Roulette lab · Lab · Context · 7 minute

Roulette computer dem

Ow a computer time di ball, who build di fus one dem, an ow accurate it haffi be.

Short ansa

A roulette computer time di ball an di rotor inna di fus few second a a spin, an name di part a di wheel weh di ball fi land. Edward Thorp an Claude Shannon build di fus one inna 1961, an di Eudaemons build digital one fram 1976. Di method need di casino fi tek bet afta di launch: live studio now close bettin at di launch, an Nevada ban dem kinda device fram 1985.

fus build
1961, Thorp an Shannon
wah it measure
ball an rotor timin
useless wen di error pass
bout 17 pocket
ban inna Nevada
fram 1985

Ow dem build di fus roulette computer system

Di idea come to Edward Thorp inna di spring a 1955, wen im did a physics graduate student at UCLA: a wheel weh mek too good fi have bias fi sen di ballgearBallDi likkle ball weh di dealer fling roun di track di opposite way to di rotor. Modern ball mek outta ivorine or Teflon, 18–21 mm.Read di whole ting → roun as predictable as a planet. Im film a cheap half-size wheel wid a stopwatch inna di frame an find seh it too irregular fi predict, an a trip to Nevada inna December 1958 show im seh casino wheel run smooth. Den blackjack tek ova. Inna November 1960, at MIT, im ask Claude Shannon fi sen im card-countin paper to di National Academy of Sciences; at di end a di meetin Shannon ask wah else im a work pon, an di roulette project move inna Shannon basement. Dem buy a regulation wheel fram Reno fi $1,500, set it pon a ole slate billiard table, an time di ball gainst a strobe light an a clock weh read to a hundredth a a second. Wid practice, dem timin error drop fram bout 0.03 to 0.01 second. By June 1961 di computer did have twelve transistor an did size like a cigarette pack: a switch inna di shoe, weh dem work wid di big toe, time di rotorgearRotorDi part a di wheel weh spin an carry di pocket dem.Read di whole ting → an di ball, an a tone inna one ear name one a di eight sectorpon di wheelSectorA piece a di wheel weh di pocket dem follow one anodda roun di arc, nuh a group pon di betting layout.Read di whole ting → a di wheel. Inna Las Vegas dat summer, Shannon time at di wheel an write dung numba like any system playa, while Thorp bet dime dem at di far end a di table. Di prediction dem hold up, but di wire dem weh thin like hair, weh run to di earpiece, keep bruk, an one time a woman beside Thorp stare inna horror at di speaker weh a stick out a im ear. Dem neva bet serious. Thorp announce di roulette computer system inna 1966 an publish di detail dem inna 1979 an 1984.

A regulation wheel pon a slate billiard table inna Shannon basement, unda a strobe.

Di fus digital computer dem

Thorp an Shannon machine did analog: di circuit a did di program, an fi change di method yuh haffi rewire it. Di microprocessor tek weh dat limit. Alan Lewis, a student at Caltech, carry a digital computer weh im build a yaad inna a camera case go a Circus Circus inna Las Vegas inna 1972, time wheel fram a balcony ova di table dem an radio prediction to a partner dung below. It run, an it neva mek nuh money; lata im learn seh im did lef out di tilt a di wheel. Fram 1976 a group a physics student at Santa Cruz, di Eudaemons, do di job proper. Dem buy a casino wheel, measure di ball, di rotor, di tilt an di bounce, an write equation wid five settin weh di playa adjust fi each wheel by clickin toe switch fi bout twenty minute bifo im bet. Dem computer name one a eight sector a few second inna di spin. At a real table inna January 1978 it gi a edge weh dem put at 25%; di hardware, weh dem strap to di body an lata build inna shoes, fail far more often dan di physics (see The Eudaemons).

Alan Lewis time wheel fram di balcony an radio prediction to a partner dung below.

Wah mek late bet did matta

A casino can stop prediction completely if it refuse bet once di ball launch. Thorp find seh almost none neva do dat: inna Reno, Las Vegas, London, Monte Carlo, Nice, Venice an Puerto Rico, di usual practice did fi call “nuh more bethow di table runNuh more betDi dealer call (or di prompt pon di screen) weh close di bettin. Nobaddy cyaan put dung, move or tek up chip again.Read di whole ting →” a revolution or two bifo di ball drop. Dat window a wah evry prediction scheme need fram den, fram di Eudaemons to di Ritz team late call pon di racetrackpon di tableRacetrackDi oval strip pon live an electronic betting layout weh show di numba dem inna wheel order, weh yuh use fi put dung neighbour an sector bet wid one tap.Read di whole ting →. Nevada outlaw prediction device inna May 1985, an live studio now close bettin at di launch.

Ow di casino dem fight back

Inna di year dem wen dem did still tek bet afta di launch, casino fight playa dem suspec through di dealer. Story bout Norman Leigh system playa dem at Nice inna 1966 an bout lata prediction team describe a new croupierpeople demCroupierDi dealer at a roulette table.Read di whole ting → evry spin, a ball weh change, spin weh mek short or fas, a table full up wid di casino own people, an at di end a table weh close. Dem story yah come fram di playa side an dem hard fi check (Goldman, 2007). Still, di tactic dem mek sense: a short or fas spin cyaan aim di ball at a pocketgearPocketOne a di 37 compartment wid numba pon di rotor. Nex name: slot, cup.Read di whole ting →, but it do mess up di timin weh a predictor measure.

New croupier, ball weh change, an a table full up wid di casino own people.

Roulette dealer can hit a numba? Di myth

Ow good di prediction haffi be

Thorp (1984) work out wah a prediction worth wen it error follow a bell curve: inna a typical spin, di pocket weh it predict off by σ pocket one way or di odda. Bet di numba it predict by itself, or it an two neighbourbet demNeighboursA numba an di two pocket pon evry side a it pon di wheel, five straight-up chip.Read di whole ting → pon each side, an di edge depend pon σ alone. A perfec prediction woulda win evry straight upbet demStraight upOne single numba, pay 35:1.Read di whole ting →. At σ = 10 pocket, di accuracy weh Thorp an Shannon reach, a straight up return bout +44%, di figure Thorp gi fi dem device; im 1998 account put it pon di eighth a di wheel weh get di most favour. Di edge drop fas as di error grow, an it gone at bout 17 pocket pon a European wheel an 16 pon a American one, wen di prediction miss by almost half di wheel. Thorp own error budget show wah mek dat limit close: separate error a 1.4 an 5.5 pocket fram timin di rotor an di ball, 7 fram di ball fall cross di deflectorgearDeflectorOne a di metal block dem pon di apron weh scatter di ball wen it a drop. Nex name: canoe, diamond, pin.Read di whole ting → an 6 fram di bounce inna di pocket dem add up to bout 10.8.

typical error σ, pocketone numbafive neighbour
0+3,500%+620%
2+611%+468%
4+258%+237%
6+139%+133%
8+79%+77%
10+44%+42%
12+22%+21%
14+8.8%+8.5%
16+2.2%+2.0%
17+0.3%+0.2%
18−0.9%−0.9%
20−2.1%−2.1%
nuh prediction−2.70%−2.70%
Wi own recalculation fi a 37-pocket wheel a Thorp table fi 38 pocket: di error normal, wrap roun di wheel, an a pocket a one unit wide.

Ow di ball slow dung

Question dem

Mi coulda build a predictor fi RNG table?

No. It nuh have nuh ball. Software draw di numba bifo di animation.

Wah bout live table weh yuh can see di launch an dem close late?

Timin coulda beat dem, an dat a why none nuh run dat way.

Fun wid maths

Try dem yuhself fus. Evry ansa fold up; open it wen yuh ready.

  1. A neighbour dembet demNeighboursA numba an di two pocket pon evry side a it pon di wheel, five straight-up chip.Read di whole ting → call put one chipmoney mattaChipA token weh stand fi a fix amount, weh yuh put pon di betting layout fi mek a bet. Nex name: check.Read di whole ting → pon a numba an pon di two pocketgearPocketOne a di 37 compartment wid numba pon di rotor. Nex name: slot, cup.Read di whole ting → each side a it pon di wheel: five chip. Ow often a prediction haffi land inna dem five pocket deh fi break even, an wah it earn if it right one time outta five?

    Show di ansa

    Ansa It break even at 5 outta 36, 13.9%, gainst 13.5% by chance alone. If it right one time outta five, it return 44% more dan it stake.

    If di call hit, one chip win 35 to 1 an come back wid it winnins, 36 chip, an di odda four lose. Suh a hit return 36 fi 5 weh yuh stake, an di call break even if it hit 5 time outta 36, 13.9%. By chance it hit 5 time outta 37, 13.5%. Dat likkle gap, less dan half a percentage point, a weh di whole 2.70% come fram.

    At one hit outta five, 0.2 × 36 = 7.2 chip come back fi evry 5 weh yuh stake, +44%. At one outta four, +80%.

    Dat a why a small gain inna accuracy worth suh much, an why casino close call betpon di tableCall betA bet yuh tell di dealer, yuh nuh put it dung wid yuh han, usually a bet pon a sector a di wheel like voisins or tiers. Nex name: “announce bet”.Read di whole ting → early at high stake: dat likkle gap a di house whole margin. Dem seh di Ritz playa dem a 2004 did bet dis way.

  2. Thorp list four independent error inna im prediction, inna pocket: 1.4 fram timin di rotorgearRotorDi part a di wheel weh spin an carry di pocket dem.Read di whole ting →, 5.5 fram timin di ballgearBallDi likkle ball weh di dealer fling roun di track di opposite way to di rotor. Modern ball mek outta ivorine or Teflon, 18–21 mm.Read di whole ting →, 7 fram di ball fall cross di deflectorgearDeflectorOne a di metal block dem pon di apron weh scatter di ball wen it a drop. Nex name: canoe, diamond, pin.Read di whole ting → an 6 fram di bounce inna di pocket dem. Ow big di total, an ow much di two timin error add?

    Show di ansa

    Ansa Bout 10.8 pocket inna all. Widout di timin error dem it woulda be 9.2, suh di two timin add only 1.6.

    Independent error nuh add up direct; a dem square add up. Di typical size a di total a di square root a di sum a di square dem: 1.4² + 5.5² + 7² + 6² = 1.96 + 30.25 + 49 + 36 = 117.2, an √117.2 ≈ 10.8.

    Widout di two timin error: √(49 + 36) = √85 ≈ 9.2.

    A total a 10.8 pocket deh inside di 17-pocket limit inna di table pon dis page an it worth bout +34% pon a single numba. Di lesson Thorp draw: wen di fall cross di deflector dem a di biggest error, betta timin nuh buy much.

Sumting fi laugh

Di physics work. Di bettin time close.

Di physics work. Di bettin time close.

Source dem

  1. Thorp, E. O. “Optimal gambling systems for favorable games.” Review of the International Statistical Institute 37(3), 1969.
  2. Thorp, E. O. The Mathematics of Gambling. Gambling Times / Lyle Stuart, 1984.
  3. Thorp, E. O. “The invention of the first wearable computer.” Second International Symposium on Wearable Computers, IEEE, 1998.
  4. Bass, T. A. The Eudaemonic Pie. Houghton Mifflin, 1985.
  5. Small, M. & Tse, C. K. “Predicting the outcome of roulette.” Chaos 22(3), 2012.
  6. Goldman, N. Great Gambling Scams. John Blake, 2007.
  7. Nevada Revised Statutes, chapter 465. Section 465.075, fi use a device fi predict di result a a game (1985).
wi check it 2026-09-28model: European wheel, single zeroow wi work it out ↗source dem ↗