"Scaling out" is inferior behavior

[Thunderdog]

Quote from GTS:

1) By scaling out you are basically breaking your strategy into two different strategies and taking the average of them. Since one of the strategies over the long term must be superior to the other, you are by definition ending up with a sub-optimal strategy (the average of the two rather than the greater of the two).

2) If you do not know whether your intermediate target or final target is the optimal point then it would benefit it you to do some analysis and figure it out and use that going forward exclusively.

3) Anyone who has done mechanical system back-testing with and without scaling out already knows that what B1S2 is saying is true – the numbers don’t lie.

4)...No need to be defensive about it...

1) It may be sub-optimal after the fact, but a priori? Perhaps I should consider my scaling out as a form of diversification within a single position. Diversification is, by definition, a sub-optimal strategy if you know full well in advance where specific markets are going to go. Few of us do. I am not among them. Therefore, a priori, I consider scaling out as prudent, and therefore optimal in an environment of uncertainty.

2) No amount of analysis of historical data will allow you to see the future with clarity. You either don't know, or you don't know that you don't know. Have you looked at any of the do-or-die top calling threads lately? At best, your testing will give you a hazy indication, which is hardly the stuff from which bold predictions are validly made. When you are operating in an environment of uncertainty, it pays to recognize that fact.

3) See No. 2.

4) Was I being defensive? I thought I was merely asserting my position on this matter. By extension, are you suggesting that B1S2 was implicitly "defensive" about exiting all at once when he started this thread?

 

[illiquid]

Quote from GTS:

I would concede that perhaps this may not apply to purely discretionary traders but is there really such a beast? How does a discretionary trader decide when that intermediate point is reached? In other words, everyone is following rules even if they aren’t written down or not easily articulated.

Anyone who has done mechanical system back-testing with and without scaling out already knows that what B1S2 is saying is true – the numbers don’t lie.

Lastly, sub-optimal isn’t a dirty word. If it makes you happy or you feel more comfortable locking in partial gains so that more of your trades are winners rather than having a lot of losers and a few big winners then scaling out is for you. No need to be defensive about it – a lot of trading is psychology anyway.

Nice post GTS.

I would just add that, although I agree that in general "numbers don't lie", backtests and average amounts are always moving with the next trade, and CAN lie, or at least prove deceptive.

The big mistake buy1sell2 makes is that he assumes scaling out will always prematurely exit a position, versus a full exit. This is true only in hindsight, comparing the scaled exit to an "optimal" exit. But in real-time, in real trades, scaling will sometimes keep you in the trade longer than you normally would, which is something Thunderdog alluded to. Scaling out is always inferior to the "optimal" figure, but the optimal is just a "theoretical" number -- this doesn't hold in real-time.

Say, based on backtested figures, you've found that an "optimal" target will net you 3 points on average for method X, based on a past series of trades; over the same time frame, scaling out only nets you 2. However, let's say for the next Y number of trades, the ranges widen quite a bit for method X -- scaling out leaves you in the trade longer, and therefore for those trades you've netted an average of 6, while your "optimal" exit yields just 3.5 for the new series.

Now, if you backtest with the new information you receive with the second series of trades, you will find that a new "optimal" target will net you 4.5 points, while scaling overall yields 4. The difference here is this: the 4 points on avg for scaling is an actual figure that you would have received for all trades; the 4.5 points for the "optimal" target is just a theoretical figure which has been adjusted for the new series -- you still only get 3.5 for using the "optimal" target from the first series. Optimal is only optimal in hindsight, and comes down to how quickly you can adapt that figure for incoming trades. It's quite possible that scaling out will yield a greater profit overall -- at least a profit more "reflective" of current conditions -- while an optimal figure can move quite slowly, depending on how many trades are used as history/how fast conditions have shifted.

edit: the converse example for a deteriorating method "X" would probably be more realistic and to the point -- that is, a method whose optimal target is progressively smaller. If method "X" began as a very high yield setup, say given for a high volatility market, but deteriorates as volatility contracts, you would see a far greater "real-time" difference in results between a scaled exit versus an "optimal" exit -- meaning, the prior higher "optimal" exit might yield 0 or worse, as opposed to an "updated" optimal figure.

 

[Thunderdog]

Excellent post, illiquid.

 

[CaptainObvious]

Quote from GTS:[/

Lastly, sub-optimal isn’t a dirty word. If it makes you happy or you feel more comfortable locking in partial gains so that more of your trades are winners rather than having a lot of losers and a few big winners then scaling out is for you. No need to be defensive about it – a lot of trading is psychology anyway. [/B]

I would only argue that trading is not just " a lot about psychology", it's 100% about psychology. Finding ones own personal comfort level, which is always a work in progress, is what will keep you in the game over the long term. The bigger the swings in your account, the greater the stress, which will ultimitely lead to burnout. Game over at that point, regardless of profitability.

 

[illiquid]

Thunder, you beat me to it.

There's always a possibility of a wide gulf between real-time vs theoretically "optimal" and "correct", etc. It's like asking a chartist to look at a chart, and tell you what he (obviously) "would have" done, then trying to replicate the results.

But really it comes down to how much faith one has in the past in the first place, and how the past is used in one's trading. "Backtests" are inferior criteria to trade off of -- now that would be an interesting thread.

 

[AAAintheBeltway]

Quote from illiquid:

Nice post GTS.

I would just add that, although I agree that in general "numbers don't lie", backtests and average amounts are always moving with the next trade, and CAN lie, or at least prove deceptive.

The big mistake buy1sell2 makes is that he assumes scaling out will always prematurely exit a position, versus a full exit. This is true only in hindsight, comparing the scaled exit to an "optimal" exit. But in real-time, in real trades, scaling will sometimes keep you in the trade longer than you normally would, which is something Thunderdog alluded to. Scaling out is always inferior to the "optimal" figure, but the optimal is just a "theoretical" number -- this doesn't hold in real-time.

Say, based on backtested figures, you've found that an "optimal" target will net you 3 points on average for method X, based on a past series of trades; over the same time frame, scaling out only nets you 2. However, let's say for the next Y number of trades, the ranges widen quite a bit for method X -- scaling out leaves you in the trade longer, and therefore for those trades you've netted an average of 6, while your "optimal" exit yields just 3.5 for the new series.

Now, if you backtest with the new information you receive with the second series of trades, you will find that a new "optimal" target will net you 4.5 points, while scaling overall yields 4. The difference here is this: the 4 points on avg for scaling is an actual figure that you would have received for all trades; the 4.5 points for the "optimal" target is just a theoretical figure which has been adjusted for the new series -- you still only get 3.5 for using the "optimal" target from the first series. Optimal is only optimal in hindsight, and comes down to how quickly you can adapt that figure for incoming trades. It's quite possible that scaling out will yield a greater profit overall -- at least a profit more "reflective" of current conditions -- while an optimal figure can move quite slowly, depending on how many trades are used as history/how fast conditions have shifted.

edit: the converse example for a deteriorating method "X" would probably be more realistic and to the point -- that is, a method whose optimal target is progressively smaller. If method "X" began as a very high yield setup, say given for a high volatility market, but deteriorates as volatility contracts, you would see a far greater "real-time" difference in results between a scaled exit versus an "optimal" exit -- meaning, the prior higher "optimal" exit might yield 0 or worse, as opposed to an "updated" optimal figure.

All true, but in fairness that really wasn't B1S2's argument. He said that of all your trades, some percentage would be huge winners and that your long term P/L depends on getting the most out of them. That means getting the most out of them with your full load, not a scaled back position. I am sympathetic to all the arguments in favor of scaling out, but I have to say that all my backtesting confirms exactly what he said. Your optimal strategy over time will always be to have an exit strategy that ensures you catch those big moves with a full boat.

A corollary of your point, and it's a perfectly valid one, is that what is a "big move" may vary over time. I think the solution to that issue is not to try to optimize targets, but rather to use an exit methodology that gives you every shot at catching the big move.

Marty Schwatz' great book Pit Bull talks about how important itis to get the big moves. He said the majority of his profits came on a handful of days each year. I realize that is totally contrary to the way most daytraders have been taught to trade, ie be consistent, take quick profits, try for a high win percentage, but it sure worked for Marty.

 

[Buy1Sell2]

Quote from AAAintheBeltway:

All true, but in fairness that really wasn't B1S2's argument. He said that of all your trades, some percentage would be huge winners and that your long term P/L depends on getting the most out of them. That means getting the most out of them with your full load, not a scaled back position.

Marty Schwatz' great book Pit Bull talks about how important itis to get the big moves. He said the majority of his profits came on a handful of days each year. I realize that is totally contrary to the way most daytraders have been taught to trade, ie be consistent, take quick profits, try for a high win percentage, but it sure worked for Marty.

I was hoping someone would point out that I was not talking about being able to catch the big move on every single trade. Someone earlier, I think it was GTS also alluded to that and thanks to you both for pointing it out. It's the overall trading that is important, not the individual trades by themselves. With regard to the Pit Bull paraphrasing, I would say this--it worked for Marty and would work for every other trader as well. It is the correct way to trade. Scaling out is deficient, inferior behavior on any time frame and should be avoided.

 

[illiquid]

Quote from AAAintheBeltway:

All true, but in fairness that really wasn't B1S2's argument. He said that of all your trades, some percentage would be huge winners and that your long term P/L depends on getting the most out of them. That means getting the most out of them with your full load, not a scaled back position. I am sympathetic to all the arguments in favor of scaling out, but I have to say that all my backtesting confirms exactly what he said. Your optimal strategy over time will always be to have an exit strategy that ensures you catch those big moves with a full boat.

I understand his point of view, and it's a natural one coming from a position trader who disdains intraday trading. By definition, he will always be going for the home run, as opposed to daily singles and doubles. But to make a blanket statement that scaling is inferior is a pretty dubious one. If you were a position trader in the dollar for the past six months, you'd be tearing your hair out from the tight range and constant pullbacks we've had compared to the first quarter of the year. And if I came along and said "position trading is a waste of capital, you miss out on all the inbetween back-and-forth -- you need to maximize your returns by trading the intraday moves as well", that doesn't mean anything at all. Scaling out also doesn't necessarily mean timidly selling partials just to get comfortable, it can also be a part of an extremely aggressive strategy of squeezing every little bit out of a major move -- getting out, getting back in, repeat. Go ask a trader like Tom Baldwin how he feels about hitting home runs, or trying to hold through all the "big moves with a full boat" -- it just doesn't fit everyone's style, nor does it fit every market and market condition.

If all b1s2 is saying is that for position traders like himself, a few big trades are where most of your profits are going to come from, I have no disagreement. But it's all relative what a "home run" is, and for some alot of singles over time can add up to more than a few home runs. A wonder what a few qualifiers can do.

 

[trendy]

Quote from Buy1Sell2:

Scaling out is deficient, inferior behavior on any time frame and should be avoided.

Unless you catch the top tick or bottom tick on every trade B1S2, should me your trades, and I'll show you how scaling out would have been more profitable.

 

[Spooz Top]

Quote from Buy1Sell2:

Scaling out is deficient, inferior behavior on any time frame and should be avoided. [/B]

b1s2.........you`re not qualified to make that determination being that you are not a successful intra day trader...........as a matter of fact ,you failed miserably in your self proffessing thread about your "epiphany" on why you stick to position trading.......because of your failure to trade intra day succesfully.
perhaps your failure is due to the fact you refuse to open your mind & your stuburness interfered with an intra day approach,such as scaling,to become successful intra day trader.

the bottom line is to make money in any time frame you choose.....period ...& to use any approach one is comfortable with to attain their objective wich is a smooth rising equity curve.

i think you are an excellent position trader & have made some great calls as well as possessing a good market feel.However to insist any successful,consistently profitable trader is using inferior technique because you don`t implement them is a bit arrogant & narrow minded.

you are as correct about this theory as you are about me being another ET member named porgie....................flat out wrong.