Terrablood's Newsletters explanation

The online newsletters uses a database of warriors to search for styles. These styles come from three sources:

1- Players enter warrior styles directly into the database.
2- The Top Ten Warriors of each style from the newsletter.
3- Terrablood's style calculation algorithm. This is automatically run when there is a new turn or data for an arena.

The style calculation uses numerous sources to narrow down the possible styles for each warrior:

1- The Top 11 fighting styles from the newsletter.
2- The Most Popular list from the newsletter.
3- The Win/Lose list from the newsletter.
4- The Last Week's fights from the newsletter.
5- Deduction of warriors that qualified as a top warrior of a style but were not. For intstance, if the Duelmaster is the top LU, and the second active warrior with an unknown style is adept or above with a winning record, then that warrior must also be a LU.

How it works:

1- Known warrior styles in the top 11 fighters are subtracted from the Top 11 list.
2- Known active warrior styles are subtracted from the Most Popular list and they are looked up in the Last Week's Fights to see if they won or lost, then subtracted from the Win/Lose list.
3- As the algorithm progresses down the warrior list in the newsletter, it does a union of the lists to deduce what they could possibly be. Only the lists that qualify are used. For instance, the Top 11 list would not be used if the warrior was not in the Top 11. Those lists with a 0 for a style cannot be a style candidate. For instance, If 4 PLs fought this turn, and we already know who the 4 PLs are, both the Popularity list and Win/Lose list for PLs would have been reduced to 0, so none of the unknown warriors could possibly be a PL.
4- If the warrior has a multiple style calculation already, there is a union of that calculation with the new one. For instance, last turn the style calculation of warrior X was ST-LU-TP-WS and this turn it is LU-TP-SL. The union of these two would leave us with LU-TP as the remaining possibilities.

A simple example:

Assume an arena with no known PS styles. If the Popularity list shows 2 PS fought this turn with neither in the Top 11, and their win/lose numbers were 0 - 2, then we know only those warriors after the Top 11 that lost this turn could possibly be the PS style.

Issues:

1- If there is more than one warrior with the same exact name, the Last Week's Fight info cannot be used for that particular warrior. Everything else is still calculated.
2- Though rare, if a player enters a warrior incorrectly, such as the Id or style, that will throw off the style calculation algorithm. However, if problems occur, the algorithm is able to catch and report them.

Misc:

1- If you have your own database of warriors, you can send the output to me and I will merge them with the database. Excel, space, comma, or tab-delimited are all acceptable.

1-	Players enter warrior styles directly into the database.
2-	The Top Ten Warriors of each style from the newsletter.
3-	Terrablood's style calculation algorithm. This is automatically run when there is a new turn or data for an arena.

1-	The Top 11 fighting styles from the newsletter.
2-	The Most Popular list from the newsletter.
3-	The Win/Lose list from the newsletter.
4-	The Last Week's fights from the newsletter.
5-	Deduction of warriors that qualified as a top warrior of a style but were not. For intstance, if the Duelmaster is the top LU, and the second active warrior with an unknown style is adept or above with a winning record, then that warrior must also be a LU.

1-	Known warrior styles in the top 11 fighters are subtracted from the Top 11 list.
2-	Known active warrior styles are subtracted from the Most Popular list and they are looked up in the Last Week's Fights to see if they won or lost, then subtracted from the Win/Lose list.
3-	As the algorithm progresses down the warrior list in the newsletter, it does a union of the lists to deduce what they could possibly be. Only the lists that qualify are used. For instance, the Top 11 list would not be used if the warrior was not in the Top 11. Those lists with a 0 for a style cannot be a style candidate. For instance, If 4 PLs fought this turn, and we already know who the 4 PLs are, both the Popularity list and Win/Lose list for PLs would have been reduced to 0, so none of the unknown warriors could possibly be a PL.
4-	If the warrior has a multiple style calculation already, there is a union of that calculation with the new one. For instance, last turn the style calculation of warrior X was ST-LU-TP-WS and this turn it is LU-TP-SL. The union of these two would leave us with LU-TP as the remaining possibilities.

1-	If there is more than one warrior with the same exact name, the Last Week's Fight info cannot be used for that particular warrior. Everything else is still calculated.
2-	Though rare, if a player enters a warrior incorrectly, such as the Id or style, that will throw off the style calculation algorithm. However, if problems occur, the algorithm is able to catch and report them.