по торговле акциями в сессию, когда предпоследний раз понизили ставку.
Все лонги были открыты за 15-30 мин перед сообщением.
.........
my not so perfect sp system pulled out about 50K from
current 30 min sp on 1 lot
with approximately 80% winners and
average winner more than loser.
this is just about as close to the best
theoretical solution as you can get.
it is very unlikely that you can beat this strategy
if you trade on your own.
the machine is much better at making proper
decisions at certain tasks
than human does. it's funny because
we build the machines but the machine is
better than us... but we built it
so that it can do it better than we can.
so there is no contradiction there. we build trading
systems so that those can help us with what
we can't do well... basically nailing the tops and bottoms
as best as we can and making as little mistakes as
possible or correcting those mistakes the best way
possible... as simple as that, conceptually.
a machine is better at this task unless you
control the market.

the interesting thing is that given the math behind
it which is complicated but not too complicated,
one could come
up with the math solution a couple hundred years
ago because then we already had the math for it.
one could take an arbitrary curve and try to
mathematically nail those price turns as best
as one could. that's all there is to it.

given the current variables that we have access to namely
price curve, volume and that's about it, there are
theoretical limitation of what a system can do given
that we only have one or two variables which are not
the ones that we would like to have in the first place.
finding the right math for it and finding those limitations
is the solution.

the ideal math solution is when we know all variables
such as supply and demand, future profit loss expectations, etc...
it does not exist and never will in financial markets.
so there is no holy
grail in a way, but there is the best grail which is closest
to the holy grail that you can find :-)
the process behind the price discovery for example is known
but some vars are missing in real time.
we simply have no and will not have
access to those even some 100 whatever years from now...

so it is pretty hard indeed to build a great system.
it is tough task for our mind that we call
the search for the holy grail of trading.
but it is possible. i'd say given the current state
of things in trading, a system can do about 60-85%
winners with ave winner larger than loser depending
on the level of noise, volatility, trendingness and
predictabilty
in a particular price curve on a particular time frame
without the need for optimization.
just has to be done right.

i hope i shed some light on this fascinating subject.
take it as pointers without the math.