What is considered an edge? - page 2
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
That's a nice plain and simple way to put it. As I like to say, testing statistics generate probabilities.
On money management I'd split it clearly. It doesn't create an edge, it decides how far you can scale one and how likely you go broke on the way. With a real edge that's the part that keeps you alive. Without one, it can only change how the result looks.
Bingo, again. When I observe average consecutive loss count patterns in the statistics, that allows me to adjust position my sizing accordingly.
On the code:
i'm on the buyer side and never see it. So i can't prove an edge, only collect evidence.
That would be a bit frustrating for me. When I observe the aforementioned patterns, being able to additionally alter the underlying trade logic as necessary is highly advantageous.
An edge only means something if you can show the result is unlikely to be luck.
Some luck will be needed since retail traders use brokers which doesn't offer centralized markets. Those brokers decide themself what spread to use to match other orders.
Yes, we can use futures instead. But that will give some other problems, cost, volume and so on.
if the live per-trade mean falls inside the backtest's confidence band, the backtest was honest; if it sits well below, look at costs, spread and fills before blaming the market. Finally, keep a period the optimizer never saw and judge only on that.
Feels like a strategy based on small pip gain/loss will get a bad outcome in your test.
Hard to make it scalable too since there need to be a buyer on the other side, else you will get slipped and requoted.
Every test you do on a backtest will only be the statistical probability of the "edge" working, going forward.
You will never know if it does, unless you have a fix in the market or some insider information.
Then compare the live trade list to the backtest over the same period
Abdullah, I think you are on to something here.
There are a lot of seller telling you that their strategy have an edge because its visible in the backtest they show you.
Or they show you a real account that has survived a year or two, but with 10 upgrades along the way making the trades differ between live and backtest.
Does it have an edge or is it only luck + statistical probability?
So the edge is just statistical probability?
On money management I'd split it clearly. It doesn't create an edge, it decides how far you can scale one and how likely you go broke on the way. With a real edge that's the part that keeps you alive. Without one, it can only change how the result looks.
A grid or martingale is the typical case: lots of small wins, a nice win rate, and the losses come together in one bad week. It looks like money management but it raises the risk of ruin instead of lowering it.
Here you take a calculated risk on the previous statistical probability and hope it bares fruit.
On the code:
i'm on the buyer side and never see it. So i can't prove an edge, only collect evidence. For me that's a live record that matches its own backtest over the same months, with enough trades. And every change of settings starts the clock again, because the earlier live months were traded by a different EA.
You are probably right. Either you take a chance (gamble) with the EA you just bought in the market here, or you try and come up with your own strategy instead.
The latter at least gives you full insight into what the EA does and doesn't do.
All and all
Isn't it still just a statistical probability that the EAs sold here will continue to give you money based on historical values on the market they trade?
No retail indicator will give better results that 50/50 over time, and most EAs here are based on just those retail indicators.
Nothing wrong with buying any of the EAs here if you like to gamble with your money or spend your time dreaming of successes. It can even most likely be fun sometimes.
As I like to say, testing statistics generate probabilities.
When I observe average consecutive loss count patterns in the statistics, that allows me to adjust position my sizing accordingly.
But you don't know if the next trade will be a homerun which you just lowered the lot size on.
When I observe the aforementioned patterns, being able to additionally alter the underlying trade logic as necessary is highly advantageous.
100% agree on this.
Using somebody else's EA without knowing what it does is like throwing knifes in the dark, in a rubber boat.
So no EA here on mql5 have any real edge, just statistical probabilities?
True to the title of this Topic/thread, that continues to beg the question of what an "edge" is. As for me, I absolutely detest dull knife blades so I guess that's my edge. The word, edge, tells me nothing about qualitative, quantitative, statistical, nor probabilistic metrics of an EA. More appropriately, "edge" is more of a word thrown around by gamblers and swingers (not the trading kind) who haven't analyzed nor deduced a darn thing.
But you don't know if the next trade will be a homerun which you just lowered the lot size on.
First, I don't bank anything on potential homeruns. Such windfalls simply are not part of my trading. Second, the word, know, strikes me as existing in a binary paradigm─one either knows or doesn't know. I operate in the markets based on the principle that nothing is 100% knowable. This brings us back to the "edgy" people referenced above, who are likely "knowy" people as well. Not very useful for any kind of metric.
Using somebody else's EA without knowing what it does is like throwing knifes in the dark, in a rubber boat.
The way that I keep my knives... absolutely.
The word, edge, tells me nothing about qualitative, quantitative, statistical, nor probabilistic metrics of an EA.
Isn't it statistics and probabilistic metrics that is used by gambles?
An edge is something like a senator buying stocks in a water meter company, before they just happened to get a large deal on water meter upgrades in the same city as the senator lives. (true story by the way)
An edge is knowing information, that will influence the market, before it is officially announced. Many senators and politicians use their knowledge to their own gain or to someone they know.
An edge is knowing where most stop orders are located. Some banks sell this information for large amounts to hedge funds.
This is a double edged sword though, because you cant just throw in a billion into the market when you know price will go to a certain level. Other hedge funds will notice this and act upon it.
Several banks have been fined to pay billions for manipulating the markets many times. Now that is a fix, but obviously illegal too.
Isn't it statistics and probabilistic metrics that is used by gambles?
IMHO, rarely. Decades ago, I exploited a statistical blackjack matrix but so did every other blackjack gambler having at least half of a brain. If I remember correctly, the matrix was developed by MIT. Lo and behold, the casinos all decreased the blackjack hand payout from 2.0 to 1.5. That was enough to blow up the money adjusted 50.2% win rate of the matrix, so that was when I quit gambling. In somewhat of the casinos' defense, was I really gambling? Was that actually what they took issue with? Aside from a situation like that, I seriously doubt that most gamblers have a well defined and empirically tested system.
Now that is a fix, but obviously illegal too.
Yeah, if insider trading were an edge, bribing a quarterback would be the Green Bay Sweep.
Yeah, if insider trading were an edge, bribing a quarterback would be a play.
You need to get caught to make it illegal though, else it never happened.
Just saying that people here seem to think they have a chance gambling with their Ma crossover or overbought/oversold oscillator, or every other combination of retail indicators.
"They've done studies you know. 60% of the time, it works every time."
You need to get caught to make it illegal though, else it never happened.
😂 I love it. The Whitey Bulger style of trading, if you will.
They've done studies you know. 60% of the time, it works every time."
A naked win rate doesn't exactly constitute data forecasting. It's possible that a trader can win only 43% of the time and still come out ahead─for example, if the average winner is 7 x the size of the average loser. Point is... There's a lot more to data forecasting than merely saying, "I have a 60% edge." Frankly, there are probably more metrics in the native MT5 Tester Report than the average trader will ever read. Additionally, I created my own auxiliary custom backtest report. Sample size is a good place to start. I won't use less than 4000 bars.