How long wold this take?

[Mr Subliminal]

While I do dabble around with VBA for Excel, for the more serious work I use Turbo Pascal 6 for DOS. Old and outdated, but still lightning fast. In order to give you my estimate of 30 seconds, I ran the following (pseudo code) in Turbo Pascal - it took 33 seconds.

do 100,000,000 times
generate random number
if less than 0.5, trade is a winner
if greater than or equal to 0.5, trade is a loser
keep running total of winners and losers
end loop

 

[ElectricSavant]

I sent you a PM.

Michael B.

P.S. Jake Bernstein, Eat your heart out!

Quote from Mr Subliminal:

While I do dabble around with VBA for Excel, for the more serious work I use Turbo Pascal 6 for DOS. Old and outdated, but still lightning fast. In order to give you my estimate of 30 seconds, I ran the following (pseudo code) in Turbo Pascal - it took 33 seconds.

do 100,000,000 times
generate random number
if less than 0.5, trade is a winner
if greater than or equal to 0.5, trade is a loser
keep running total of winners and losers
end loop

 

[damir00]

Quote from ElectricSavant:

Do you understand what I am asking?....we are getting into the millions of backtests.

i think you underestimate what some folks do here on a routine basis. anybody doing montecarlos is probably running through a million (simulated) market years a run.

so yes, it can be done, and yes, it would be quicker than making a pot of tea.

 

[ElectricSavant]

You mean I am not the first one to think about this?...darn!

Michael B.

Quote from damir00:

i think you underestimate what some folks do here on a routine basis. anybody doing montecarlos is probably running through a million (simulated) market years a run.

so yes, it can be done, and yes, it would be quicker than making a pot of tea.

 

[G_Morgan]

Isn't it easier just to track the average change from the open at different times in the day? That gives you a better visual, and you get more definition in the series (can do it every .01 seconds with no problem)

 

[ElectricSavant]

ROC, Velocity and momentum indicators could be matched I suppose. But might be hard to visualize, unless you did a "NihabaAshi" and printed out charts and indicators.

Michael B.

Quote from G_Morgan:

Isn't it easier just to track the average change from the open at different times in the day? That gives you a better visual, and you get more definition in the series (can do it every .01 seconds with no problem)

 

[axeman]

This stuff is trivial to do on a SINGLE instrument like ES
when you can load it into memory ONCE and crunch the
hell out of it.

Now try doing that with 1 min candles for 18,000 stocks
over a 5 year period

Whole different ball game when you cant fit it ALL into
memory at once.


peace

axeman