What constitutes a viable system?

Let's say I run a simple system against the data files from the WOO Baccarat section and get the following results:

StandardFileStream: '/Squeak/bac-sim-25k-1.txt'
Winners: 13895 Losers: 11109
Net bankroll = 81006 Average win per shoe: 3.239721644536874

StandardFileStream: '/Squeak/bac-sim-25k-2.txt'
Winners: 14095 Losers: 10909
Net bankroll = 89362 Average win per shoe: 3.573908174692049

StandardFileStream: '/Squeak/bac-sim-25k-3.txt'
Winners: 14113 Losers: 10891
Net bankroll = 85500 Average win per shoe: 3.419452887537994

StandardFileStream: '/Squeak/bac-sim-25k-4.txt'
Winners: 14062 Losers: 10942
Net bankroll = 83617 Average win per shoe: 3.3441449368101104

StandardFileStream: '/Squeak/bac-sim-25k-5.txt'
Winners: 14014 Losers: 10990
Net bankroll = 86310 Average win per shoe: 3.4518477043673013

StandardFileStream: '/Squeak/bac-sim-25k-6.txt'
Winners: 14073 Losers: 10931
Net bankroll = 88678 Average win per shoe: 3.5465525515917453

StandardFileStream: '/Squeak/bac-sim-25k-7.txt'
Winners: 14019 Losers: 10985
Net bankroll = 79774 Average win per shoe: 3.190449528075508

StandardFileStream: '/Squeak/bac-sim-25k-8.txt'
Winners: 14113 Losers: 10891
Net bankroll = 91855 Average win per shoe: 3.673612222044473

StandardFileStream: '/Squeak/bac-sim-25k-9.txt'
Winners: 13913 Losers: 11091
Net bankroll = 81316 Average win per shoe: 3.2521196608542633

StandardFileStream: '/Squeak/bac-sim-25k-10.txt'
Winners: 13967 Losers: 11037
Net bankroll = 80605 Average win per shoe: 3.223684210526316

If you are betting in $10 units that yields about $30 / shoe.

The first question is do you think this is viable?
A side question: Do these results warrant (that is, are they statistically significant) the effort to run the system against a billion shoes?

Next let's look at a sub-sample of the data in more detail:

Shoe: 1 Hands: 79 Bank: -35 Min: -55 Max: 8
Shoe: 2 Hands: 81 Bank: 21 Min: -3 Max: 39
Shoe: 3 Hands: 85 Bank: -2 Min: -15 Max: 18
Shoe: 4 Hands: 82 Bank: 39 Min: -5 Max: 50
Shoe: 5 Hands: 80 Bank: 60 Min: -7 Max: 64
Shoe: 6 Hands: 81 Bank: 59 Min: -21 Max: 64
Shoe: 7 Hands: 80 Bank: -26 Min: -26 Max: 25
Shoe: 8 Hands: 79 Bank: 34 Min: -9 Max: 40
Shoe: 9 Hands: 82 Bank: 31 Min: -9 Max: 40
Shoe: 10 Hands: 80 Bank: 7 Min: -22 Max: 29
Shoe: 11 Hands: 79 Bank: 3 Min: -15 Max: 16
Shoe: 12 Hands: 82 Bank: -4 Min: -9 Max: 19
Shoe: 13 Hands: 79 Bank: 30 Min: 0 Max: 34
Shoe: 14 Hands: 80 Bank: 58 Min: 0 Max: 58
Shoe: 15 Hands: 80 Bank: 0 Min: -24 Max: 25
Shoe: 16 Hands: 78 Bank: -39 Min: -39 Max: 6
Shoe: 17 Hands: 78 Bank: -16 Min: -28 Max: 2
Shoe: 18 Hands: 80 Bank: 29 Min: -2 Max: 31
Shoe: 19 Hands: 82 Bank: 23 Min: -28 Max: 37
Shoe: 20 Hands: 82 Bank: 17 Min: -26 Max: 17
Shoe: 21 Hands: 80 Bank: -25 Min: -44 Max: 6
Shoe: 22 Hands: 81 Bank: -2 Min: -14 Max: 18
Shoe: 23 Hands: 80 Bank: -24 Min: -29 Max: 25
Shoe: 24 Hands: 79 Bank: -28 Min: -41 Max: 9
Shoe: 25 Hands: 81 Bank: -18 Min: -28 Max: 25
Shoe: 26 Hands: 82 Bank: -8 Min: -40 Max: 2
Shoe: 27 Hands: 84 Bank: 42 Min: -12 Max: 44
Shoe: 28 Hands: 84 Bank: 23 Min: -1 Max: 29
Shoe: 29 Hands: 81 Bank: -8 Min: -34 Max: 1
Shoe: 30 Hands: 81 Bank: 2 Min: -14 Max: 15
StandardFileStream: '/Squeak/30shoes.txt' (The first 30 shoes from bac-sim-25k-1.txt)
Winners: 17 Losers: 13
Total bankroll = 243

Given that the largest net loss was 39 units and the minimum bankroll at any given point was -55 units,
and that there were six losing shoes in a row (shoes 21 - 26) is this system useful?

This can also bring up questions about what constitutes a system.
Most definitions (including Michael's) state that the system must be followed to the (bitter) end,
but what if we apply a simple stop-loss rule on the above? If we have a cut-off of -20 units then
we have a net loss on shoes 6, 10, 19, & 20 , but the stop-loss saves us on shoes 1, 7, 16, 17, 21, 24, 25, 26, & 29.
Is this no longer a system because we don't bet on every hand, or is it still s a system because the stop-loss rule is absolute?

Cheers

Quote: Kendig

but what if we apply a simple stop-loss rule on the above? If we have a cut-off of -20 units then
we have a net loss on shoes 6, 10, 19, & 20 , but the stop-loss saves us on shoes 1, 7, 16, 17, 21, 24, 25, 26, & 29.
Is this no longer a system because we don't bet on every hand, or is it still s a system because the stop-loss rule is absolute?

What makes the stop loss helpful? Does your system count cards or follow some other method to determine that the remainder of a given shoe is unfavorable? If not, what is the benefit of sitting out the remainder of a shoe in order to start fresh on the next one? Aren't you just wasting time on a potentially winning system in that case?

I'm not sure a stop loss would do much either. If you come up with a system (for any game) where you find your standard deviations (or top loss, top win), it doesn't matter if you when you stop so long as everything is within those bounds. That's like me saying "I know my SD for blackjack is -$1,000 in any given session, so I'll always quit when I lose $500 and I'll make more money!" This just isn't true. You're taking away all of the scenarios where you're down more than $500, then come back to be UP that trip/session/etc. You're not only taking the bad out, you're taking the good out too. Statistically, a stop loss shouldn't help. This is something that people should use when they can only afford to gamble a certain amount, or want to spread their trip bankroll out over a few days, etc...

So what's this system that consistently yields approx 3.4 units per shoe? I'd love to see your system run for a couple hundred million shoes, although from the data you have above 3.2 to 3.5 does seem to be a converging point.

Quote: rdw4potus

What makes the stop loss helpful? Does your system count cards or follow some other method to determine that the remainder of a given shoe is unfavorable? If not, what is the benefit of sitting out the remainder of a shoe in order to start fresh on the next one? Aren't you just wasting time on a potentially winning system in that case?

Because people want to lose everything over a long period of time instead of all at once.

Quote: rdw4potus

Aren't you just wasting time on a potentially winning system in that case?

I chose the -20 unit limit based on a cursory look at the data in that only four shoes which reached -20 units came back for a net win. Admittedly, shoe 6 came all the way back to +59, so I would have to do more testing to find out if the net result is worth it.

I'm more interested in the philosophy: is a system that doesn't bet on every hand still a system?

Quote: Kendig

I'm more interested in the philosophy: is a system that doesn't bet on every hand still a system?

Yes.

Quote: Romes

So what's this system that consistently yields approx 3.4 units per shoe? I'd love to see your system run for a couple hundred million shoes, although from the data you have above 3.2 to 3.5 does seem to be a converging point.

It isn't really a system yet. I've just been playing around with some ideas.
What I'm trying to find is the answer to the question posed in the title of the thread.

If I run the simulation against some (very large) number of shoes and come up with a positive result,
is that the only thing that matters?
What about variance? In the data above, we saw six losing shoes in a row.
If I look through the results of the 250k shoes and find 15 losing shoes in a row does that matter?

So besides the fact that my ideas are in the very early stages of development there are two practical reasons that I'm not going to disclose what I'm doing.
1) I don't want be halfway through a shoe at my local casino and find myself getting banned because of what I write here, and,
2) if I do get banned simply because I win too frequently, I don't want to wake up one day and find that someone else is selling my ideas for $500 a pop.

I trust you understand.

Quote: Kendig

It isn't really a system yet. I've just been playing around with some ideas.
What I'm trying to find is the answer to the question posed in the title of the thread.

If I run the simulation against some (very large) number of shoes and come up with a positive result,
is that the only thing that matters?
What about variance? In the data above, we saw six losing shoes in a row.
If I look through the results of the 250k shoes and find 15 losing shoes in a row does that matter?

So besides the fact that my ideas are in the very early stages of development there are two practical reasons that I'm not going to disclose what I'm doing.
1) I don't want be halfway through a shoe at my local casino and find myself getting banned because of what I write here, and,
2) if I do get banned simply because I win too frequently, I don't want to wake up one day and find that someone else is selling my ideas for $500 a pop.

I trust you understand.

You gotta have a winning system before you have to worry about any of that stuff.

Quote: Kendig

It isn't really a system yet. I've just been playing around with some ideas.
What I'm trying to find is the answer to the question posed in the title of the thread.

If I run the simulation against some (very large) number of shoes and come up with a positive result,
is that the only thing that matters?
What about variance? In the data above, we saw six losing shoes in a row.
If I look through the results of the 250k shoes and find 15 losing shoes in a row does that matter?

So besides the fact that my ideas are in the very early stages of development there are two practical reasons that I'm not going to disclose what I'm doing.
1) I don't want be halfway through a shoe at my local casino and find myself getting banned because of what I write here, and,
2) if I do get banned simply because I win too frequently, I don't want to wake up one day and find that someone else is selling my ideas for $500 a pop.

I trust you understand.

You have absolutely nothing to worry about for many many reason's.

I think the fallacy is that systems can be viable. No system can overcome negative expectation games, especially when run to infinity. When you run any system to infinity the expected loss always manifests itself.

I remember how many wall street Quant's professed how.accurate their systems were 99% of the time. That 1% of the time cost them their jobs, and the rest of ours too.

So that leads us to redefine what your limits are that qualify for viability. As a craps player, we all love and hate the iron cross betting strategy.

Quote: Kendig

So besides the fact that my ideas are in the very early stages of development there are two practical reasons that I'm not going to disclose what I'm doing.
1) I don't want be halfway through a shoe at my local casino and find myself getting banned because of what I write here, and,
2) if I do get banned simply because I win too frequently, I don't want to wake up one day and find that someone else is selling my ideas for $500 a pop.

It looks like you're working up a baccarat system. I think you have very little to worry about.

Baccarat has fewer identifiable moments of opportunity than blackjack, and they're harder to exploit. If your system was somehow able to accurately predict future results, and you were exploiting it heavily (table max bets), then maybe you have something to worry about.

I hope your system works as well as you fear...

Quote: Asswhoopermcdaddy

I think the fallacy is that systems can be viable. No system can overcome negative expectation games, especially when run to infinity. When you run any system to infinity the expected loss always manifests itself.

Actually, when you run a system to "infinity", occasionally they work. That's the problem - long before you can get to infinity, the sun burns up all of its hydrogen, and Earth becomes inhabitable, so every system has a stop point somewhere. ("Infinity" usually also always means an infinite amount of money, but if you have that, then why are you worried about making more through gambling?)