Entry, all in (due to risk control), exits, scaleouts letting winners run.
Scale in/out vs. all-in/all-out
- Thread starter Overnight
- Start date
Thanks Overnight,
I manually trail like this. I enter, then wait for 20 ticks. then move stop in favor by 10 ticks from entry. Once market move another 10 ticks in favor I move another 10 ticks. I do this til break even.
Out of about 40 trades, the most I got was 50 ticks.
I honestly just guess to do it that way.
Is there an logical or someway I can test the optimal trailing distance for my system? I do forward testing for now.
No that sucks, if the area is real good, it could leave with a tiny position, yet if it fails miserably, you are all in, how in the world is this good money management?
Like, I said, it's just massaging it in. It won't leave with a tiny position, you're going to be all-in regardless. If you see it speeding up, you finish out the entry quicker. Either way, you're all-in, but your entry may be a little better (or, yes, worse), if your timing is off. But I think this allows for some flexibility.
Consider the "scale-in/entry" time in proportion to the overall timing of the intended trade; instead of going all in instantly, perhaps you'll scale in over the next hour if that suits a trade that will span a day or a week. And this reduces your initial exposure while allowing you to wait out if the signal stays. The tradeoff is obviously the difference in profits you would've made from an all-in position (if you were late on an entry), but I think the more granularity you can add to your decision-making, the better, since it is more flexible. More complicated, but more flexible. And obviously if you were early on your signal and it pans out, you've improved your entry.
Also, if you're advocating all-in, won't that loss be worse than scaling in, if we consider the first signal is where one would've gone all-in?
Consider the "scale-in/entry" time in proportion to the overall timing of the intended trade; instead of going all in instantly, perhaps you'll scale in over the next hour if that suits a trade that will span a day or a week. And this reduces your initial exposure while allowing you to wait out if the signal stays. The tradeoff is obviously the difference in profits you would've made from an all-in position (if you were late on an entry), but I think the more granularity you can add to your decision-making, the better, since it is more flexible. More complicated, but more flexible. And obviously if you were early on your signal and it pans out, you've improved your entry.
Also, if you're advocating all-in, won't that loss be worse than scaling in, if we consider the first signal is where one would've gone all-in?
Last edited:
Your massaging wont work, only sounds good on paper, you know how most good areas fail? They react *you are filled 100%, but now with even bigger stop*, then they fail.
How most good areas fail? How do the other good areas fail? How much is most?
Point being, you can construct a worst-case scenario for every setup, but that isn't constructive until taking all possibilities and their probability relative to each other, no? Problem becomes Bayesian, in a way.
I'd assume that, all else equal, more flexibility is better.
I imagine your trading platform has a way where you could implement an ATM and run it through replay data. NinjaT has it and it works pretty well. The only problem with that is related to the exact same reason that you mentioned above. Forward testing is the only true test of a system, but it takes so much time. And the day you may decide to implement the system, the market conditions could have changed to the point where your system no longer works in the current market condition. But all-in-all, I figure that if you find a system that works 80% of the time in sim with back testing, it should work that much time in forward testing, given enough backtesting timeframe. Say, a couple of years.
Thanks for your response.
That's a good idea to use Ninja Trader through market replay daily and their ATM strategy.
I will give this some thoughts, cause forward testing , but ultimately I eventually need a simple way to back test my ideas. not sure where to start, but I will look.
I kind of agree with Both CyJack and FON. I use 0.75% risk on each trade at initial entry. Most of the time (~80%) that is my only entry until the trade is profitable. If the trade proves profitable and my trailing stop is NOT hit yet there is a retracement and I have another entry trigger I'll add 0.25% to that. That is my max. That's if the trade is profitable. If it's NOT profitable I take initial 0.75% risk then if price drops a little lower but still above my initial stop loss AND I get another entry signal I allow another 0.25% for max of 1% risk. The benefit of this is my entry signal continues if the trade is going my direction. As the trade goes against me for a small amount (but still above my stop) I no longer have my entry signal (we will call this a neutral signal) and I ride it out until I either get a sell signal (in which case I sell even before my stop is hit) or my stop is hit and lose my initial 0.75%. If the neutral signal now becomes another long entry signal before the stop is hit I add the 0.25% but now at a lower price. Since my entry is lower than initial entry I averaged down. Normally I do not add to losers except in this one and only scenario. This may happen 10% of the time. In this scenario I could lose a max of 1% all in so FON is right but that rarely happens to me. Usually I get my trade to move in the intended direction and now I have more profit to show for it. This is especially true because I have a tighter stop on the 0.25% 2nd entry since I can put more money to work on the 2nd entry trade (my stop loss is the same for both 1st & 2nd entries). Most trades happen with only one (0.75%) entry for me. So in this case I am all in at the beginning.