Oscar's Grind Question

I was recently turned on to this site when a friend was telling me about Oscar's grind and how he rarely loses playing this system. Now I have done the online research and seen lots of info on this site telling me its a bad idea. I tested it out myself and found that the system seems to work. The system is just a Oscar's grind with these limits: 1. Start with exactly 20 times the minimum bet. 2. Walk away from the table when you double your money. I tried it out on the simulator and these were my results:

Out of 16 sessions, I lost my money 4 times.
857 hands. 425 wins, 432 losses. Pushes were not counted.

At a $10 table this would have been a $1600 gain. The math seems to add up (49.6% win ratio). Was I just lucky? Can anyone tell me what is the odds are of losing your $200 if you play this system?

May I refer you to FoolsGold's comments in blog.oddhead.com of eons ago:

Mohammad Mahdian says:
I don’t know much about craps, but there is a 1965 paper by Tom Ferguson that solves a very similar problem for general MDPs:
http://www.math.ucla.edu/~tom/papers/ruin.pdf

Fools Gold says:
March 18, 2007 at 9:45 am

A Statistical Characterization and Comparison of Selected Craps Money Management and Bet Selection Systems. Ken Elliot Kbeiico. Columbia, MD 21046

As to the ‘max odds’ question: the mathematical answer seems to be that you should never make a line bet on which you are unable or unwilling to take full odds. Now ofcourse this can make the experience short and sweet if you lose that first bet or the first few bets. 100x odds are often publicity promotions with most players sticking to five or ten times odds, yet Casino Royale (LV-strip) offers one dollar line bets and 100 times odds. Best strategy: $606 PutBet. After three hits, take it down and leave the flat bets for the dealers.

“Better” depends upon goals and preferences. Few people actually trek to Vegas to put their entire bankroll on that first bet despite what mathematicians say is optimal strategy. More bets means more entertainment time, more free drinks, etc. The answer would also depend upon bet progression strategies that you intended to employ. If all you are ‘comfortable’ with is a table minimum bet with five times odds, then go for that. A line bet followed by two come bets would be adequate to stay in the game with your bankroll and catch any ‘hot streaks’. Hedge systems don’t make sense. Grinders don’t really do well anyway.

what you want to maximize is the overall mean length: that is the number of rolls before exhausting your bankroll or making your pre-determined win-goal.
At 5 dollar bets with two times odds, Hoyle’s Press and Oscar would be the two best systems, though Ponzer, in third place, would be preferable overall.
Best thing would be to play at a slower table where there are lots of grinders who will slow things down even more. That maximizes your time at the table more cheaply than switching to a Patrick Right system.

The reference in comment4 carried out simulations to 800 rolls. I would ofcourse question whether this is really enough.
When a statistics professor had a pair of shaved dice manufactured to precision standards he found that the ‘edge’ could not be detected in fewer than one million rolls, so I wonder if any Monte Carlo simulation would really prove anything.
URL for comment Four:
http://www.conjelco.com/downloads/elliott-paper.pdf
A statistical Characterization and Comparison of Selected Craps Money Management and Bet Selection Systems. Ken Elliott.
Elliott paper is also available in:
Finding the Edge: Mathematical and Quantitative Analysis of Gambling (Institute of Gambling & Commercial Gaming) (Hardcover)
by Olaf Vancura (Editor), William Eadington (Editor), Judy Cornelius (Editor)

This can be "solved" in something like 150 simultaneous equations.

However, I need a clarification: if your bankroll is reduced to 20 but your next bet should be 30:
(a) do you bet 20, or exceed your bankroll and bet 30;
(b) if you bet 20, and then win, is your next bet 30 (i.e. 10 more than what you actually bet) or 40 (i.e. 10 more than what you "should have" bet)?

A. The way I tested the system I would never risk more than the 20 units. So in that scenario where my bankroll was reduced to 2 units and the progression bet was at 3 units, I would just bet the 2 units I had left. If that hand was a win,, I would continue the progression by betting 4 units. Thanks for your help.

Quote: jfire353

I tested it out myself and found that the system seems to work. The system is just a Oscar's grind with these limits: 1. Start with exactly 20 times the minimum bet. 2. Walk away from the table when you double your money. I tried it out on the simulator and these were my results:

the original Oscar played Craps from what I have read.
what game did you play?

This is easily simulated BTW. $200 bankroll and $10 unit bets in Craps
I get 55% chance of a session ruin playing with no odds. The odds would make this way more interesting I would think.
(this can also be setup as a Markov chain for ruin or quit at $400 as an example)

Stewart N. Ethier
The Doctrine of Chances
Probabilistic Aspects of Gambling
Has a few sections on this Oscar system for the probability of winning 1 unit with a max table bet of 500 units
(requires a very large bankroll too)
more than the OP
Fun and interesting reading. from page 288

"This leads to
a recursive algorithm that is well suited to numerical computation for large
M. For example, if M = 500, the system has 41,917,000 equations. Nevertheless,
a numerical solution is feasible, and when p = 244/495, it has been
found that Q(1, 1) ≈ 0.999740807832."
he uses some other parameters for his calculation too about when to abort the system

a very high probability of winning one unit one time while risking a massive loss with a low probability of losing
sounds all too familiar

Sally

I'm sorry I should have clarified... The game was Blackjack. I am neither an avid gambler or a mathematician. I used a strategy card to play a perfect strategy. I have a Vegas trip coming up and I'm just looking for a way to maximize my play time and win some money if I can. I understand that this system is probably flawed in some way, I just don't have the mathematical skills to prove or disprove it. The data I collected had peeked my interest. If I am only risking $200 at a time, my risk of ruin seems quite low.

I resorted to Monte Carloing the problem SHAME...

For even-money bets on double-zero roulette, it's about a 35.2% success rate

For a game with 1% house edge (winning probability = 49.5%), it's about 46.22% success.

Quote: mustangsally

the original Oscar played Craps from what I have read.
what game did you play?

This is easily simulated BTW. $200 bankroll and $10 unit bets in Craps
I get 55% chance of a session ruin playing with no odds.

just to compare one system to another system

a craps player just flat betting $10 trying to double a $200 bankroll
success = 36.22%
=1/((q/p)^b_x+1)
b(x)units: 20
p=244/495
q=1-p

But we know the Oscar bettor raises the bet after a win one unit or less to hit the one unit win goal before starting the progression all over again
The Oscar example I showed had an average bet of almost $20
That is how he was able to have a higher probability of doubling the $200 bankroll
(betting it all one time would give a 49.3% success rate)

How about the flat bettor now betting $20. That gives a 10 unit bankroll
success = 42.96%
closer to Oscar system

Is the Oscar in my example "better" than flat betting?
better = ?
how fast is ruin with Oscar compared to flat betting?
that is another question some may be interested in. time playing instead of how much can I win.

different bankrolls (and bet sizes) and win goals and time goals mean way different results
your mileage and fun will vary

Oscar
"Named after a hot dog, you poor man, you poor, poor man"
many LOVE hot dogs
Sally

Quote: jfire353

I used a strategy card to play a perfect strategy. I have a Vegas trip coming up and I'm just looking for a way to maximize my play time and win some money if I can.

just flat bet to maximize play time (or no bet at times if they will allow this) and have lots of fun. you win by luck and basic strategy. better than a rabbit's foot.

Quote: jfire353

I understand that this system is probably flawed in some way, I just don't have the mathematical skills to prove or disprove it. The data I collected had peeked my interest. If I am only risking $200 at a time, my risk of ruin seems quite low.

all systems are "flawed" playing against a house edge game.
the number of times you played was low.

like flipping a coin 10 times and betting on Heads. one session you could get 6 or 7 or even more Heads.
You may now think you are a master coin flipper and can win money flipping coins.
Play more sessions and you will see those sessions that have more Tails than Heads.
until then, have fun playing and if you win, winning is only because
and not because you have some system that if everyone played would close all the casinos or force then to change all the games.

some more info from one that runs simulations for blackjack using many different betting systems
7-systems-and-3-ways-to-play-blackjack

I too will be in Las Vegas at the end of this month.
do not win all the money, I want to play and win some too
Sally

urgh, don't try to change the EV with betting systems. If you're using a betting system it'd better be for other reasons.

This Oscar System was first introduced by Allan Wilson in his "Gambler´s guide" by 1965.

Wilson( a mathematician and nuclear physicic) felt surprised how Oscar stood strong fluctuations.

He tested it, with his early computers and realised fluctuations would raise up to 50 units with 10k units requiered.

He knew the odd perfectly and knew you could not win without an edge.

Eventhough, oscar´s system stood too much, he tested this system with 0% desadvantage too.

Now, what if we have got at least 1% edge over 18 numbers played using Oscar´s system?

How could we find out it performance against flattbetting +1% on these 18 numbers?

regards