The feed of wins, and the denominator it is missing

A gambling site shows you wins constantly. A lobby feed scrolls, a celebration screen fires, a meter ticks, a streamer posts a clip, a push message says you won. Every one of those is a numerator with the denominator filed off. This desk puts the denominator back, and computes what the same hour looks like for the players rather than for the display.

Open the partner account Read a feed in six minutes

Feed stub

The window: one rolling hour
Rows shown: 12
Spins pooled: 1,900,000
Your share: 0.021% of them

the barThe multiple a win must reach to be eligible for the display - on the sample, 250 times the stake

the poolThe players and spins the rows were drawn from - on the sample, 4,750 players placing 1,900,000 spins in the hour

the intervalThe hours between one player's own qualifying wins - on the sample, 10.4 hours at 400 spins an hour

Direct answerA recent-wins feed is a small, non-random sample of the wins that happened while you were watching. Covering the same hour, the sample lobby shows 12 rows drawn from 457 qualifying wins across 4,750 players, and it draws them from the rarest win band at sixteen times that band's real share. What the display tells you is that wins happen to somebody; it cannot tell you how often one happens to you.

A numerator with no denominator

Every claim in a wins display is true. Somebody did win that amount, and the amount is on the screen. What is missing is the other half of a rate: the number of attempts, the number of players making them, and the width of the window the rows are drawn from. A rate needs a denominator, and a display is built to show the numerator.

The numerator

One row: a stake, a return and a multiple. It is a fact, and it is checkable in principle - which is what makes a display built from rows so persuasive and so incomplete at the same time.

The denominator

How many spins were placed to produce that row, and by how many people. This is the number a feed does not carry, and on the sample it is 1,900,000 spins and 4,750 players for twelve rows.

The threshold

Which wins are eligible. On the sample the stated bar is a 250-times win, and the feed shows 2.6% of the wins that cleared it - so the threshold is necessary but nowhere near sufficient.

The selection

Of the eligible wins, which twelve. The sample feed takes its rows from the top and middle bands far out of proportion to how often those bands occur, which is the part of a display a reader can actually measure.

The hour, computed

Everything on this desk is derived from one rolling hour on one invented casino, and it is small enough to re-derive by hand. The rate of a 250-times win is the only externally-shaped number in it; every other figure follows from that rate and the hour's volume.

The same hour, from three directionsplayers in the hour                                    4,750
spins per player                                        400
spins in the hour                      4,750 x 400 = 1,900,000
rate of a 250x win or better                      1 in 4,155
qualifying wins                    1,900,000 / 4,155 =     457
rows the feed shows                                      12
share of qualifying wins shown        12 / 457 = 2.6% (1 in 38)
qualifying wins per player       457 / 4,750 = 0.096, or 1 in 10.4 hours

The feed is not lying and it is not empty. It is a 2.6% sample of a number that is large because 4,750 people are playing, and it is displayed to each of them as though it were the weather.

What this desk is not saying. Nothing here accuses a site of inventing a win. The rows in a feed are real rounds, and the arithmetic below works precisely because they are - a display does not need a single false row to leave a reader with the wrong rate. No real operator, game, studio or streamer is named as misleading anywhere on this site, and every figure is derived from six invented samples.

Where every number on this desk comes from

No operator, game, studio, streamer or regulator is named anywhere on this site, and no real site's lobby is described as misleading. Six invented samples carry the arithmetic, and every other figure is derived from them.

S1 - the feed

12 rows in a rolling hour; the smallest shown is a 250x win and the largest a 7,400x one, on a base stake of 1.00.

S2 - the volume

1,900,000 spins and 4,750 players in the same hour, at 400 spins each. A 250-times win lands once in 4,155 spins, so the hour contains 457 of them.

S3 - the shape

Of the 457: 375 are 250-500x, 68 are 500-2,000x and 14 are 2,000x or more. The 12 rows shown split 2 / 4 / 6 across those bands.

S4 - the reader

At 400 spins an hour a player meets a 250-times win every 10.4 hours, about 10 times a year, and sees one of them in a feed about once every 3.8 years.

S5 - the celebration

The in-game celebration fires at 25x or more - once in 380 spins, so 1.05 times in a 400-spin session - while 71% of sessions end below their starting balance.

S6 - the desk's book

40 sites: 12 run a recent-wins feed, 0 of the 12 state the threshold or the pooled volume, and 3 offer a "verify this win" link of which 1 verifies a named round.

The twelve rows, and the shape behind them

The first table is the display as a reader meets it - twelve rows and the bands they came from. The second is the same hour counted properly: what each band's real share was, and how much of the feed it took. Read together they are the whole method of this desk, because the defect is never in a row.

Where to go next

What a feed is made of

How the display is assembled, and the three things a row never carries.

The denominator

The player count the writer's news is pooled over, and one player's own line in the same ledger.

Read a feed in six minutes

The routine that produces the one number a display is arranged not to show.

What is left out

The losses, the median and the attempts a wins display omits by construction.

Sample 1 - the twelve rows of one rolling hour, each tagged with the band it came from: 2 low, 4 middle, 6 top
Time	Stake	Return	Multiple	Band
14:59	1.00	7,400.00	7,400x	7,400x
14:52	0.50	1,860.00	3,720x	3,720x
14:47	1.00	2,310.00	2,310x	2,310x
14:39	1.00	1,155.00	1,155x	1,155x
14:33	2.00	2,480.00	1,240x	1,240x
14:26	1.00	912.00	912x	912x
14:20	0.50	490.00	980x	980x
14:15	1.00	2,040.00	2,040x	2,040x
14:09	1.00	3,275.00	3,275x	3,275x
14:04	1.00	415.00	415x	415x
13:58	1.00	250.00	250x	250x
13:51	1.00	4,110.00	4,110x	4,110x
Twelve rows, one hour	13.00 staked	26,697.00 returned	median 1,640x	6 of 12 in the top band

Every row is true, and half of them come from the band that produces 3.1% of the hour's wins. The twelve rows sat behind 457 qualifying wins, 1,900,000 spins and 4,750 players in the same hour, and the display carried none of those four numbers. That is the difference between a record and a display.

The 457 qualifying wins in the hour, by band, against the twelve rows the feed showed - and the ratio between them
Band	Wins	Share of wins	Rows shown	Share of feed	Ratio
250 - 500x	375	82.1%	2	16.7%	0.20x
500 - 2,000x	68	14.9%	4	33.3%	2.24x
2,000x and up	14	3.1%	6	50.0%	16.3x
The same hour, both ways	457	100.0%	12	100.0%

The rarest band is sixteen times over-represented and the commonest is a fifth of its share. Wins of 2,000 times or more are 3.1% of the hour's qualifying wins and 50.0% of the rows; wins of 250 to 500 times are 82.1% of the wins and 16.7% of the rows. Neither band is the display's fault and both are the display's design.