What are the botchs dices reduction available for certamen ?
Because my players will soon participate in the Normandy Tribunal Tourney, and assuming an aura of 3, the chances for a lot of magus to enter twilight during the event are very high :
3 rolls for the first turn per magus, so 6 rolls.
1+3 = 4 botchs dices.
Chance for a botch with 4 dices = 3.4%
So chances for a botch during the first turn : 1-(96.6^6) = 0.22. 22%
I know that a botch is not automatically a twilight, but it seems very high.
P.S. Sorry if my english is flawed : it is not my native language.
Well. First of all, and I guess everybody will agree on this one, a magic aura does not add botch dice to magical activities. So, unless other circumstances conspire against the duelist, it's one botch die per roll -- and thus, even if you assume that a certamen botch results in warping (but see below), no chance of twilight for most magi.
Second, and on this a number of posters will disagree with me, I'll claim that a botch in certamen does not cause warping. The book does not say so explicitly either in regard to certamen specifically or in regard to magic use in general. A number of posters extrapolate that if it's the case for spells, it must be the case for certamen too, but everybody seems to accept that it's just a subjective extrapolation (just those people who assume that vis in certamen adds botch dice to rolls admit it's a subjective extrapolation rather than an explicit rule).
So! Even if you take the "nastiest" interpretation, certamen is unlikely to produce a twilight. And a strict reading of the rules as written seems to rule out the possibility of getting even a single warping point from any number of certamen matches. But ultimately it's up to your troupe!
The section on Realm aligned spells in HoH: true lineages assumes that one does take extra botch dice from magic auras. We hashed this out on the board for a bit shortly after that book came out and, as I recall it, pretty much everyone was convinced that this particular section was in error.