what is trading system edge?

http://elitetrader.com/vb/showthread.php?s=&threadid=74217&highlight=edge+test

tireg   Registered: Jan 2006 Posts: 184

08-03-06 08:52 PM A few months ago, I wrote about my thoughts on the Edge. Ever since that question was posed to me, I have been seeking what defines an edge. Before, in my naivete I thought an edge was positive expectancy, or discipline, or money management. Those who were in the know scoffed at the idea and stated that these are things that are part of a successful system, but do not constitute a 'real edge'. For the last few months, the concept of an edge was bugging me. It seemed everyone was talking about pressing the edge and edge this and edge that, as if it were the most obvious thing in the world. It seemed everyone knew what an edge was. Yet I didn't. I mean, I had a general idea of what the results of a profitable system were, and looking back at my own strategies I could see what worked, and what didn't, but I could not for the life of me pinpoint the exact edge, and why the systems were profitable. I've finally come to a definition of 'Edge' that I am comfortable with, and clarified how it relates to expectancy. THE EDGE dictionary.com defines edge as it applies as: "A margin of superiority; an advantage: a slight edge over the opposition." Many things can be edges; the most prevalent image I had in my mind was that of a casino, in which the house has a built in 'edge' - due to the probabilities of the casino games being in slight favor of the casino, over time they are profitable. This is a statistical edge. An example of a statistical edge for trading was demonstrated in Schwager's interview with Gil Blake. Blake recognized that the markets were not completely random - there exists pockets of nonrandom price behavior which can be taken advantage of; in his case, more than 70% chance of follow through after x days, allowing him to swap funds. He took this a step further and applied it to many trades throughout the year, applying his edge over many markets. Another example is faster execution - someone posted once that if all traders could pick tops and bottoms, those that could do it fastest would make money. This is a competitive advantage; an execution edge. Many funds' edge is an analytical one. An analytical edge or informational edge is having analysis that others (re: the market) have not taken into account, thus locating inefficiencies in price. The product of the edge, that is, how it manifests itself, is positive expectancy. Positive expectancy is the result; not the edge itself. Tying this all together, in a profitable trading system with positive expectancy, the edge is the system's ability to locate and generate trading signals based on nonrandom price behavior, no matter what time frame - seconds, hours, days, weeks, months, years, etc, or style - fundamental, technical, statistical, etc. As retired EliteTrader forum poster Acrary has shown, multiple uncorrelated trading systems/strategies come together to smooth out the equity curve and create consistent profitability. Probability also plays a role, as the more vehicles or trades generated, the more the effect of the edge, i.e. expectancy, plays out.

shanoballs   Registered: May 2005 Posts: 48

08-07-06 01:20 AM Edge: This is one of the most basic concepts in trading, yet it seems like most starting out pay very little attention to this, so i will make an attempt to define it. When it comes to trading, an edge is defined as something that will put the odds in your favor. Quantitatively or mathematically speaking, having an edge would mean that whichever methodologies that you are using to enter and exit positions yield a positive mathematical expectancy. To simplify it further, this would mean that during the trading session you were trading with a higher probability of being net positive than being net negative. The most common examples of negative expectancy games are the games played in casinos. They are presented with the odds in the favor of the house, meaning the house has the positive expectancy and you, the gambler, have the negative expectancy. The only reason that one might win in a negative expectancy game is because probabilities are distributed randomly. Simply put, people win in casinos for the same reason why you might get 4 heads in a row instead of heads-tails-heads-tails while flipping a coin, though the chances of it being heads or tails is 50/50. Formula for Expectancy: Expectancy = [(((WIN: LOSS ratio) + 1) x (Percent Profitable)) - 1] So for example if your average win is $600 and your average loss is $200 then you have a WIN: LOSS ratio of 3, and if on average out of a sample of 100 trades 40 were winners then your percent profitable would be 40% and your expectancy would be .60. In this case we would be trading with a positive expectancy. Expectancy = [(((3) + 1) x (.40)) -1] = .60 The key is to have a positive expectancy when you trade. Of course, by now you may notice that to even determine if you are trading with an edge or not, you need a good sized sample of trades, and you have to be applying your methodology without any deviations from the rules throughout the entire sample for this to mean anything. If you were inconsistent in your methodology or approach, then it would be very difficult to come to any conclusions by analyzing your results because they will be random at best. Of course in this case you will not know if what you are doing (your methodology) is actually putting the odds in your favor or not. This is why many starting out fool themselves with "discretionary" trading, but that is an entirely different discussion. In a world where market returns are normally distributed the equity curves of numerous expectancies would look similar to this: But returns are not normally distributed therefore a positive expectancy equity curve in reality might look like something like this: The expectancy of an edge can vary anywhere from -1 to 1 and beyond. However, for you to be profitable, your expectancy must be positive, and of course the higher above zero the better. A realistic expectancy number to shoot for, at least in my opinion, would be .50. So there you have it. That is my two cents on what an edge is.

tireg   Registered: Jan 2006 Posts: 184

08-07-06 01:44 AM Quote from shanoballs: Formula for Expectancy: Expectancy = [(((WIN: LOSS ratio) + 1) x (Percent Profitable)) - 1] Also simplified as: (Average win size * %tage wins) - (Average loss size * %tage loss) Money management and risk management have to do with avg win and avg loss size, whereas %tage wins and %tage loss have to do with Edge and market conditions. Note that positive expectancy is NOT an edge, but is the result of many things coming together, part of which is an edge. Also note the distinguishing of Edge and market conditions. Great way to tie in everything.

tireg   Registered: Jan 2006 Posts: 184

08-07-06 01:58 AM Add-in: Notice how u can still have positive expectacy - ie make money, even though you may not have an edge, if market conditions prove favorable. This is why it is important to perform proper 'edge-tests' to confirm that one does have an edge (proving statistical significance of edge vs random). Interesting because it brings into the debate if strategy or tactics should be weighted or given emphasis on - strategy meaning correctly defining the market conditions to employ the proper system, be it trend following, reversal, etc. or tactical meaning finding edges and exploiting them. Acrary in the end discovered that strategy played a much larger part b/c he spent most of his later years finding and exploiting edges.

inCom   Registered: Apr 2005 Posts: 100	08-07-06 02:28 AM Quote from steve46: The comment about proper testing and evaluation of a system to determine an edge is THE POINT. Without that, you certainly can make money, and you may THINK you have the ability to make more than you lose, UNTIL mysteriously it all closes down on you. THAT is the difference between having an edge, and simply being the beneficiary of random good luck. When one runs out (luck) you don't get any advance warning. When an edge stops working, IF you know how to test, you DO get an advance warning. Steve [/B] Excuse me, but even if you have a proven statistical edge, your properly tested system can start losing money anytime. Also I think to compare casinos' edge to a trader's edge is misleading because casino games are based on a very restricted rule set. Market are changing continuously. So even if you think you've found a statistical edge, it may disappear at any time. GS
		Edit/Delete • Quote • Complain

tireg   Registered: Jan 2006 Posts: 184	08-07-06 02:42 AM Quote from inCom: Excuse me, but even if you have a proven statistical edge, your properly tested system can start losing money anytime... So even if you think you've found a statistical edge, it may disappear at any time. GS Thus emphasizing the necessity of proper edge testing over time. Once your edge starts to degrade, it is time to move on to a new one. This is the same for any type of system - even an edgeless trend following system will make money in a trending market - think the late 90's bull market - but in a trendless range-bound market, it will suffer. I would rather trade with an edge and properly test the significance of it, having advance warning if it starts to weaken, than to rely purely on market conditions. Think about it. If you are trading without an edge, how will you know when your system is suffering? A drawdown would be a rude awakening.
		Edit/Delete • Quote • Complain

steve46   Registered: Mar 2003 Posts: 3575

08-07-06 02:46 AM You are excused. You are correct. Even a properly tested system MAY lose money, and it can start to lose money at any time. IF you have understood my comment, you will remember that there is a way TO TEST, so that you know if your edge is degrading or disappearing. IF you didnt have an edge to begin with, because for instance you believed that YOUR edge was the ability to make money, then IF and WHEN you start to lose money, you will have no idea why... I hope this makes it clearer. IN a casino, periodically they will lose more than they take in on specific games. They watch this process carefully because they know what the limits of their losses should be if the game is being run correctly. IF the losses exceed what the math projects as an upper limit, they KNOW that someone is doing something wrong and it is affecting the profits. Thanks Steve http://elitetrader.com/vb/showthread.php?s=&postid=1633345#post1633345 acrary   Registered: Apr 2002 Posts: 700   10-08-07 01:13 AM Edge is simply taking advantage of the non-random nature of a market. They are quantifiable and persistent through time. A simple example of a edge in price data is "trend". Every market I've ever checked has a trending tendency beyond random. Some more than others. Incorporating that tendency would be to simply place trades in the direction of the trend. The best edges (largest degree of non-random behavior) are found external from price data, however. Something as simple as observing money flows into or out of mutual funds several days prior to price moves has been a exploitable edge for myself. Some people believe edge is nothing more than positive expectancy in a trading strategy. While this could be the case, you should satisfy yourself that the model captures persistent, and non-random opportunities. Otherwise the model will go under when the short-term "fitted" data reverts to it's more random nature. tireg   Registered: Jan 2006 Posts: 184   10-14-07 12:45 PM Through my experiences, my understanding of an 'edge' has been refined. In terms of hedge funds, or businesses, an edge is one fund's competitive advantage over its peers. This can manifest itself through experience, analytical capability, execution capability, technology platform, connections, and access to information. Holding all other variables constant, an increase of each of these will mean a manager has an edge over another. These manager/fund-specific qualitative factors are one of the primary reasons why hedge fund replication strategies do not work. Simply replicating exposures of a 'typical' hedge fund misses this point. In terms of a trading system, an edge is the system's ability to locate and generate trading signals based on nonrandom price behavior over market conditions. It is important to distinguish a system's behavior over market conditions (re: proper backtesting) and randomness because a random 'buying' strategy will appear to have an edge when the mean of the returns over the period is positive (bull market). In this context, 'money management and position sizing', 'positive expectancy' and 'discipline' are NOT sufficient to be an edge, though they may be assumed necessary. One of my biggest pet peeves is when people confuse these terms with what is necessary or sufficient to an edge. The reason that positive expectancy is not sufficient to be an edge is that again, in a bull market, a random buying strategy will tend to exhibit positive expectancy, as will a trading system with an 'edge'. At the same time, in choppy markets, a trend-following system with an edge may exhibit flat to negative expectancy. However, having positive expectancy over the aggregate market conditions is necessary. Discipline is not an edge in any form. If you don't have discipline you should not be putting money at risk; you are no better than a lazy gambler. It's as ridiculous as saying having an account at a broker is an edge for trading. Most retail traders struggle with this; this is why they are the amateurs at the bottom of the totem pole - risk, and quite often losses - gets transferred to these market participants. Money management/position sizing is necessary to maximize an edge, however it is not required as one form of backtesting involves trading 1-lots. In addition, normalized-risk trading systems incorporate position sizing/money management into their trading rule, so this could be thought of as a piece of the system itself. Therefore, position sizing is neither necessary nor sufficient for an edge.

[undeadrogue]

谢谢岁月老师 =)