Hannes' system


I was wondering if anyone else have tried “Hannes’ system” (used by Johannes Mallow). Essentially, it uses 1000 images to encode 3 cards per locus, first using the cards suits , then the cards values. You can see a description here:


I saw a post by Benoit explaining his system (3 cards | 2 images) which would be similar. Is is the case? (if it is, could you tell me if you use French language to encode - Major system or Ben | Simon system)

My thought is that some people have a 3-order system for numbers, but use a PAO system for cards. Why? Is it because they are not aware of this system or it is too long to translate the cards in images? Another example would be Boris who decided to use Ben system for longer events over Hannes’ system. Just wondering (i over thinker with systems…) .

Thanks in advance

J-C Vinet

The Ben system for cards requires at least 2,652 images, which requires a lot more work than the 1,000 images required for the 3 order system for numbers. By comparison the PAO system for cards is more easily learnt - only 156 images needed).
World Champion Wang Feng has shown that despite the seemingly greater efficiency of the Ben system great results can result from simpler systems.
The Hannes system (which involves having to match up three letter pairs from two different images) seems a little more clumsy than the Ben system, but I think with practice it would be very efficient. Johannes deserves great credit for developing it - it is a brilliant way to maximise the use of 1,000 images for card memorisation.

Sorry for replying late.
My sytem looks similar to Hannes’ system. I use a major system to encode which is based on the French language but I use some English and German words too.

From the link above, it’s clear that Hannes using system of 2 images for 3 cards. However, i read around here that Hannes uses 2-card system, it means 2 cards per image. Could anyone here advise how he can do that using 1000 images only or what his method? Thanks a lot.

I use something essentially the same as Hannes’s system (I based mine on Part I of Lance Tschirhart’s Shadow System, but I believe Hannes’s is the exact same just with slightly different phonetics). It needs 1352 images though, not 1000. I can’t think of any reasonably simple way to have a 2-card system with only 1000. And I know for a fact that Hannes uses 1352 images. He might have used that “3 cards = 2 images” system a while back, but for the last few years at least he’s used a 1352-image system. But the great thing is, 1352’s really not that much more than 1000…

Lance’s Shadow System: (I used Part I and the first bit of Part II where he talks about Mallow’s system)

Some comments by Hannes himself (found here on the forums in the Speed Cards thread):
"I use a Double-Card System with consonants. But only with 1352 images instead of 2704. I was always too lazy to create a real big system so I use that trick:

Every image works for two different card combinations. So for example Kd-8c has the same image as Ks-8c. This gives you two blocks of 1352 combinations.

If I memorize cards I put images on one location and make up a story as long as combinations from block 1 appear. If a combination from block 2 appears I put it on the actual location but then move to the next one and start again with process. So I have a different number of images on each location

This gives me all information I need."

I do the exact same thing, with all my block 2 card-pairs being red-first (diamonds or heart as the first card), and all my block 1 pairs being black-first. I like it because it’s kind of like having a built-in STOP sign. You see a red-first pair, you stop there and move on to the next locus.


Thanks for the answer, AlexM. I just have 1000 images. Could you plz guide me to build up the balance, 353 images? If possible, i’d like to know your system.

Read the very link he pointed you to. Lance’s Shadow System. It shows you how to.


For those interested in making a system like this but may have been discouraged by its variable nature, here are some stats I generated a while back using a Matlab simulation. Hopefully this can mitigate any concerns you might have as to the riskiness of the system.

I’ll preface it by saying that I’m very glad I chose this system. I struggled for a while contemplating different “higher-order” systems. Obviously, different systems work for different people and none is inherently “the best,” but I would re-choose this one if I had to do it over again.

Here goes:

Multi-deck: (say, the hour cards event)
If you were to try for 25 decks 10,000 times (ie. doing 10,000 runs of hour cards with 25 decks each time), here are the stats (for average total number of loci, standard deviation, etc.) you might get.

average total loci needed: 337.6690 (~13.5 loci per deck)
standard deviation: 9.6876*
max (how many loci the longest of the 10000 runs took): 376
min ( " shortest " ): 302

So, the use of 15 loci per deck (1525 = 375) isn’t a bad idea at all. Not one of 10000 runs even made it 377. In fact, playing with the standard deviation, using 15 loci per deck (for a 25-deck run, that is) will work well more than 99% of the time (99.996% of the time to be precise).

Single-deck: (ie. speed cards)
I simulated 1,000,000 decks (ie. you did speed cards 1,000,000 times). Here’s what i got:

max of 3 images on a location: 10.84% (ie. about 11% of the time, you’ll have a maximum of 3 images on any given location)
" 4 ": 31.93%
" 5 ": 28.19%
" 6 ": 15.82% (so, about 86.92% of the time, you won’t have >6 images on any given location*)
" 7 ": 7.50%
" 8 ": 3.29%
" 9 ": 1.38%
" 10 ": 0.56%

*Something you can do also is set a cap at, say, 6 images/locus (ie. move on after 6 images even if the 6th is still black-first). As it says, this only happens once every 7-8 decks, so it’s not all that difficult to just remember by rote that you did that. Then you’ll never have to worry about having an insane amount of images on one locus.

Also, as with multi-deck, the average # of loci needed for speed cards is 13.5. However, many competitors don’t really memorize the last few images, so it really only takes about 11.

As I said, I really like this system and would love to see more people give it a shot. It’s really a great option for anyone serious about memory sports but who’s reluctant about a full 2704 system. Plus, following Lance’s guidelines, you have an easy way to upgrade to 2704 if you feel inclined…

Hope this helps :slight_smile:

1 Like

Yes, i’m re-reading the Shadow system to understand it more. It’s quite clear for me now :slight_smile: But i still want to know the balance 352 images. Could you plz not mind sharing?

Duyhoa83, I think I may be misunderstanding your question. Lance explains very clearly how to make the extra 352 images toward the end of the “Part I” section. Are you looking for something else?

I made a ‘cheat sheet’ of sorts for the extra 352 images for myself a while back:

(S)(Combination of suits)(Number of first card)(Either an R if it’s a queen, or a T/D if its a Jack)
(H,)(Suit combination)(Second card’s number)
(H,)(Face card, so H or R or T/D)(Suit combination).
(Suit combination)(First cards Number)

Each thing in parentheses is a number. (#)(#)(#)(#)

These extra ones come from second cards being jack or queen, and from kings in any position. This is because combinations of A-10 make up 800 out of 1000 images, and the other 200 are from first cards being jacks or queens.


1 Like

An example of actual use would really be helpful here. You know, a demonstration. We are in a place where a lot of visualization is required.
Ever taught chess to a beginner without a board?

It’s not the best description ever :), but here’s a video interview with me and Nelson talking through an actual example. We start to discuss the cards system around 17:30, then get into the example around 23:00. Hope that’s helpful.

1 Like

Thank you Mr.Mullen. Nice to know the World Champion has time for the common folk! Joking. Really appreciate it. Turning fifty soon and I’d like to win the Canadian Championships.
Will you be covering card memorization techniques in your video tutorial?

1 Like

What if your locus was an object (eg:sofa in a house)and you get 3 images on that locus