I think most people, including academics, lose sight of the fact that a market is made of people, not cards. There are no stationary stochastic processes that can be milked ad infinitum, markets are not some machine in an equilibrium that can be figured out. For the sake of argument assume there was some stationary edge, then somebody will figure it it out and eventually trade it with great size. That process guarantees most such 'edges' will vary with time.
I suppose, I just don't do it that way. Certain behaviors just seem to repeat, although not always exactly the same. That's my odds calculation.
Yes, but it's still possible to broadly estimate certain reliable setups, and for this it is helpful to know the largest % you can bet without much risk of hitting your drawdown threshold over the long-term. If you are a trader, then by definition you have made an estimate or calculation of trade odds, since it's irrational to trade if the expectation is not positive.
Let's try and move things on to new ground on this question - anyone here know how to use a spreadsheet to make a basic Monte Carlo simulation?
Why a spreadsheet? I would personally use MATLAB or something like it, excel doesn't seem like a good tool for serious analysis.
I don't have matlab, whereas I do have a spreadsheet program. Also, the thing I'm trying to do is very basic (i.e. see drawdown graphs at different % bet sizes for a given trade expectation).
I don't see where that is Monte Carlo, but I think you want to do is use the 'get external data' function in the data tab to import your time and sales (or whatever you are importing). Perform your row or column operations, and whatever, however I really don't understand the details of your problem .
Ok - let's say you have a typical setup, 40% win rate, 2:1 payout. Over 1000 trades, how many iterations will hit a 20% or greater drawdown if you bet x% per trade. I.e. if you risk 0.5%, how likely is the drawdown, ditto for 1%, 2%, 3% and so on. Ditto for any given win rate and payout ratio.
Most monte carlo is done by running trading schemes over sets of real market data, but this type of calculation seems more straightforward. You know that I would argue that you can't know the 40%, but I assume it. This seems like you could just solve it using brownian motion with drift, or maybe some sort of weighted poisson distribution, but since 1000 is a pretty small number to a computer you can brute force it by running 10000 (or so?) simulations and look at the results. Exe: Set up a loop to run 1000 times with T= a random variable that can only have integer values 1 to 10. If T = 1-4 then increment the starting value by 1% (for risk = 0.5%), if T=6-10 then decrease by 0.5%. Set up a tracking variable using logical operators that tracks the absolute minimum and maximum value in the imaginary account over the 1000 trades. Nest the previous loop in another loop that will run 10,000 times and plot a distribution of the absolute account minimums achieved for each 1000 trade run. Should be fairly easy in excel. You can easily adjust the win rate and payout ratios. please post the results.
I checked excel, I don't see how to loop. I did see that the RAND function returns 0-1 values. Getting 1000 trades is pretty easy. Pick a box, say B2, and enter in the RAND function. Do a copy-paste from B2 into B3-B1002. You should see a column of random numbers. Then enter some account start value in C1, say 10,000. For C2 enter the logical function IF B2<0.4 (then) C2(1+0.01) (else) C2(1-.005). Copy and paste into C3-C1002. In D2 enter the min function MIN(C2:C1002) which gives the minimum accounct value in the C column. I changed the random column about 10 times and the lowest D2 was 9020. I do not know how to loop this 10,000 times in excel, but I'm sure there is some way.