I believe strongly that one needs the flexibility of averaging. Few are good enough to consistently "nail a level". For instance I know a great trader who posts here (more and more infrequently) who sold ES at 75 yesterday and had more to go up at 1280. Yea that would have been adding to a loser. Technically. Also woulda been the right play. As it is, turned out well. I've ALWAYS scaled. Here's the problems though. 1. If you're a scaled down buyer and the market's in good shape, you'll rarely have your max size on. Conversely if the market's getting hit and you're ultimately wrong then you will get filled on everything. 2. It's a bit different when adding to winners but just as dangerous. Let's say unit one shows a profit. Now you go for your add. BUT your add buys the swing high. Now as the market comes off and retests your original buy, you're a loser instead of a scratch. Adding to winners is a horrible counter trend strategy. Just think of how many buyers on monday's lows decided to buy more yesterday on the strength of the higher opening. Death. Turning winners into losers is very difficult psychologically. I consider position sizing and the "doubling cube" to be ultra important components in successful speculation. However THE essential ingredient is being right on direction and knowing when to hold em' and when to fold em'.
I agree and for me the main impetus for holding is that you are not near your loss comfort barrier at the time you hold or add.
Scale inversely to the odds of getting the shares. So higher the odds of getting it then the lower the shares. This will give you small shares for most of the trades but when it goes further then expected you will benefit from higher number of shares at higher prices. Consider 10% 30% 60% etc. So if trading 1000 shares do 100 then 300 then 600 then pray your ass off, etc. Sounds backwards but understand with mean reversion stuff you are open to hit rates but big big losers. The idea is to minimize the big losses. Fading is the same as selling naked options. They work high % of time but give big loses. John
i think i used to side with a similar view... however, i think the right answers may be strategy-dependent: let's take the Dollar Index as an example, i may see a +-16-year base/support at 80+-; so i may chose my stop arbitrarily below that level and if it gets hit i know i am wrong about support on the next test... - however, i trade equity index markets on an intraday basis - i may have an idea of my perfect level for entry, but the market may under-shoot this level many times, thus affording me plenty of opportunities to profit from partial scaling-in, while as when it finally gets to my level, that's when it may "surely" go for my stop too.... - what i am saying is that the "one shot one kill" approach may not work best in these liquid, inter-connected instruments, i.e. you may not be able to get the kind of perfect "precision" you are looking for.. - hence, scaling may be not just important but essential. this discussion is great; also, there have been plenty of threads on ET debating the pro's and con's of scaling-in; i am decidedly past that stage, and i am hoping to have some brainstorming discussion here on some specific scaling-in strategies, rather than argue about the general benefits of scaling - i understand it may not apply to everyone as the benefits may be strategy dependent - all the best.
agreed, i am looking to take a more specific, analytical probability-based approach to optimize the solution to the dilemma you are referring to. i do not see any benefits in this either, i am not considering this type of scaling-in, i am looking at scaling-in only at progressively better levels than the immediately preceding partial entry.
Ever play backgammon? Because of the randomness of rolls and betting angle, backgammon is perhaps closer to trading than poker. As defined: Double An offer made by one player to his opponent during the course of a game (just before the player rolls the dice) to continue the game at twice the current stakes. The opponent may refuse the double, in which case he resigns the game and loses the current (undoubled) stakes. Otherwise, he must accept the double and the game continues at double the previous stakes. A player who accepts a double becomes owner of the cube and only he may make the next double in the same game. http://www.bkgm.com/gloss/lookup.cgi?double
thanks for the input, i have gone through similar iterations - let me show how i think about this (still very much work in progress) i am looking at this on a risk/return basis. on the return side, i am assuming that the probability-weighted return is constant/equal at any level within the scaling-in window, which is the reason for scaling-in; in statistical terms, given the log-normal distribution of returns, i think you have to assume either constant or diminishing returns (i use constant for simplicity) but not increasing returns, i.e. as the price goes more against you, the expected return can't go higher... then on the risk side, the probability-weighted risk is basically equal to the value of an option struck at the stop price and valued at-the-money at each respective scale-in level. therefore, continuing with the option analogy, as the market price approaches the stop level, the ATM price will approach the strike price, hence the value of each respective option (and therefore, risk at each respective scale-in level) will get progressively higher. hence if you want to keep a constant risk/return ratio throughout the scaling-in window (which is what i meant by "proportional" in my opening post), and if the return side of the equation is fixed, then as the risk rises at each respective scale-in level, you will want to scale-in at a decreasing rate.
Lately I have been wrong about when to call a bottom and enter in to the position, but I have managed to turn a profit on 90% of my trades..... How is this possible..... Lets take IIIN as my most recent transaction..... You could just feel it was getting oversold after it's recent earnings that actually were above consensus.....I missed the first bounce from 37 to 42 so I put in a position at 39.52 hoping to get this stock on a steel sell off.....Of course the stock gaps from 41 to 34 over the two big sell off day's....I buy 3 times my original purchase 33.90 and it still continues to fall.... I get happy and buy more at 31.10 order filled for 5 times the original purchase..... So my average Price is 32.97 and it closes today at 36.74....I'm up 11% on a steel stock during this horrible market in the last 10 market days.... Had I never done any dollar costing I would be down 7%...(Some would call this increasing your risk, but the way I look at it is being able to get in the green faster)....I mean if you like it at 39.50 you got to love it at 31....And that's why I buy 5 times more once it has dropped +25% from my original purchase....Now of course capital and risk tolerance are factors, but one should apply dollar costing/scaling in to all long term positions or even trades...
thanks a lot, i remember backgammon from my childhood i guess, as applied to trading, you are referring to some sort of a martingale angle here; it's a great subject, although personally i haven't gotten around to it yet.