A pick fell to 40c. Does it come back? And could a loser have been sold at 82c first?
Two questions, two charts, one set of real in-play tapes. Chart 1 asks how often a pick
that got cheap still won. Chart 2 asks how many losing picks ever traded high enough to have been sold
at a profit first. Prices and counts only. No money.
The research numbers are not on this server yet.
This page draws agreed_curve.json, produced by build_agreed_curve.py from the
match tapes. Nothing is estimated here, so with the file missing the page shows nothing rather than a guess.
Start here: two facts that decide what is worth chartingshort version, then the charts
What Chart 1 says at the point you picked
Click any point on Chart 1 to move this reading. The wording changes with the reading.
Every band has its own sentence, including the unremarkable middle.?
Chart 1 — A pick got cheap. How often did it still win?
LinesZoomShow
drag to resize chart height
How to read Chart 1.
Pick a price on the bottom axis. Each point counts every match that ever touched it.?
The line shows how many of those the pick won. The slanted line is fair: a 40c pick is fair if it wins 40%.?
Below the line, that side cost more than it delivered. The other side, at about 100c minus the price, was the bargain.
The shaded band is the 95% range.?
Dot size is the number of matches behind the point.
Chart 1 in five numberscomputed from the same file the chart draws
What Chart 2 says at the point you picked
Click any point on Chart 2 to move this reading.?
Chart 2 — Could a losing pick have been sold high before it went to zero?
LinesMeasureZoomShow
drag to resize chart height
How to read Chart 2.
Pick a price on the bottom axis. Each point counts every loser that ever touched it.?
One line shows what share of that cohort's losers reached it. The other shows, of everything trading up there,
what share went on to lose. There is no winners' line here on purpose.?
Every winner reaches every price on its way to 100c, so a winners' line would be 100% everywhere and would say nothing.
Switch Measure to compare each loser against its own opening price instead of a flat number.?
The answer, in one tabledarker = more losers could have got out there
What Chart 3 says at the level you picked
Click any point on Chart 3 to move this reading.?
Chart 3 — At that level, held every time, would it have paid?
LinesOrderZoomShow
drag to resize chart height
How to read Chart 3.
The height is money, not a rate. It is the average result of staking $1, over and over, at that level.?
Above the zero line the level made money in this sample. Below it, it lost.
Standing buy leaves an order waiting at a price. Every match that ever traded there fills it.?Buy at open, sell high buys at what the pick opened at and leaves a sell waiting up at a price.
Both assume the level held long enough to fill.?
Prices here are polled midpoints, so a fast wick through a level may never have filled anything.
Are there enough winners at each price to pay for all the losers?green = enough · red = short
What Chart 4 says at the square you picked
Click any square on Chart 4 to move this reading. Click an empty part of the
chart to clear it.?
Chart 4 — Where is the entry–exit sweet spot?
BasketSells?Check
drag to resize chart height
How to read Chart 4.
Every square is one plan. Buy $1 when the price first touches the row price.?
Then leave a sell order at the column price. The colour is the average result of $1 for that plan.
Green made money, red lost it, and the middle of the key is break even.?
The grid runs all the way down to a 5c buy. A 5c buy is not a cheap version of a 45c buy:
almost everything that gets that low is on its way to zero, so each square says who filled it.?
There is no zoom here on purpose: all 112 squares are on screen at every width.
Narrow screens show colour only — tap a square for its numbers, or drag the handle above to make the squares taller.
Both assume the level held long enough to fill.?
How cheap did it get before it first printed the sell level?the dip you had to sit through first
The old 82c study, checked against the full universeconfirm or revise
Touched is not bottomed. This is the thing people get wrong.every number here is computed
Table 1 — Chart 1 as numbersclick a header to sort · click a row to pin it on Chart 1
Table 2 — Chart 2 as numbersclick a header to sort · click a row to pin it on Chart 2
What this is, and what it is notread before using any of it
Research page for the pmtrade Polymarket project. Prices are polled midpoints from the bot's own match
tapes. Outcomes are as resolved by the market. Nothing on this page is advice. Nothing here describes what
any price will do next. It describes what already happened in this sample.