Frosty Strategy Development

Here's a thought, given the caveat that I am NOT an ATS trader and I have NO experience with ATS programming, although I was/am a professional commercial quality software developer/architect.

Rather than starting out being in a position at all times, why not start out programming 1 maybe 2, specific strategy entry/exit pairs. Perfect this/these entry/exit pairs before adding a reverse situation and/or other entry/exit. It may even be determined the current incarnation of style and strategy are not well suited to being in the market at all times. Which would provide a different, yet completely viable set of options, for ongoing development.

All the best
Osorico :)
 
osorico,

That is basically what I am trying to do. Previously when I developed strategies I always focused on both long and short at the same time, entry and exit at the same time.... I think trying to fit it all together at once was the wrong way to go..

This time, I am focusing only on entry, and only for long trades first. So right now i'm using a time exit to try and perfect the entries until they have a true edge over those random entry results I posted previously.

Once I have an entry that has a true edge, I will then work on perfecting an exit system BASED on how that entry works, so that I can extract the optimal number of points out of each trade.

I feel this way of strategy design is a lot better than the way I used to do it.
 
Quote from JimmyJam:

TSGG is correct.

To tell you the truth, I never knew trading could be so difficult.

Never.

JJ
More...

Stay with it long enough without suffering massive psychological damage and it becomes easy. I'm more than a decade into the subject and it is looking easy, I have a neighbor that draws some lines on charts and trades only when he needs some money. The guy has a nice paid for house, 'Vette in the garage, he's not greedy and he is not worried and he is not stressed at all.

Back on topic, Frosty, some have stated, and I agree, that 1 minute bars are too short to have meaning. You might want to set a higher minimum for testing and explore faster bar intervals after you get a system in place.

One widely practiced approach to a very complex problem is to break it into pieces and solve them individually, there is little doubt that will help.
 
Quote from frostengine:

osorico,

That is basically what I am trying to do. Previously when I developed strategies I always focused on both long and short at the same time, entry and exit at the same time.... I think trying to fit it all together at once was the wrong way to go..

This time, I am focusing only on entry, and only for long trades first. So right now i'm using a time exit to try and perfect the entries until they have a true edge over those random entry results I posted previously.

Once I have an entry that has a true edge, I will then work on perfecting an exit system BASED on how that entry works, so that I can extract the optimal number of points out of each trade.

I feel this way of strategy design is a lot better than the way I used to do it.
More...

I think this level of granularity is the right approach. And it's within the realm of the "test everything" mantra professed by some of those knowledgeable with ATS development. The OOPness of such granularity should also be a boon.

Osorico :)
 
This is a lot harder than I thought it would be... or you can just say i'm pretty dissapointed with most of the entry strategies I have been testing. Here is the results from the best long entry strategy I've gotton thus far.

Exit After 5 Bars:
Trades 574 PL +2 pts, Win % 37

Exit After 10 Bars:
Trades 560, PL +20 pts, Win% 39

Exit After 20 Bars:
Trades 550, PL -20 pts, Win % 37

Not sure what to think about this. The win% is fairly low compared to alot of the random strategies, but the fact it was profitable over this time period WITH commissions and slippage included tells me it may have a slight edge. I'm guessing that although it may not be a good entry for a high % win system, it appears it must be a good entry for picking when the ES will rise substantially over the next 10 bars. At least enough to overcome slippage/commissions.

More than anything it tells me I got a lot more work to do before I have an entry strategy that I feel has a very strong edge.

Any thoughts? Looking at the random strategy results and then this one, what suggestions do you have? Would it be wise to assume that this entry may have an edge, however slight? Or is it not enough to rule out random chance at this point?
 
Right after posting this I actually tried looking harder at the random results and realized that it would probably make more sense to compare these results with random if the # of trades taken matched closer.

So I reran the random trading with a less % chance of entering a trade. Here are the results:

Exit After 10 bars:
Trades 784 PL -39 Win% 32
Trades 774 PL -9 Win % 33
Trades 808 PL -26 Win % 34
Trades 726 PL -3 Win % 33
Trades 682 PL -10 Win % 34
Trades 708 PL +1 Win % 35
Trades 656 PL +16 Win %35
Trades 644 PL +10 Win %36
Trades 632 PL +19 Win %36
Trades 632 PL -2 Win % 35
Trades 624 PL +15 Win %36
Trades 596 PL +12 Win % 37
Trades 594 PL +18 Win % 38
Trades 590 Pl +9 Win %38


These results makes my entry strategy not look as good. Shows that my results more than likely are nothing more than just random chance........ So a lot more work to do. Which is the way it goes. Hopefully by time i'm done with this thread I will have refined my strategy development skills greatly and walk away with a good strategy.
 
i was reviewing the thread and was trying to determine the rationale behind testing the entry(edge) w/ timed exits. it's an interesting concept to verify the signal. maybe i'll try it. is there any reason to trust this result, or is it more of a logical conclusion we thought up?

:confused:
 
yes.. it would seem logical. i agree.
another concept that at least makes sense to me..
test w/ a hard stop + hard profit target.
hypothetical-

200$ stop
400$ target

if it then hits 70% winners over 800 trades..
well, i'd be encouraged.
the example might be extreme, but you see the logic i hope.
(if not, i'm back to the drawing board,lol)
 
You must log in or register to reply here.
Back
Top