# What are the odds you win the WSOP? — methodology

reelgorithm.py · pulled and checked 2026-09-25, extended 2026-10-04

**What are the odds you win the World Series of Poker Main Event, and what is
your $10,000 ticket actually buying?**

Two files ship with this write-up:

| file | rows | what it is |
|---|---|---|
| `wsop-main-events.csv` | 57 | every Main Event, 1970–2026, with the derived per-ticket figures |
| `wsop-2026-payouts.csv` | 1,383 | the complete 2026 payout ladder, every paid place |

---

## 1 · The question, and what it is not

Not "how good do you have to be to win" — nobody's skill is in this table. This
is the price of a lottery ticket that happens to also be a poker tournament: how
many people you are actually up against, and what a flat $10,000 buy-in has been
worth to the average entrant across 56 years of the field getting bigger.

---

## 2 · Sources

| what | source |
|---|---|
| every Main Event 1970–2026: field size, first prize, champion, runner-up | [Wikipedia, List of WSOP Main Event champions](https://en.wikipedia.org/wiki/List_of_World_Series_of_Poker_Main_Event_champions) (pinned revision 1368075126) |
| 2026 champion | [WSOP.com](https://www.wsop.com/news/22-year-old-lucas-jumalon-wins-the-2026-wsop-main-event/) |
| 2026 payout ladder (1,383 places) | [PokerNews](https://www.pokernews.com/tours/wsop/2026-wsop/2026-wsop-main-event/payouts.htm) |
| 2026 field size (9,208) | [CardPlayer](https://www.cardplayer.com/poker-news/1680417-2026-wsop-main-event-draws-9208-to-become-fourth-largest-in-history) |
| 2026 result, cross-check | [ESPN](https://www.espn.com/espn/betting/story/_/id/49542526/washington-lucas-jumalon-22-captures-wsop-main-event) |

Buy-in: $5,000 in 1971, $10,000 every year since 1972. **1970 was a vote of the
players, not a freezeout tournament** — it has no buy-in and is excluded from
anything that needs one.

Nothing upstream is redistributed. These files are counts and the arithmetic we
did on them.

---

## 3 · The numbers

**The odds.** 9,208 entered the 2026 Main Event. A player exactly as good as the
field has a 1-in-9,208 chance — that is holding every chip on the table, which
takes log₂(9,208) ≈ **13.17 doublings**. Thirteen straight coin flips land
1-in-8,192, which is why "roughly thirteen coin flips in a row" is the honest way
to say it, not the exact one. It is a restatement of 1/N, not a second
independent estimate of it.

**The 2004 cliff.** The field went 839 → 2,576 between 2003 and 2004 — **×3.07** —
and every field since has beaten the pre-2004 record of 839. 2003 is the year
Chris Moneymaker, an accountant, won his Main Event seat through an **$86
PokerStars satellite** and then won the whole thing (Wikipedia, "Chris
Moneymaker", citing PokerStars).

**The hidden layer — what the ticket is worth.** In 1997 the field was 312 and
the first prize $1,000,000: for an average player the title was worth $3,205 of a
$10,000 ticket — **32.1%**. In 2026 the arithmetic is $10,000,000 ÷ 9,208 =
**$1,086, or 10.9%**. The field grew **×29.5** since 1997; the prize only **×10**.

| year | field | first prize | title EV per ticket | as % of a ticket |
|---|---|---|---|---|
| 1997 | 312 | $1,000,000 | $3,205 | **32.1%** |
| 2003 | 839 | $2,500,000 | $2,980 | 29.8% |
| 2004 | 2,576 | $5,000,000 | $1,941 | 19.4% |
| 2026 | 9,208 | $10,000,000 | $1,086 | **10.9%** |

**Also true, not in the video:**

- Cashing is 1 in 6.66 — 1,383 of 9,208 entrants.
- Gross buy-ins were $92,080,000 against a pool of $85,634,400, so the house
  keeps **7.0%** and the average entrant gets back **$9,300** of their $10,000.
- There have been **23 different champions in 23 Main Events of 1,000+
  entrants** — nobody has repeated in the modern era. With only 23 events that is
  not yet strong evidence of anything, which is why it is a footnote.
- A skill edge compounds across all 13.17 doublings: 55% per double-up instead of
  50% is 1 in 2,625, about **3.5×** an average player; 60% is 1 in 835. This is a
  **model, not a measurement** — there is no per-player edge in this dataset, and
  it is labelled as a model wherever it appears.

---

## 4 · Data traps, stated rather than fixed quietly

- **1970 was a vote**, and **1971's buy-in was $5,000** — both are excluded from
  any buy-in-based calculation. Rows without a buy-in or a recorded first prize
  have their derived columns left empty rather than guessed.
- **2020 (1,379 entrants) was an online + live hybrid Main Event**, the one dip
  in an otherwise-monotonic post-2004 climb. It still beats the pre-2004 record
  of 839 by 64%.
- **Two sources disagree about the bottom of the ladder, and the disagreement is
  real.** The published tournament structure says **1,382 places paid with a
  $15,000 minimum cash**. The payouts PokerNews actually recorded list **1,383
  places**, the bottom three tied at **$10,000** — the buy-in back — with $15,000
  the tier immediately above, covering places 1,151–1,380. `wsop-2026-payouts.csv`
  ships PokerNews's numbers as published, and that as-recorded ladder is the one
  that sums to the $85,634,400 pool **to the dollar**, which is why the derived
  figures use it. Both are stated here rather than silently reconciled, and
  `analysis.py` now carries both pairs as separate named constants with an
  assertion that the gap is still exactly one place.
- **Field size is entries, and the Main Event is a freezeout**, so entries equal
  players — nobody re-enters. That is a property of this event and does not
  transfer to re-entry tournaments.
- **Every dollar is nominal.** No inflation adjustment is applied anywhere. In
  real terms the first prize has grown by less than ×10, which makes the gap
  between the two multipliers wider — so the nominal version is the conservative
  one.
- **This prices the title, not the entry.** It is not the expected value of
  playing, which would include all 1,383 paid places. The pool has spread
  downward as the field grew, so cashing is a larger share of what a modern
  ticket buys even as the title is a smaller one.
- **One chip on screen is 40 players**, a rounding chosen so the 2026 stack stays
  legible next to 1997's. The chart's shape is the data; individual chip counts
  are a rendering.

---

## 5 · Columns

**`wsop-main-events.csv`** — `year`, `entrants`, `buy_in_usd`,
`gross_buy_ins_usd`, `first_prize_usd`, `winner`, `runner_up`,
`odds_of_winning_1_in`, `double_ups_to_hold_every_chip`,
`title_ev_per_ticket_usd`, `title_ev_pct_of_ticket`, `note` (flags 1970's format
and any other year-specific caveat).

**`wsop-2026-payouts.csv`** — `place`, `prize_usd`, `multiple_of_buy_in`,
`pct_of_prize_pool`, `cumulative_paid_usd`, for all 1,383 paid places. Sums to
**$85,634,400**, the published 2026 prize pool, to the dollar.

---

## 6 · Reproduce it

```
python3 data/wsop/build.py      # raw -> main_event.csv
python3 data/wsop/analysis.py   # every number above, with the ladder assertions
python3 data/wsop/kit.py        # the two CSVs in this download
```

*reelgorithm.py · every figure above is reproducible from the CSVs.*