Single Comment

The 2 Guanacas in the end beat the 2 Alpacas by 56.3% in 404 games. I now
gave them an additional Pawn odds for a new 400-game run. As predicted from
the previous result, the Guanacas seem to be losing that substantially.

I stopped the 2 Knights vs 2 Guanacas match after 300 gmes, with the
Knights leading by 77.8%. Such an extremely unbalanced result cannot really
tell me anything quatitative, So it makes no sense to try to reduce the
error from 2.5% to 2%. In stead I started a match of 2 Knights vs 3
Alpacas, where I put the third Alpaca on d2,e2,d7 or e7, advancing the
corresponding Pawn by 1 square. e.g.

rnbqkbnr/pppppppp/8/8/8/4P3/PPPPAPPP/RABQKBAR w KQkq - 0 1

where A = Alpaca. This seems to go pretty even (but only 20 games so far,
so can still easily be anything between 35% and 65%).

I remember I did some divergent pieces sometime ago (in particular mNcK,
mKcN, mQcN and mNcQ), and I thought I posted the rsults at CVP. But I
don't remember where, and I cannot find it. What I recall was that mQcN
was 450 and mNcQ 750, when N=325 and Q=950. I never did mQcK and mKcQ, but
snce K and N are pretty close in value, one might expect nearly the same.
Now R is about half a Q, and W is about half a K, so it does not seem that
silly if mRcW would be about half a mQcK, which would be about equal to
mQcN = 450. That would put mRcW around 225. Now the Guanaca is very similar
to mRcW: its lacks the odd-stride distant non-captures, but as a
compensation it jumps, and so it cannot be blocked on these odd-distance
squares. So actually something around 225 would seem a quite reasonable
value for something like the Guanaco.