This guide assumes you are already well acquainted with creating quality positions and you can correctly test position very deeply and make logical decisions about improving it.
It is well known that everyone wants their position to be as balanced and fair as only possible, yet usually, this is not achieved. Firstly, let me define the goal that one should be following when reading the guide:
In position, one should be able to successfully defend from 2v1 against two intermediate strength players (approximately 1800 each).
In position, one should always be able to clinch for 2nd place or sometimes even 1st in case a 2100 rated player targets only you and no one else targets him.
You should be able to get on top with second or first in case you are being blatantly 3v1ed by three 1600 players.
In a game of three 2400 players, you should stand a nearly equal chance to win even if your opposite resigns on move 3.
All of this subsumes your rating is 2400, and the rating of other players directly correlates to their playing strength in the variant, assuming no inflation.
Variant should be enough well-made and unique so that CGAs will accept it.
Achieving this is far not as simple as it looks like, yet let me make it clear what aspects add up to the position being notably closer to such a balance.
Most importantly, the easiest and most commonly used way of creating such a position is by using custom royals. King as a royal just does not work as well with other pieces, so creating a balanced position with the king will be way harder. Royal should be easy to checkmate for a strong player, yet at the same time enough defendable against weaker players.
The royal piece should be limited in movement: ferz, wazir, alfil, dabbaba, or pawns work already well for this. General is also a perfect choice if used correctly, it is a universal piece. In the case of a pawn, 8th to 10th rank promotion and KOTH also work very well, but if you use KOTH you should almost definitely use N-check in conjunction with it.
Next, the number of pieces. There should be a lot of pieces, about as many pawns - this is a well-tested setup that allows you to opt for rational power concentration. Sergeants, berolinas, and stone generals are the best pawns you can think of usually. Also, use all the 14*14 board: do not turn it into 12*12 or 13*13, do not minify it. Walls like in the Battle of the Castles are OK.
To arrange the pieces, you will need a concept. Here you are not limited by the concept itself, but power concentration should be adequate. Take a look at the Battle of Castles, Laboratory of Knights-Sergeants, and Unison to see what correlation of pawns and pieces should approximately be.
When you are done with the overall placement of the wall structure, and overall piece type proportions, you should finally begin to fill out the pieces. You should take the following points into account:
The position should be very complicated and very variable. Complicatedness can help to combat problems of balance by a whole lot, so opt for variants that are overall slightly less or more complex than standard.
Limit 2v1 attacks that win material to only one line. Why only one line instead of zero, you may ask? Such a line should grant you positional advantage via losing some material so that you can knock out a weak player very easily.
The game should be slow-paced: the more moves to prepare and the more quiet position is, the better. Players tend to learn positional and slow-paced games at a way noticeably slower than fast-paced games. Yet, your position must remain sharp.
Point spread like OxN or N-point checkmate is an outstanding way to compensate for weak flanks/opposites, so point spread while not being a must is a very good idea usually.
Now, pick a piece set that will correspond to all of the above. You will fail a lot with the piece set, usually due to variability or overall lack of strategy. Do not worry, this is normal if you aim for the inordinately balanced position, so 4-5 tries should do it (I found optimal piece set on the second try due to myself being lucky here, you will likely need more).
Another thing that you should note in the final piece set proportions is that there should be three well-defined piece categories. Having a piece that cannot influence KOTH or threaten royal is a good idea. This piece should be strong by itself, yet have its notable weaknesses.
When the preparation is done, the most important part begins testing. That's where you will have ↈ possible setups that all seem to work, so you will need to filter them out. Here is a good way that I observed from creations of laboratory and from my own experience:
Three-move checkmates should be present but be incredibly restricted that any of three players no matter his color can easily stop them. This is a good way to strike down 3v1 potential.
Subdivide positions into two flanks, either from start or allow players to opt for it via different promotion options (fast promotion). Two separate flanks and royal being a weak piece is a good way to let each player define what is the attacking flank and what is the defending flank (not like in Collateral Worlds, something like that will only break balance)
Checks should not gain compensation, and the only checking pieces should be important in middlegame and opening. Laboratory of Knights-Sergeants and Unison perfectly demonstrate this aspect.
Calculate 2v1 attacks up to depth 17(!) so that there will be no material gains without resulting in worse positional weaknesses. This will take a lot of time to get right, yet also this allows you to filter out wrong positions. Take a look at the archive of qilp while he created and tested T, a very good example for position creators on how to correctly build up a position.
The next step is TEST, TEST, AND TEST. Yes, now when you are down to 3-5 position setups, test all of them and ask for opinions. They all will be likely very close in terms of balance, yet you want to land a perfect shot. Also, it is normal for a position to be pretty boring: beginner-friendliness and balance will compensate for it a lot.
One interesting aspect is that balanced and complex positions tend to be more beginner-friendly, so you will carry out on two aspects instead of one. IMO, but trading balance and friendliness in an exchange for excitement is a very good choice.
As an additional idea, test your position with diplomacy. If it is balanced in diplomacy enough to be accepted were it not for the prohibition of admins, then it is enough balanced to be played without diplomacy for sure, and you'll end up with an inordinately balanced position.
Now about the opposite, ideally they should not be your friends. Instead, they should be the ones you try to weaken to gain an advantage, this brings a good amount of balance to the position.
While all of this does not contribute to balance directly, these aspects influence balance in a different way and allow your position to be extremely balanced. Each of these aspects is compensation for an element of imbalance and luck present in each of the normal FFA positions (cough cough dead king labyrinth cough), and contributes to the overall balance.
You may also like to make your position solo by default if you managed to succeed in creating a very balanced FFA position. It might happen one will follow all of these guidelines randomly, yet this happens extremely rarely and you should instead focus on the aspects themselves.
And of course, in the process of this, make sure that the position is original and complicated enough, and not having issues so it will not be declined, but this applies to every position reviewed by CGAs.
While creating a very balanced position is not that difficult, it does require a lot of power investment and a good amount of determination to correctly implement it. Even weak players may create such a position - you will need to ask for opinions a lot and test with strong players. Again, diplomacy in testing helps a lot with this: you see problems of 2v1 and 3v1 far away.