history3

Plays out $1 per match on the recorded prices (the tape) IF the betting rules had been the settings the owner picks. Nobody on the AI side analyses.

favcohort data for the next build: favcohort.json (two rows per match, each side labelled tp_and_market_fav / tp_only_fav / market_only_fav / nobodys_fav) · counts and win rates per cohort (report) · one rising span per match, as arrows · when the lows and highs happen

Reading published files.

Matches with recorded prices

height ↕ 60vh rows/text 1.00× x zoom 100% y zoom 100%

Controls

SPORT

COHORT

BUY LIMIT buying stops at the game's end

SELL LIMIT sells cancelled and sold at market at the game's end

HIDE TOP / BOTTOM hide the top X matches 0 hide the bottom X matches 0 RANK BY

BUY PRICE

SELL PRICE

AVERAGE GAME TIME PER SPORT

Scoreboard

No cell selected. Click a cell to highlight it. Enter or the link opens detail.php. Arrow keys move the highlight.

height ↕ 24vh rows/text 1.00× x zoom 100% y zoom 100%

open the highlighted cell on detail.php

Total net of $1 per match IF we had bought at x and sold at y — values in $

COLOUR BY

height ↕ 55vh rows/text 1.00× x zoom 100% y zoom 100%

requirements check

Questions for additional charts (brainstorming; nothing here is built until the owner names it)

Time rules or price rules — which finds the higher sweet spot?One chart with two x axes (buy minutes after the start, and buy price in cents) and two net lines, circles sized by net per fillable match, peaks labelled.
Percent of the opening price or fixed cents — which basis gives a shape that holds across sports?Two panels of net, one per basis, one line per sport plus pooled, circles sized by fillable count.
At each buy price, which sell rule nets most: a fixed sell price, a multiple of the buy price, or holding to the end?x = buy price; three net lines (best fixed-cent sell, best multiple-of-buy sell, hold); the winning rule's marker is solid.
When in the game does the low happen, and does buying the low up to that minute pay?x = minutes after the start; left y = share of matches whose low had already happened; right y = net if we had bought that low.
Does the market overreact? After a sudden dump, where is the price p minutes later?x = minutes after the dump; y = median price change; one line per dump size, with a net panel if we had bought after each dump.
How long does a fillable buy take to reach each sell price, and how many never do?x = minutes since the buy; y = cumulative count that had reached each sell price, never decreasing, on a fixed population.
Volume against profit — as the buy price rises, more matches fill and each fill earns less; where does the total peak?x = buy price; left y = total net; right y = fillable matches; circles sized by net per fillable match.
Which favourite cohort carries the edge, and at what price?x = buy price; one net line per favcohort plus pooled; circles sized by the win rate.
Which sport, and does the shape differ by sport?The same as the cohort chart, per sport, with the pooled line as the reference in every panel.
How concentrated is the profit — how many matches carry it?x = the top k matches by net; y = the share of total positive net they hold; a tick box on the table row hides the match.
Is the best window a number of minutes, or a share of how long that sport usually runs?Two lines on two x axes, one y (net): bottom x = minutes, top x = percent of the sport's usual length.
Does re-entering after a sale add net, or just stake?x = allowed entries per match; y = net with dollars staked as a second bar; groups by spacing.
Buying at the open or waiting for a dip — at what discount from the open does waiting beat buying at once?x = discount from the opening price; two net lines (buy at the open, buy only if the price falls to the discount); the crossing labelled.
Does the time of day or the weekday matter?Bars by start hour (KST) and by weekday, y = net, circles sized by fillable count; pooled and per cohort.
Which combinations of the two clocks produce the most profit?proposed by astra, not built. A tile grid of buying-stop time against forced-sale time, coloured by net of $1 per match.
Which matches leave the most possible profit uncaptured?proposed by astra, not built. Circles of each match's max-winnable ceiling against that cell's net, sized by the gap, with an equality line.
Where do our clocks fall against each match's best buying and selling moments?proposed by astra, not built. Circles of best-buy time against best later sell time, sized by max winnable, with both clock boundaries drawn.
How much missed profit comes from the clocks, and how much remains after those limits?proposed by astra, not built. Waterfall bars from the unrestricted ceiling to the best possible under both clocks to the chosen grid net.
Does the same price mean the same chance of winning as time runs out?proposed by astra, not built. Tiles of price band against elapsed minutes, coloured by eventual win share minus price, one panel per favourite label.
How much money must be available to fund overlapping $1 entries?proposed by astra, not built. Stacked area of dollars tied up in open positions over calendar time, with a running grid-net line.
How much profit survives a delay in acting on a recorded price?proposed by astra, not built. Lines of grid net against assumed delay in seconds, separately for delayed buys, delayed sells, and both.
Do prices and clocks chosen from earlier matches still work on later matches?proposed by astra, not built. Paired bars of settings selected on earlier matches against that later period's hindsight best. Listed only so it was considered; a time-period comparison the owner ruled out.
How deep and long are the losing stretches behind a profitable grid total?proposed by astra, not built. Running grid net over calendar time of tape-based sales, with the previous-peak line and shaded losses below it.
Does the apparent opportunity depend on poorly recorded stretches around our deadlines?proposed by astra, not built. Circles of the gap between recorded prices spanning each clock against max-winnable minus grid net, one panel per clock.

history3

▾ Matches with recorded prices

▾ Controls

AVERAGE GAME TIME PER SPORT

▾ Scoreboard

▾ Total net of $1 per match IF we had bought at x and sold at y — values in $

▸ Questions for additional charts (brainstorming; nothing here is built until the owner names it)

Matches with recorded prices

Controls

Scoreboard

Total net of $1 per match IF we had bought at x and sold at y — values in $

Questions for additional charts (brainstorming; nothing here is built until the owner names it)