Two-card system with Shadows and variable linking strategy


I’m trying to develop my two-card system at the moment. Having just completed my 3-digit major system, I figured that I wasn’t willing to spend the time necessary to build a full 2700-image system so I read a bit about the different existing variations. I’m a bit fuzzy now, has anyone tried the following variation?

One image per [Value_1] + [Suit_1] + [Value_2]
If Value_X is between Ace and 10, use the (modified) major system (Ace=“T/D”, 2=“N” … 10=0=“Q”)

Spades = 1 = “T/D”
Hearts = 2 = “N”
Clubs = 3 = “M”
Diamonds = 4 = “R”

So that’s 400 images that I get from my 3-digit system.

For knights, queens and kings I’m thinking of using vowels (“I/Y/U”, “A” and “O”) and creating the missing 276 images this way. Some of these 276 images could be reuses from the 1000-400=600 images that aren’t initally used.

Then for [Suit_2] I was thinking of the following, inspired by system descriptions from Johannes M., Alex M. and Lance T. among others.

Spades =Original image + move on to next locus
Hearts = Shadow image + link with current locus
Clubs = Original image + link with current locus
Diamonds = Shadow image + move on to next locus

Essentially: Red=“shadow image” and Suit with a pointy top=“move on”

So on top of the 276 images, I will need to create 676 shadows, but that’s easier than creating images that fit (a version of) the major system.

I’ve only tested quickly on some smaller decks (without knights, queens and kings), it works, as expected, but is very slow at the moment …

Has anyone tested such a system? Any obvious flaws?

Card deck memorisation - post your times here!

I’m actually working on something similar myself. My Major concern would be the fact you have to keep moving locus, I think this will get confusing during recall working out why you moved. Was it because you moved to the next set of card or because of the combination?


Hmm, not sure if I understand what you mean. There are three potential reasons for moving to another locus:

  • Pointy suit
  • New deck (but actually, nothing says that you have to start on a new locus at that stage)
  • Too many images on one particular locus (I think that Alex M. said that he had some kind of threshold there and that after linking 5-6 images to the same locus he would make a mental note and move on to the next one)

So basically, most of the time moving to a new locus is just an indication that the [Suit_2] is either diamonds or spades.

Or did I misunderstand your comment?


Ah I understand now. I was thinking that you were using a lower number of images per locus and therefor having more changes. I guess if your only doing say 1 in 6 it just isn’t that big of an issue.

My biggest issue with systems of this type (and i say this as someone who is testing one out as we speak) is justifying why I’m complicating my images when I could just use 16 different sounds (aka the Ben system) at the start as opposed to four. Both system require you to make two sounds from two playing cards, do you really think it will take you less time to form your images than it would to think of the sound for each two card combination?


16 sounds as in one sound per combination of suits, right? That’s pretty much what I’m trying to avoid. I found the process of finding 1000 words matching the consonnant patterns of the major system such a tedious task that I didn’t want to have to repeat this for an extra 1700 images.

Having the second suit as a sort of multiplier makes it easier for me. I need 676 words that match this extended major system and then the shadow images are easier to create. They just need to be related to the original image, but there aren’t any phonetical requirements.


Sylle, keep in mind that although you might use sounds to make your images, that doesn’t mean you’ll access them as sounds later. I have Viper for 8H, for example, based on a Major Method grounding. But that in no way requires me to reconstruct it from the sound basis.

Your post is giving me some ideas on how to permutate shadows based on my existing pseudo-system (I saw pseudo because many cards already have 2 or more possible figures and I don’t practice card memorization for competition purposes).

Fascinating stuff! :slight_smile:


Yes indeed, the idea is to - as much as possible - free oneself from subvocalizing.

I finished my system (well, 99% of it) a few days ago. I’m learning it at the moment. The 1-10 cards are quite easy as they are taken directly from the major system for numbers, their shadows were also quite easy to recall. J/Q/K are trickier, but I’m getting there.

At the moment I’m wasting quite a lot of time looking at the second suit to determine which locus to use next while at the same time trying to finish off the previous linking. The timing isn’t quite right. Plus I don’t associate the cards to their images quickly enough, but that’ll come!

Today I managed 3:15 for my first attempt on a complete deck with this system, which gives me a lot of hope :slight_smile:


Question to those using a two-card system with 1352 images, what is the sequence of “events” when seeing two cards:

  1. Look at the features giving the image (for me: Value_1 + Suit_1 + Value_2 + Color_2)
  2. Generate the image
  3. Identify the locus (either empty one or already featuring one or more images)
  4. Connect to locus
  5. Look at feature indicating the next step (diamonds/spades or hearts/clubs)
  6. Make a note of where to go next …

… or are you already aware of the next step before generating the image?

My timing for the question might not be fantastic as all the persons using such systems might be competing in Singapore now … or maybe not?


Little update on “my” system (more like my tweaks on existing systems), it seems to work :slight_smile:

3:15 for my first deck on the 12/12
2:00 on the 21/12
1:30 on the 11/01
1:20 on the 18/01
1:10 on the 25/01
… and 1:02 today

Like Mr Métivier said, in the end, it’s not about sounds or digit-to-letter correspondance, your brain gets rewired to associate the actual card combinations to the images. I think that having that kind of shadow images to reduce the number of “keywords” to 676 was a good way of getting started very quickly.


hi sylle.
lets say that your images for J,Q,K will be taken from your 1000 digit images(as what you have posted) then you will have 1676 images in total (1000 for your number system and 676 for your shadow images). Is that correct?
and also 9s2h 2s9h 4c3s 9d8d 5c6s
then: locus A = ( 9s2h 2s9h 4c3s)
locus B = ( 9d8d)
locus C = ( 5c6s)
am i interpreting correctly? thanks


Really cool ideas :slight_smile:

It looks like this is working for you so far, and you’re getting what you want from it. I hope that it takes you as far as you want to go.

I am an admirer of innovation, but the bane of the inventor is to experience a life-long chain of failures with brief respites sprinkled in. Trailblazers, beware.

Just want to make sure that you do know Alex has memorized digital cards in under 15 seconds, and analog in 15, and that times in that ballpark are very regular for him. In my opinion, that is as close to a conclusive demonstration that the method he is using is the best method to use that we could get. Really, we’re lucky to have such a bright star to illuminate us. Since you are in the perfect position to emulate him, I urge you to take a second look at what it is you want to get out of memorizing cards, and how you are most likely to get it.
Best :slight_smile:


Not exactly:

There are 1600 combinations following the pattern [1-10] [1-10]
Half of them (the ones with a black card in second) are taken directly from the number system and one image corresponds to two different card combinations (the one with spades in second and the one with clubs in second).
These combinations are the only ones that can be read as if they were numbers.
So that’s 400 images from the number system + 400 shadows (most of them also taken from the number system)

There are 480 combinations following the pattern [JQK] [1-10]
There are 480 combinations following the pattern [1-10] [JQK]
There are 144 combinations following the pattern [JQK] [JQK]

All these other combinations cannot be read as if they were numbers, i.e. they have associated keywords that are not part of the number system BUT most of the images used for them also come from the number system, albeit with a different keyword. For instance:

164 is TanGeRine, which I visualize as an orange.
QdKs and QdKc are O-R-A-nge. Same image but different keyword.
QdKh and QdKd are Susan from “Orange is the new black”.

So, in summary, I have exactly 1352 images in my cards system, and about 925 of them are coming from my numbers system, which means that I have in total 1352+75=1427 images.

I hope I didn’t make it more complicated now! :wink:

That part is correct, yes :slight_smile:


Thanks! As I mentioned, I got some inspiration from all the card experts who were kind enough to share information about their systems, including yourself :slight_smile:

I’m wondering if this system and Alex’s and Johannes’s (and other’s) aren’t ultimately leading to the same point, where the encoding help (i.e. the reliance on the numbers system or else) is irrelevant and where, in the end, we are all 1) seeing two cards, 2) seeing an image, 3) understanding the next linking and 4) putting it all together. In that sense, with enough practice in other to become very familiar with all cards combinations, my system should be equivalent to the aforementioned (making me not much of an innovator!). What do you think?

I’m still trying to figure out what I want out of it, so I don’t really have a good answer right now.

The point of having a limited number of keywords was to make it possible to start using this system asap after finishing it. Minimizing the learning phase, but surely limiting my progress rate whilst I’m still relying on the original-image/shadow-image associations (instead of seeing the shadow images as images in their own right).

There’s also a chance that I will less easily forget my system were I to take an extended break (not that I’ve planned any).

Thanks again for your message as well as for your sharing details about your techniques and training methods. This has been really helpful and inspiring!


Sylle wrote:
So that’s 400 images from the number system + 400 shadows (most of them also taken from the number system).

let’s say for example that the 400 images that are not initially used do not have any relationship to the first 400 images (cannot be used as shadows).
What would be the case then? thanks


It required a bit of mental gymnastics to get the non-used images to become shadows. For instance:

212 is Indiana Jones
558 is a Boulder

A boulder isn’t necessarily what you associate directly to Indiana Jones, but I chose to remember that scene where he runs away from a boulder and as a result Boulder is the shadow of Indiana Jones.
A ballerina has a Spinning Top as its shadow.

Some associations might be quite obscure to others than myself.




Reaching some kind of plateau now it seems. Maybe it was the realization that I’m quite a bit slower with a physical deck than online which made me lose motivation for a few days. Maybe I don’t know my combinations well enough…




Not the plateau part though :frowning:


I’ve been playing around with a digital deck a bit and I think one strengthens the other.

If you find the physical deck to slow, I’d suggest delving into that slowness. Enjoy the physicality of it and don’t force. Study your reactions and work on accuracy before coming back to speed.

I predict that will translate into some huge wins for you going forward in both realms. :slight_smile: