Microgaming slots - are the odds adjustable?

[steinhaug]

If the "variance was extremely predictable", then it is not a random game.

You are completely missing the point! Setting up a predictable variance in my example was to make my example so much clearer, not to confuse you with streaks and dry runs.

Here's something else, I think it says they can change the paytable per coin size, but I could be daft and i would love to be corrected if I'm reading it wrong.

Thanks for looking into this. Using different variance, or paytables on different coinsizes does NOT need to change the 95% payout. It only changes what type of wins, lots of small wins on low wagering and more high wins on high wagering.

My point in it all - is that MG slots must work the same way, or they will go bust. I would think it's in the nature of all video slots with different denominations to do this.

With these schemes, every time a player gets one over on the house by hitting the big features and wins on big bets, there will be one who gets the opposite, with big wins on small bets given back with bad spells when betting large.

As you see in some documents, it's normal to do this. (Different variance on different consizes.) The problem is however - that the reels are still the same... What does that this tell us about the RNG in the background? Are we truly playing against RNG or are we playing against RNG which is weighted as Zoozie likes to say it. (The answer is ofcource weighted RNG since true RNG with same reels must give same variance)

I think the people in here which still doesn't understand what I am talking about, or still are locked on thinking I am trying to bust RNG here must read my post all over again. I might try to rewamp a new post with more examples so all get the feel of what I am explicitly trying to discuss here.

 

[aka23]

Thanks for looking into this. Using different variance, or paytables on different coinsizes does NOT need to change the 95% payout. It only changes what type of wins, lots of small wins on low wagering and more high wins on high wagering.

My point in it all - is that MG slots must work the same way, or they will go bust. I would think it's in the nature of all video slots with different denominations to do this.

You are correct that I am still not certain what you are getting at. The variance per spin/roll/hand of a casino game is dependent on the probability of each payout. These probabilities should not change with bet size. For example the standard deviation per hand of BJ is ~1.15. For slots, the expected standard deviation per spin is much higher, often >10. These numbers do not vary with bet size since they are expressed in terms of the full initial bet. Yet the variance in range of return increases in a predictable way as bet size increases since range of return is usually not expressed in terms of initial bet. If "big wins" are defined in terms of total size, you will get big wins more often with larger bets. However, if "big wins" are defined as a proportion of the initial bet, you should get the big wins equally often with any bet size.

 

[spearmaster]

I'm not sure I understand... variable betting units. Almost all slots have multidenominational input - coin size. I may have misunderstood.

Different size coins. I disagree completely, very few slots in Vegas allow you to switch from pennies to nickels to quarters. VP is a different story entirely.

Here's something else, I think it says they can change the paytable per coin size, but I could be daft and i would love to be corrected if I'm reading it wrong.

Again, this is usually done on VP machines - I have played on these machines and have seen the paytables change according to coin size.

Note that for slot machines, what you pointed out applies to single-coin vs. multi-coin - and here we're almost certainly talking three-reel classic machines.

 

[steinhaug]

Let my try another example, looking back (not to confuse the
random people in here) say we had this sequence of events.

0 0 0 x 0 0 0 0 0 0

Every spin is at $10, and the x symbolizes a win on $95. This would
be 95% payout on the $10 wager since it's 10 spins.

Now say what some of you people state is true, true random whatever
coinsize. My point is, say that you get the luck of squeezing out
three spins on $1 after the big win. Your wager would then be:

$10 $10 $10 $10 $1 $1 $1 $10 $10 $10

Total wager : $73
Total payout: $95

As you see, you totally skewed the payout from 95% to 130%, hence
it must be different variance on every consize.

Is it me - or is it you? I'm trying to get my head around this.

 

[vinylweatherman]

Let my try another example, looking back (not to confuse the
random people in here) say we had this sequence of events.

0 0 0 x 0 0 0 0 0 0

Every spin is at $10, and the x symbolizes a win on $95. This would
be 95% payout on the $10 wager since it's 10 spins.

Now say what some of you people state is true, true random whatever
coinsize. My point is, say that you get the luck of squeezing out
three spins on $1 after the big win. Your wager would then be:

$10 $10 $10 $10 $1 $1 $1 $10 $10 $10

Total wager : $73
Total payout: $95

As you see, you totally skewed the payout from 95% to 130%, hence
it must be different variance on every consize.

Is it me - or is it you? I'm trying to get my head around this.

You could have got another big payout in one of the three $1 spins, followed by a dry spell after going back to $10. Overall, the advantage would evaporate back to the expected return. This would only be a workable system if you could KNOW that the three spins following a big win were destined to lose, and that returning to $10 after a given number of games returns you from a preset recovery period.
This kind of thing would only work on UK Fruit Machines, and has no place in a genuinely random games.

The argument about MG slots not being entirely random is more complicated than this, and boils down to the fact they can exhibit long term tendencies, such as being predominantly "hot" for an extended number of spins (thousands), or predominantly "cold" over a similar period.
The coin size argument is that changing denomination and number of coins can switch a game from predominantly "cold" to predominantly "hot". Where a big win, or series of wins has occured, the slot may appear to go very cold (no scatters when many used to appear). The argument is that by changing coin at this point can "find" a new "hot" streak, rather than playing through the "cold" streak after a run of big wins. The principle of the game being random means this simply won't work long term, although it will work sometimes in the short term. If the game used seeds that were non-random, it would mean the RNG output would not itself be 100% random, but would be biased towards the seeds, and could indeed produce predominantly "hot" or "cold" streaks, rather than a "flat" near 95% game. Add to this the high variance on slots, and you can get dramatic swings.

Better off looking at Blackjack, a low variance game, but MG BJ often exhibits long streaks where the dealer CONSTANTLY beats you with Blackjacks galore, 20/21 outs on really bad "up" cards, while you cannot get past 16 without drawing a bust card almost every time.
I have noticed that during these bad streaks, fewer 10 value cards than expected are produced, and the dealer thus survives more bad hands. A "10 poor" shoe in a land casino does indeed favour the dealer.

When the PLAYER is on a winning streak, there are many more 10 cards drawn, these often bust the dealer, and help the player win many of the double down hands.

 

[Zoozie]

Y Better off looking at Blackjack, a low variance game, but MG BJ often exhibits log streaks where the dealer CONSTANTLY beats you with Blackjacks galore, 20/21 outs on really bad "up" cards, while you cannot get past 16 without drawing a bust card almost every time.

I believe you mean LONG streaks, but you are actually also quite right in your mistake. Flipping a coin n times will on average have the longest streak
log2(n). And blackjack is close to this coin flipping analogy.

So for 256 coin flips you will on average find a streak of 8 identical results.
But you will not be able to predict anything of course.

 

[vinylweatherman]

I believe you mean LONG streaks, but you are actually also quite right in your mistake. Flipping a coin n times will on average have the longest streak
log2(n). And blackjack is close to this coin flipping analogy.

So for 256 coin flips you will on average find a streak of 8 identical results.
But you will not be able to predict anything of course.

That's what I wrote, you must have "Monday Morning Eyes"

 

[aka23]

As you see, you totally skewed the payout from 95% to 130%, hence
it must be different variance on every consize.

Is it me - or is it you? I'm trying to get my head around this.

Perhaps we need to define what variance means.

Variance is defined as the sum of the squares of deviation from the mean value. For a casino game, variance is the sum of the squares of the difference between all possible payouts and the mean value * the probability of that payout. For example, the variance of a 1:1 bet in single zero roulette is calculated as below, where HE is the house edge.
18/37*(1 - HE)^2 + 19/37*(-1 - HE)^2 = .9993

A single number bet has the same house edge, but a much higher variance:
1/37*(36 - HE)^2 + 36/37*(-1 - HE)^2 = 36.001

If you want to plug in particular coin values instead of referencing payouts to the initial bet, then you can see that the variance calculation increases as bet size increases. For example with $10 coins, then the first calculation becomes:
18/37*($10 - HE*$10)^2 + 19/37*(-$10 - HE*$10)^2 = $99.27

Lets assume the variance of a particular slot is 100 and house edge 5% for easy calculations. Standard deviation is sqrt(variance) = 10 in this example.

If you make 10, $10 bets (your first example), then the mean return is -$5. However, you are unlikely to get exactly -$5. You may win big or lose big. Approximating as a normal curve, produces a 1 standard deviation range of $316. This suggests that losing or gaining $100 is normal result. If you make 10, $1 bets (instead of $10), then the mean return is -$0.50. The 1SD range of return is -$31.6... exactly 1/10 of the $316 SD with 10x larger bets.

In your both of your examples, there is a huge normal range of return because of the huge variance of slots. The variance is the same with different coin sizes, if defined in terms of initial bet. The variance increases with larger coin sizes, if defined in terms of $s.