the total replay net in dollars and the bought / evaluable counts
This is where the population and the three rules are declared. Grids 3.1 and 3.2 replay the first rule of this calculator on this population. Its own six sub-sections are numbered 4.1 to 4.6.
How to read the headline table: one column per rule, all three scored on the same population, at $1 per match, before fees. "reconciled" means the matches it bought add up exactly: bought = sold at the target + sold at the $1 settlement + sold at market or at the deadline + lost at $0. Every rate prints its fraction, as "51/60". Each rule carries one tag — its slot number, its name and its colour in words, beside a swatch of that colour — and the same tag stands over its column in the headline table, over its group of columns in the per-match table, and beside its line and its mark shape in the two charts, so no colour has to be guessed. The six sub-sections below are numbered 4.1 to 4.6.
The rules and the population are frozen exactly as the calculator in section 4 applied them; only the scheduled-start date range changes, so the three totals are the same ruleset read over three stretches of the calendar.
Would the same three rules have netted the same thing in different stretches of the calendar? The rules and the population are frozen exactly as the calculator in section 4 applied them; only the scheduled-start date range changes.
One row per match in the selection from section 1, with the two extremes the charts drew, the window they were computed for, and the tape conditions behind them. Hover any column heading for its full name; click a row to pin its whole record.