Just before the season ends where we might see a new club winning the trophy an interview again
Continuing with the series of our Champions - before the end of season 48
Rabona folks - we have completed season 48 it is time to give you some more details on the reigning champion and 9 time winner of the trophy. Please read about Team Beerbased Lifeforms - some great insights that might help new players as well. The manager @twinner is a legend of the blockchain and one of the nicest guys in the entire ecosystem as well.
Interview with @twinner
Photo via pixabay, free license
1) What was the biggest challenge throughout your manager career?
The moment when I realized that my first team will go to retirement and I have to say goodbye. Season 1 till 18 I played manually with same team and same formation. I knewed every player and relied on them. Was not able to drop them. At the end I was relegated to league 2
2) When did you realize you can win the season?
Can't remember, but were though fights with Phoenix Incantatalem at this time
3)What was the toughest game?
4) What opponent do you fear?
My wife when she is drunken. joke
5) what was your personal highlight of the season?
No highlight when you play by bot
6) What is your favourite team member and why?
Difficult question, because I probably even don't know all team members.
7) What is your strategy to win?
First of all, it is important to find a way to calculate the value of players in order to compare different players.
In the beginning, I used Excel to check the stats of the players, a very useful tool for this is https://rte.emre.sh/ to download the current team. I then created a team setup in Excel for each formation (433, 4231, ...) and looked to see if there were players who could be used on multiple formations. I did this manually in Excel for 2 seasons, but the new scoutings as well as the stat changes at the end of the season and after each training forced me to create the table again and again.
In parallel, I then developed simple formulas to calculate an overall player value analogous to OS from the players' stats, and then to place the strongest players in the most important positions. I then compared this placement with my manual team setups from Excel, and in about 85% of the cases the placements were identical. This gave me hope that this could be automated.
An essential factor in the formula for the player value is the team player value TP in addition to the formation stats. You can also say that I have overweighted TP. So it happens that I scout new players with high OS but don't sign them simply because the TP is so bad.
Another advantage of a formula for player evaluation is that you can then also calculate a value for the scoutings, and thus compare the scoutings with the current players. The stats of old players get worse, and as soon as a scout has a better overall value, he is signed and also automatically used in the game. If an old player slips down in the ranking, his contract is no longer extended.
For each team part (goalkeeper, defence, midfield, attack) I always keep one player more than necessary, i.e. I always have a substitute for each of the 4 parts. These are then the players with the lowest total value for the formation played.
Another point is mass scouting. I scout 200 players per day, which takes about 5 minutes. Here, the players should also be provided with a mass scouting function in the UI instead of being forced to click on "Scout" 200 times.
After the scouting, my formula comes into play again, the 200 scouts are compared with the current team and the script then tells me which scoutings in which team section in which formation are better than the regular players + the one substitute player. These players are then signed.
Now we have our players, and come to the game itself.
The crucial thing at the moment is to choose the right counter-attacking formation to the formation of the opponent. Although the formation is encoded in the custom json of the game setup transaction, you can guess the chosen formation with a high probability based on the lined up players and their stats. When rondras announced that the setups would be encrypted, I had built an alternative that analyses the opponent's history and chooses a counter formation accordingly. However, since the set-up players are still visible in the custom-json, this variant has not been used so far.
In my opinion, the formation advantages should be reduced, as it makes a clear difference in the result whether one plays the same formation as the opponent or the counter formation.
Just as an example, in season 44 I always tried to play the same formation as the opponent, it ended with 3rd place.
In contrast, in season 46, I always tried to find the counter formation, and it ended with a clear lead.
Of course, it is particularly annoying for human players when the counter formation is set by a bot only seconds before the game, and then they have no chance to do anything about it. That was also the reason why I played several seasons with exactly the same formation as the opponents, that was the fairest approach for me. But since I was countered by bots myself, I activated the more aggressive counter mode again from season 45 onwards, which then led to the results.
I haven't changed my formula and code since season 25 or so, which means the bot plays with the same setup as in the last seasons of Phoenix Incantalem. Back then, the LoC's were always very close, usually decided in the last or penultimate game, so you can only guess what level Phoenix was playing at back then.
Season 46 was also the reason why I'm now trying to only play with 400 RBN rookies and see how far I get with that. It makes no sense to dominate the league permanently, human players hardly stand a chance against bots and end up turning away from the game in frustration. Since I don't renew contracts, the players are only at the club for 2,5 years on average, which means they hardly have time to improve their stats at the end of the season. Currently I only have 3 older players with more than 400 RBN, and it's already going down in the table. Yes, and that was it. I also gave the bot code to mciszczon months ago (the only thing missing in it is the formula I use to calculate the player values), he's welcome to add it to the UIs for all. Then the players could experiment with their own formulas 🙂
Thanks for the interview - much appreciated!
Rabona is a massive-multiplayer online soccer management game built on the Hive blockchain. Decide how to build and develop your team, take on the role of the head coach of your team, and manage the financials of your club. Lead your club to the top of the league, and become the champion!
If you have any questions, join our Discord server or use the comments below.
Every single reblog saves lives.