Free Friday 8 – Randomized Monte Carlo

8 thoughts on “Free Friday 8 – Randomized Monte Carlo”

1. What exactly does radomized exits mean? Exit after x days?

   Reply

2. What exactly does radomized exits mean?

   Exit after x days? Where x is a random number of days for each trade?

   Reply

   1. Yeah, exactly. It is a simple test to check if we have discovered curve fit exits or an actual edge in regards to our entry signal(s).

      Reply

3. The length of the position hold may have a material effect on the trading system though, don’t you think? As an extreme example, someone attempting a momentum trade based on fundamental changes to the stock, but then exiting the same day, will show poor returns. This sort of trade is dependent on a long term hold. Likewise, someone developing a 24 hour trade would experience poor returns if randomizing the length of position hold for too long.

   I think the key is that returns should show a smooth variation around the optimal exit delay, rather than a spike. If five days is optimal, four and six days ought be reasonable as well. If a four and six day hold kill the system, then a five day hold is likely curve-fit. the randomization of exits should be within very strict ranges that are applicable to the trade type.

   Great blog btw. I’ve added it to my reading list.

   Reply

   1. You’re absolutely right in that certain tests lend themselves better to certain systems; this is certainly where the human is still important in the automated trading process.

      Also, this is just one of the tests available at our disposal. The test you’re describing is more similar to Build Alpha’s “Delayed Entry and Exits” test. This test shows a variety of results if you were to delay either or both the entry or exit by -3 to +3 bars. This would cover/accomplish the 4-6 period test with original exit at 5 as you described, for example (and more). I really appreciate you reading and the words of praise. Thanks for taking the time!

      Reply

4. I think there is an error here:

   The first part of your entry Open(0) does not match up with the MC Analysis which say Open (3)?

   Could you double check the formula.

   Thanks

   Rick

   Reply

   1. Lol obviously we all make type-os!

      The formula in the excel table should read the same as the Build Alpha charts (which are correct)

      Close[3]>Low[9] and High[2]>Close[3] and Open[0]<=Open[2] and Open[5]<=High[7] Thanks, David

      Reply

5. That should be Close(0) and Close(3)

   Reply