That is true, but then you would have to really sum over different contributions/populations in a way that Harry suggests in his example with people of very different heights. My point was about the measure you used for the opportunity factor, it does not have to be per day. I see your point, but rather from the example that Harry gave. Only then the opportunity factors are really necessary, for a homogeneous distribution it just boils down to rescaling my formula by multiplying it by a number that can have some measure (like 1/day) or not. That does not give you anything really new unlike in the example given by Harry.
Wally, The interaction of opportunity factors with an understanding of a heterogeneous return distribution (this latter item applies for anything but scalping) helps to better elaborate upon your initial expectancy formulation. Without an imputation of the opportunity factor, it is meaningless to compare per trade expectancies between strategies. I hope this helps. Tharp discusses expectancy in significant depth in his publications. Some of his books are in the Elitetrader books section. If you need any further advice gained from my experiences, feel free to private message me. Good luck with your trading and best wishes, Candle
Candletrader, thanks for your remarks. I now perfectly understand what you meant, I was just being simplistic in a way Harry was, but I do realize that when you deal with heteregenous distributions the opportunity factors are bound to enter into the formula we are talking about. I have one of Tharp's books, but I know this topic as someone who professionally worked on casino game strategies, where the expectancy is what casino players simply call the edge. Just like traders.
Allen: I agree that as an intraday player we are sometimes missing larger moves which extend over a few days but... a.) Playing intraday I can trade a position size I would not dream to carry overnite. b.) Sure, in a strong trending mkt that has gaps in the direction of the trend it will be unlikely you will catch a large % of the move,but what about the times the dailys are in a trading range and 2 weeks later the mkt has bounced between 2 points and is back where it started, but has had 20 pt +S&P swings intraday?.The swing traders are frustrated and the intraday players are smiling.Just my thoughts.
I can automate fixed point trailing stops but was trying to see if there were better ideas out there defining where your trailing stops were and your initial stops. The problem is not following the stops but seeing if I can improve upon the results with using a better stop and trailing stop placement.
First, you got some really fine responses in my opinion. I particularly liked AAAintheBeltway and Profitseer's answers. Setting a stop has alot to do with your style of trading, the volatility of the market, and how you enter the market. That said, a 2 point stop wouldn't work for me. Really, the idea of setting a stop like this....a particular point amount....is really trying to "game" the market. But it really has nothing to do with the market per se. The market can fluctuate 2 points for absolutely no reason at all. Take today for instance: you could have shorted the ES at the high of the day just after the opening. It moved 3 points in your favor, and then back to the high of the day. Somewhere in that process you would have been stopped out. It then took the lows out, and rallied another 2+ points, before continuing on lower. See the point? It's just fluctuating in the process of moving lower. It continued this process until about 2PM CST, when it hit it's low of the day. Now, I don't know what type of trading you're wanting to do. But in order to ride that move down today, you had to do something less mechanical than trail a 2 point stop. If you're trying to "game" the market, find a stop that's ideal, then I think you'll be in for a long, frustrating process. But that's just my opinion...other's here may have a different view. I set a kind of "failsafe" stop similar to that described by AAA and Profitseer. It's a very wide stop, which is there more or less for disaster protection. and nothing more. I set it with an eye toward volatility, but these days I've been mostly using 10 points. I can almost here the wails going up over a stop of that size. And certainly I have no real intent that this stop would normally be triggered. Again, it's just there in case something unusual happens during the time I'm in the market. That 10 points is as much as I'm willing to lose on a trade. But hear this too: I trust my market judgment. But I also know that sometimes I'm too early. Sometimes it needs to top or base first. And therefore, I give it room to do so. I let it develop. At the same time, I watch carefully to see if it's doing what I expected. If it starts to give me feedback that is not consistent with my expectation, then I exit immediately. I don't wait for the stop to be executed. Sometimes, what will happen is that I'll get on the wrong side of the market, and let's say it moves 5 points against me....but at that point, looks due for a reaction. I might wait a little for the reaction, and then exit at that time. But as long as the market is acting in ways that are consistent with my expectation, I stick with the position. I don't move a stop in there for them to hit. A 2-3 point move against me I'll sit through as long as the underlying reason for holding the position has not changed. Now let me tell you something: I don't know of any other way to stay in a big move. Because sometime during that move it's going to move against you, and if you have a tight stop in there it will knock you out. What I've found is that if you have the direction right, if you get out on a very tight stop, chances are you won't be able to get back in at a better price. Now, this is just me. And again, I'm not a scalper. I believe in watching the market, exiting when it starts to act in ways that lead you to believe the move is over....and not just because it reacted a couple of points. OldTrader
MAE means the largest amount by which a position goes against you, and is usually separated into winning trades losing trades So if your winning trades according to backtesting have gone against you by 3 points at the most, before turning profitable, then setting a stop at 3.5 won't stop you out of any of the winners. (If history repeats itself) . MAE is established through backtesting and Wealth-lab for example gives you a nice graphical representation. The trick is (obviously) to use the numbers to set a stop that minimizes losing trades while staying in winners. If your system shows losing trades generally have a MAE LARGER than the winning trades, then tightening stops to just outside the winners MAE works fine. If else, tightening stops also will take away some winners.
I said that because statistics can be used for descriptive purposes of course but is above all useful for predictive purposes (or inferential - to use a probability term). As I have worked in SPC field (Statistical Process Control for Quality Engineering), I am used to distinguish between specifications (what you would like) and capabilities of the system (what the system can do in fact). Generally you want the moon and the system can't do it. The first step would consist to revise the specifications to more realistic adequation. In this first step you won't decompose and take aggregate datas (ie you mix the Giants and the Lilliputians). This will give you a rough idea of the capability of the system. Then come the other steps of improvment. In theses steps you'll have to refine so as to distinguish random factors from controllable factors. To do so you have to split your datas into subgroups, for example between diffrent machines, workers, procedures or process. You will have to eliminate outliers so as to get what is called the intrinsic capability of the system. By doing so you want to get the usual (or normal or predictable) behavior of the system and eliminate the exceptional part. Then you will study exceptional parts either good or bad and if you can find factors you will make correction to the process and improve the intrinseque capability of the system that is to say exceptional performance will become a normal part of the system. This is a rather general improvment process which is part of a more global scheme called PDCA of Deming's wheel continuous progress circle. It can be used in any field so trading could use it also. So my use of statistics in trading is rather for methodology than for modelisation of stock markets. Statistical model of stock markets give poor results because of non-linearity. I prefer to use a determistic approach which is hugely more difficult to find: in fact you have to have ... chance to find the root idea