RideTheEdge
RideTheEdge
Joined: Apr 30, 2020
  • Threads: 1
  • Posts: 95
April 30th, 2020 at 2:47:28 PM permalink
I'm thinking about calculating the RTP for non-bonus rounds on video slots by figuring out which symbols are on the reels from youtube videos and then applying the paytables to all possible outcomes weighted by their probability. I'll need to figure out if there is a virtual reel to weight properly (likely punt on this idea of there is a virtual reel). Does anyone have advice they are willing to share?

My first questions are about the symbols flying by between spins. I suppose I have to figure out the answers, but am sharing my challenges and seeking comment. As a proof of concept I've recorded the symbols for a few spins, and it isn't hard (though it is time consuming) to advance the video a few frames at a time and catch every symbol. Questions:
- Are these symbols "spinning by" actually part of the reel, or are they random?
- If they are random, do they occur at the same frequency (almost as good as knowing the actual reel) when they are flying by and at the end of the spin?
- For the machine I looked at first, it seemed like each spin had a different "stack" symbol that appeared in groups of four. Same stack symbol for every reel on any given spin. This complicates figuring out the reel since it is unclear how the stack is added:
* Stacks added randomly or at deterministic spots on the reel?
* Stacks added where that symbol would normally appear (expand single symbol to four)?
* Stacks "overwrite" symbols that are on the reel or inserted between them?
* For symbols that stack (some don't), is the probability of a symbol appearing as a stack the same as that symbol appearing alone?
* In any case, I played this machine for a couple months and didn't realize it had stacked symbols at all so looking at the symbols between spins was good for that.
- How many symbols are there likely to be in a reel on a video slot? After looking at about 350 symbols across 7 spins on 1 reel I couldn't spot a repeating pattern, so I think this machine either has (much) more than 64 symbols, or the symbols between spins are irrelevant (except for revealing the stack symbol and perhaps symbol probabilities), or I was confused by the stacking and missed any repeating pattern.
- Not a question: anticipating the reels may be different depending on the RTP for a specific machine, I probably can't combine multiple videos to get a larger sample size. Maybe after getting a large enough sample across multiple machines there will be a way to combine the spins (e.g. if I identify how the RTP is manipulated).

Apologies if I got some of the terminology wrong. I've only been playing slots for a few months.

Steve
DRich
DRich
Joined: Jul 6, 2012
  • Threads: 79
  • Posts: 8112
Thanks for this post from:
heatmap
April 30th, 2020 at 3:24:14 PM permalink
I have written video slots both ways. Some actually show the exact symbols rolling down the reels and other ones just had a standard "blur" effect showing symbols rolling down that do not correlate to the actual reels.
Order from chaos
RideTheEdge
RideTheEdge
Joined: Apr 30, 2020
  • Threads: 1
  • Posts: 95
April 30th, 2020 at 3:30:38 PM permalink
Thanks!
rsactuary
rsactuary 
Joined: Sep 6, 2014
  • Threads: 22
  • Posts: 1880
Thanks for this post from:
RideTheEdge
April 30th, 2020 at 5:00:35 PM permalink
One of the tricks is understanding where do they "cut" the reel? Note that all reels stop at precisely the same time interval apart, except when you perhaps need the third symbol for the bonus. They can do this because at some point in time they cut to a part in the reel so that when it rolls around it stops at the right time interval. Hope that makes sense.

I do believe some actually show the reel, others show a blur and are not likely following the actual reel (Green machine is a good example of this - if you watch you tube videos, on each spin you are likely to see a high limit number show up passing by - but they rarely land).
RideTheEdge
RideTheEdge
Joined: Apr 30, 2020
  • Threads: 1
  • Posts: 95
April 30th, 2020 at 5:31:21 PM permalink
That's very interesting rsactuary. On the machine I looked at closely, about 6 or 7 additional symbols went by on successive reels. For instance, 24 symbols on reel 1, 30 on reel 2, 36 on reel 3, 43 on reel 4, and 50 on reel 5. I thought, "ah reel 5 has the most symbols, I'll stitch a few spins together and I'll have hundreds in a row, easily enough to find a 64 symbol repeating pattern." I guess not.

So in figuring out the "cut," I guess there are several possibilities:
- The cut is made between the spins. The last symbol in spin 1 does not precede the first symbol in spin 2. For the machine I looked at this doesn't seem right because a partial stack of symbols in one spin is completed at the beginning of the next.
- Or perhaps the cut is made a fixed distance from the end of the next spin. If it is 20 symbols, then the last 20 symbols of the spin are a continuous portion of the reel, and the symbols prior to that are also a continuous portion of the wheel, but at some point there is a "jump" so the spin finishes out where the RNG told it to. This actually makes a lot of sense if the reel has hundreds of symbols on it but each spin is only 20-50 symbols. I was wondering how the results are really random if a small subset of the reel is advanced on each spin; a cut explains that.
- Or maybe the cut is somewhere else (a random location? a different fixed location for each reel?)

Both you and DRich mentioned a "blur." Does that imply the symbols are not legible as they roll by? Can I tell if they are not following the actual reel because I can't see each symbol clearly at all?
rsactuary
rsactuary 
Joined: Sep 6, 2014
  • Threads: 22
  • Posts: 1880
April 30th, 2020 at 5:42:13 PM permalink
Quote: RideTheEdge

That's very interesting rsactuary. On the machine I looked at closely, about 6 or 7 additional symbols went by on successive reels. For instance, 24 symbols on reel 1, 30 on reel 2, 36 on reel 3, 43 on reel 4, and 50 on reel 5. I thought, "ah reel 5 has the most symbols, I'll stitch a few spins together and I'll have hundreds in a row, easily enough to find a 64 symbol repeating pattern." I guess not.

So in figuring out the "cut," I guess there are several possibilities:
- The cut is made between the spins. The last symbol in spin 1 does not precede the first symbol in spin 2. For the machine I looked at this doesn't seem right because a partial stack of symbols in one spin is completed at the beginning of the next.
- Or perhaps the cut is made a fixed distance from the end of the next spin. If it is 20 symbols, then the last 20 symbols of the spin are a continuous portion of the reel, and the symbols prior to that are also a continuous portion of the wheel, but at some point there is a "jump" so the spin finishes out where the RNG told it to. This actually makes a lot of sense if the reel has hundreds of symbols on it but each spin is only 20-50 symbols. I was wondering how the results are really random if a small subset of the reel is advanced on each spin; a cut explains that.
- Or maybe the cut is somewhere else (a random location? a different fixed location for each reel?)

Both you and DRich mentioned a "blur." Does that imply the symbols are not legible as they roll by? Can I tell if they are not following the actual reel because I can't see each symbol clearly at all?



yes, you get the idea of a cut. From watching myself, I feel like Dragon Link and Lightning Link cut right after the first symbol above the top of the reel is displayed. But any machine could handle that differently. It could cut right before the final symbols roll into place. By blur I mean that some symbols are visible. If you watch some highlight Green Machine on youtube, you will see a very high frequency of large hit amounts show up during the blur. Far more than would be on the actual reels.
DRich
DRich
Joined: Jul 6, 2012
  • Threads: 79
  • Posts: 8112
Thanks for this post from:
RideTheEdge
April 30th, 2020 at 6:30:02 PM permalink
I would guess most of the newer machines will show the actual reel symbols in order. We used to do the blur on older machines that weren't as fast graphically.
Order from chaos
p13man
p13man
Joined: Jul 14, 2019
  • Threads: 4
  • Posts: 29
Thanks for this post from:
RideTheEdge
May 28th, 2020 at 11:29:48 PM permalink
I have an AI engine that can read the reels of any video slot as they spin, stitch them together and report the reel length and symbol composition of all reels. As you know, the outcome of each spin is determined before the reels start spinning so the slot can then calculate how far 'up' the reel to start so that when the reels stop, they are displaying the correct symbol(s). Each reel spins for a set number of symbols and that gets progressively larger as you move from reel #1 to reel #5. On Aristocrat (Dragon Cash for example) games, the first 4 symbols of each spin are fake and then the reel 'cuts' to the actual reel at the point determined by the RNG selected stopping position and the number of symbols in the spin.

The number of symbols per reel varies by game, and within a game. Aristocrat Golden Century (for example) has 64 symbols on the first 4 reels and 85 on the last. Games with symbol stacks will have hundreds of symbols. Reels with stacks are difficult to read as the length of the stack is hard to determine if longer than 5 or 6 symbols.

I don't think the reels are different for different RTP settings. Certain jurisdictions state the reels must always stay the same and virtual reels aren't permitted. So my guess (and it is only a guess) is that the RTP for the base game is fixed, and changes to the RTP in the machine's config are handled by loosening/tightening the bonus rounds (e.g. hold and spin coin values). Happy for someone to confirm/shoot this down this independently. To answer your question, yes you can combine games from different sources as the reels are the same.
tringlomane
tringlomane
Joined: Aug 25, 2012
  • Threads: 8
  • Posts: 6272
Thanks for this post from:
charliepatrickRideTheEdge
May 29th, 2020 at 12:43:32 AM permalink
Quote: p13man



I don't think the reels are different for different RTP settings. Certain jurisdictions state the reels must always stay the same and virtual reels aren't permitted. So my guess (and it is only a guess) is that the RTP for the base game is fixed, and changes to the RTP in the machine's config are handled by loosening/tightening the bonus rounds (e.g. hold and spin coin values). Happy for someone to confirm/shoot this down this independently. To answer your question, yes you can combine games from different sources as the reels are the same.



In Nevada, I think this is incorrect. The reels only need to stay the same when playing the same denomination, iirc.

Here's a paper that shows the difference of the payback between a 85% and a 96.2% game of "Lucky Larry's Lobstermania" The payback difference is achieved by mostly unbalancing the reels while mildly starving the more valuable reels. The bonus round frequency was left constant between the two paybacks.

See page 90 for the reel breakdowns. (Page 10 of PDF)

http://www.nh.gov/gsc/calendar/documents/20091117_harrigan_dixon.pdf
p13man
p13man
Joined: Jul 14, 2019
  • Threads: 4
  • Posts: 29
Thanks for this post from:
tringlomaneRideTheEdge
May 29th, 2020 at 1:01:03 AM permalink
Different rules in different jurisdictions for sure. If you can find (or record) Golden Century games played in a Nevada casino on youtube (at different denoms) I can tell you for sure if they conform to my reel definitions for that game. When I devised the reel reader app I checked the reported reel composition against youtube videos and they aligned, but I don't remember whether or not the denom changed in the session or where in the world the session was recorded.

It's also worth noting that LLL is by no means a modern game, and perhaps variable reels (base games) were the only way to control RTP in the olden days.

Thanks for the reference. And the shoot down!

I'm keen to get to the bottom of this. All assistance appreciated.

  • Jump to: