Welcome back #splinterlands fans! Star💫 here (
Cover art in honor of 
OUR CURRENT SYSTEM
Up until brawl cycle 98, SoG has used a brawl assignment system similar in style to the system used by many guilds today. At a high level, the system rewards top brawl performers by letting them keep their spot for the next cycle, and punishes bottom performers by taking away their spot (otherwise referred to as "benching") for the next brawl. In the case of a benching, another member in the guild takes the benched player's spot and tries to hold onto it by performing well in the next brawl.
Under our current system, we typically allowed to the TOP 5 performers to hold their spot, whereas the BOTTOM 5 were benched. Over the course of time, exceptions were made to the rules based on specific situations that felt unfair, or which just didn't make sense strategically.
As an example of the former (i.e., felt unfair), at some point we instituted an exception to the BOTTOM 5 rule so that, if your winrate was more than 65%, you cannot be benched---even if you end up in the BOTTOM 5. This situation arose when everyone in the guild performed well in the same brawl, and even good performers were being benched simply by default.
As an example of the latter (i.e., didn't make strategic sense), at another point we realized that in order to compete in our gold foil frays, we needed to keep the gold foil fray assignments relatively stable. This is simply because--even in a guild like ours, where some players have been playing for a long time---we don't have a deep enough bench of players with fully stacked gold foil decks to be benching from those frays. Therefore, another exception under the current system is that gold foil brawlers cannot be benched, regardless of their performance. Luckily for us, the performance of our gold foil brawlers has rarely if ever been an issue 💪
PROBLEMS WITH OUR CURRENT SYSTEM
In principle, the current system is setup to reward the best performing brawlers and punish the worst performers. The underlying principle behind the system is therefore sound. How have the results been though? Well, the results haven't actually been that reliable. In other words, sometimes we dominate, but other times we significantly underperform.
One problem we were concerned with was how the remaining open fray assignments were being filled. Obviously the TOP 5 and BOTTOM 5 were easy. They either kept their spot, or warmed the bench, respectively. What about everybody else though? We are a 30 member guild with only 16 fray assignments (for now, until we move up tiers). Under the current system, we simply allowed the remaining spots to be filled on a first-come, first-serve basis. A sort of free-for-all, depending solely on who was quick enough to sign up. As you might imagine, this is where our consistency problems seem to have been coming from.
For example, imagine the situation of hypothetical "Player X." Player X plays mostly modern, and has a chaos legion heavy deck that is light in the older editions. The frays open when it's 2:00am in Player X's country. When they wake up and go to check for open frays, the only ones left are Frays 7 and 8 (silver/regular foil/wild). Player X is all about brawls, wants to participate, and wants to help the guild. All good, admirable attributes in a guild-member. So they join Fray 7 and hope for the best. How does it go? Long story short, usually it does not go well.
It is because of situations like this that we decided to take a step back and analyze our fray-specific winrates.
FRAY-SPECIFIC WINRATES
So we took a close hard look at the fray specific winrates for each member of our guild from the last 35 brawls. Interestingly, some very clear patterns arose right away. Members fell into three general categories referred to here as SPECIALISTS, UTILITY PLAYERS, and JEKYLLS.
SPECIALISTS
The first category are fray-specific specialists. These members tend to dominate the same one or two fray spots week-in and week-out, thereby keeping them for each subsequent brawl. In some cases, this is due to deck constraints (e.g., 
UTILITY PLAYERS
The next category are referred to as "utility players." These members tended to play in a wide variety of frays---generally, in at least four to seven different frays. However, what really distinguishes the utility players as group was their tendency to win in whichever fray they played (e.g., 
JEKYLLS
Finally, we have what are referred to here as our "jekylls" 😂 Yes, as in dr. jekyll and mr. hyde. What we're referring to are guild members that---like the utility players---play in a variety of frays, but---unlike the utility players---do not perform consistently well everywhere. Instead, jekylls perform very well in some frays, while significantly underperforming in others.
Simply as an example, take 
COMPARING THE DIFFERENT SYSTEMS
To test out the new system, we identified the top performers in winrate for each fray, and then specifically assigned them to those frays. Kind of like setting a batting order in baseball, or the starting lineup in a football match. This new approach contrasts with our current system in which most frays were taken in a first-come, first-serve free for all, without strategic thought as to which members were best for which fray.
For this experiment, two of the identified 'top-performers' in their frays (
Yes, in fact, it seems like it did. Not only did we turn in a blow-out first place win under the new system, but we increased the percentage of players with a winrate of >75% from 25% to 63% in just one cycle 👀 Moreover, the replacements for
Btw that winrate (>75%) is an important threshold for us. It is where we believe top tier brawlers should be performing.
Of course, the data here in this experiment is very limited---we are only comparing two brawls. It's very possible that other factors contributed to the difference in outcomes. Further testing and refinement of the system is no doubt needed. However, it is hard to argue that this wasn't a substantial turnaround. One we needed too. Getting us out of a slump and an unusually bad, four-brawl losing streak.
CONCLUSION
That's it for now folks. Please keep an eye out for our next special END OF SEASON edition (co-authored with 