Do you think statistics work in trading? | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Statistics combined with system. Profitable? What do you think 71 replies

If you haven't made it yet, what makes you think you will? 212 replies

Do YOU think EA's work (POLL) 70 replies

Do you really think a broker will let you get away with that? 4 replies

Trader Statistics and Broker Statistics 0 replies

Attachments: Do you think statistics work in trading?
Exit Attachments

Do you think statistics work in trading?

  • #21
  • Jul 30, 2013 11:02am Jul 30, 2013 11:02am
Hi wookey,

I'll try to answer you the best I can.

Statistics and odds are everything in trading, as you always enter the market with an estimation (probabilities) of your risks and potential rewards.

What you are trying to do is study data series in order to obtain patterns that give you better probabilities than 50%+spread in order to make profits. The first thing you must take into account is that this field has been studied for decades with statistical models that take into account price data + fundamentals data + intermarket correlations data. Also the big boys add to the model their flows data which they have because they are the backbone of the flows.

Given this, you have to understand that any pure technical advantage is eliminated when a few companies get to know it... so by now, you can expect the probabilities of finding a pure technical edge to be quite low.

If you are capable of mixing your mathematical understanding with markets behaviour, economic fundamentals knowledge and psichology you can do well. Also you can try to go for options trading, I think your math skills add value in that field.

Good luck mate!
  • #22
  • Edited 11:50am Jul 30, 2013 11:39am | Edited 11:50am
Quoting Warburg
Read Post 19 again and accept it for that's just the way it is. Someone mentioned support and resistance. Looking at this or any a number of (indicators) and my first question to you is "What time frame are you talking about?" As we all know - one time frame is telling us to buy, another to sell, while another is telling us to do nothing. We must keep one thing in mind - even the brokers say 95% lose in this game - I suspect it to be closer to 99%. With this being said - something has to be terribly wrong. Name me another business where 99% lose...
A while back, I stopped asking myself "Who is in control, bulls or bears?" and began asking "Why are dealers moving their quotes?" When it comes to indicators, I'm sure you've noticed, they always have us selling and buying "overbought" and "oversold" levels - as if something can legitimately be "overbought" and "oversold". The price of Twinkies doesn't fall when people buy the heck out of 'em, right? Warburg is absolutely right about time frames - for ten seconds price can be in a different trend than it is every minute, five minutes, thirty minutes, and so-on. I've learned that the trick is capital risked and time frame must be proportional to desired holding time. They tell us "Cut your losses and let your winners run" well, how many people have let a winner run real well for a few hours (or days) only to lose the position (or take less profit if you moved your stop loss above break even)? This happens for a reason, and it's because we, as traders - are gasoline for a giant, complex engine that finances the global economy. Hell, I played a "$12.00 an hour" strategy a while back where'd I'd buy or sell (depending on if the daily bar was up or down) every hour and take profit every 12 points and cut losses every -6 points. Surprisingly this works better farting around all day waiting to sell into upward momentum.

There is a structure - like Warburg is saying:

Quote
What you see is maybe 2 white and 3 red in a row, then 7 white and 2 red, then white, white, red, red - you get the idea.

I began looking at the market in a similar way except using quarters. The market will move up and down in increments of 25 points over and over again, all the time on smaller time frames. I usually take into consideration which side covers more quarters - up or down - into establishing a bias. On much larger time frames, I look at 250 point moves. Same principal, just bigger.

Quote
The problem is that everyone starts trading the same way - MACD's and Bollingers.

I completely agree. There is so much evidence that validates this statement it's crazy. If MACD, Stochastics, Bollingers, et al really worked effectively, long-term everyone would be using them and we'd be ransacking dealers left and right. There's a reason our respective broker(s) offer trading seminars and classes. It isn't to make you a better trader, it's to amplify your confidence to get you to trade more because they make money on the spread. I remember when I first opened my account with $2,000, my broker could care less I existed. I was still working a full-time job so I never traded. I was building capital so I could retire and trade full-time. When I hit the $10,000 mark, like clockwork, they called me every Monday (I was still working back then) asking me if there was a reason why I wasn't trading and if there was anything they can do for me. As the equity went up, the phone calls came more and more. Then when I retired and started trading (making and losing money) - the phone calls stopped.

The whole thing really is a racket - and this isn't coming from someone who's "blown their account" and is pissed about it. I still have yet to blow an account and frankly, am surprised how quickly some people can accomplish that feat regardless as to how much capital they have. But it's the reality of it. In the world of finance, no one cares about your success, failure, or feelings. It's all about providing you a computerized casino so they can gradually sap money out of you and move it to (as Jesse Livemore said) "It's rightful owners".

I remember having an "ah-ha" moment when watching the movie "Casino" where "Ace" (De Niro) talked about the Japanese "whale" who came in and bet $30,000 a hand on Baccarat and took the casino for $2 million.

Quote
It's a whale like K.K. Ichikawa who plays $30,000 a hand in Baccarat - that's the one you really gotta watch. He plays fast and big and has the cash and the credit to turn out your lights. About a year ago he cleaned out a couple of casinos in the Cayman Islands. Downstairs he takes us for $2 million and upstairs he takes free soap, shampoo and towels. Another billionaire cheapskate who loves his free rooms, free private jets and $2 million of our money.

But we got him back.

I had our pilot tell him the plane was on the fritz. I don't know what went wrong. I'm awfully sorry. I can't understand it. These mechanical things, you know they happen. Hey, but better here than up there, you know what I mean. Then he missed the commercial flights connecting with Japan. I can't believe that there's a convention; there's no room. We got him back with a whole floor of rooms to himself.

I'm sorry you missed your plane.
Glad to get to know you.
Oh, no. No gambling.

And once he was back, he played small. He bet $1,000 a hand, instead of his usual $30,000 a hand - to the bank with a natural lead over a five. But I knew the trick with whales like Ichikawa was that they can't bet small for long. He didn't think of it as winning $10,000 - he thought of it as losing $90,000.

So he upped his bets until he dropped his winnings back and gave up a million of his own cash. In the casino, the cardinal rule is to keep them playing and keep them coming back. The longer they play the more they lose. In the end, we get it all.

That's how it rolls.
  • #23
  • Jul 30, 2013 11:41am Jul 30, 2013 11:41am
AS I see it, there are 3 dimensions to trading

TA/PA
Fundamentals
Psychology

Using only 1 of these, usually TA or PA, your odds of winning low, same as observing object movement in 3D vs line (1D). There will be lots of fake edges visible that will start and stop working without notice.

One can say, use MACD to measure masses psychology and gaps for fundamentals. As this is heavily studied, I believe there is too little edge left to exploit for long term success.
  • #24
  • Jul 30, 2013 12:00pm Jul 30, 2013 12:00pm
Quoting seaman2
AS I see it, there are 3 dimensions to trading TA/PA Fundamentals Psychology Using only 1 of these, usually TA or PA, your odds of winning low, same as observing object movement in 3D vs line (1D). There will be lots of fake edges visible that will start and stop working without notice. One can say, use MACD to measure masses psychology and gaps for fundamentals. As this is heavily studied, I believe there is too little edge left to exploit for long term success.
I agree to an extent. The "missing" dimension I noticed when I was on my "grail quest" years back was how the machine operates. You'll occasionally see once-popular threads about market micro structure and dealer operations that most (MOST) traders quickly get bored with because it isn't telling them how to make money. I was in that boat once and by NO MEANS am I implying that I'm some kinda pro because I'm not. There's new stuff to learn every day and largely, I learn it by losing money. My neighbor works at Morgan Stanley as a manager and we were talking trading on the 4th of July (while we were blowing things up, lol). He was laughing about all the emphasis people put on demo and paper trading because when you buy and sell nothing, your order doesn't get put into the "machine" - and the market will 9-times-out-of-10 react differently when you're actually in it then when you're not. It's a coin toss. You do the best you can to stack the "odds" in your favor but in the end, all of it means nothing if the market decides to go the other way. Now, if the market decides to go in your favor, that's when we start to think highly of ourselves; how good our strategy is, how smart we are - and this keeps us trading until that strategy inevitably falls apart.

That's when the dip in the emotional roller coaster comes for most retail traders. Then they hit the internet and the forums looking for the next big strategy that they can utilize.
  • #25
  • Jul 30, 2013 12:03pm Jul 30, 2013 12:03pm
Quoting Warburg
Statisticians have gone back over 60 years examining distribution curves for all markets traded. The results - the markets are indeed random.......
Very good post.
The whole thread is great and would make 99% of other threads in this forum useless if read.

I used to play poker for a living for six years. The game is very similar to trading. 95%+ people are losing long term, but people don't stop playing...why? Because they don't realize that they dont have an edge. They win some hands and think it's skill. Then they lose some hands and now they think it's bad luck or lacking of concentration or whatever
  • #26
  • Jul 30, 2013 12:04pm Jul 30, 2013 12:04pm
Quoting wookey
Hello, I am a mathematician, so I like when every action is explained mathematically. I use various statistics in my trading systems starting from support/resistance (to see which level statistically more accurate) to validating patterns (for instance if a certain kind of engulfing patter has more probability to work). My systems have moderate success trading automatically. So I am pretty happy with them and math behind them. However few weeks ago my friend, also a math graduate, pointed out to me that his research shows that statistics does not...

Statistics will help you determine what the idea of trading is .. but won't help you trade.. I know thats a confusing answer but its the truth .. most people look at trading like if trade with stop 1 and tp 1 hits 51% or more its profitable.. but that doesnt work. They seem statistically to work but the problem is that they all revert back to the mean over the long term minus the spread so it wont continue to make money as the number of trades approach infinity. I have been trading for a long time now.. I have watched the rise and fall of trading ideas and gurus. Everything from candlesticking, trend is your friend, holy grail indicators, bar fading, wave analysis, the naturalists(you know the cant see the forest through the trees people, or rivers run around rocks etc...), to the statisticians, price action trading, gailing, grid trading, god the list goes on I even went to school to program genetic algos and hft. I have watched profitable traders .. eat it after 7 years profitable with the excuse the market has changed and isn't moving the same... what they experienced was their strategies reverting back to the mean minus consumption by the spread drawing them closer to 0.

What I can tell you is this, price is based on how people feel about the market based on events that are happening or are perceived to be happening. Most of these events are generally pretty random.. a tsunami hitting japan causing nuclear melt downs, war breaking out based on social issues in the mid east, massive earthquakes, and etc.. the list of random event go on forever. the future is random which means perception is based on the random events or possibility of random events. That perception or changing perception however has a pattern or groups of patterns that can be traded initially about how long those perceptions last is anyones guess. It is beyond anyone to know the future or be able to predict it based on statistics.

As a statistician tho you should be familiar with intentional skewing of normal bell curve distributions and this to some extent is what trading is.. this is what we get paid to do. the ability to trade random future perception surrounding random events with the ability to minimize risk and maximize profit and skew normal distributions of profits so they don't return to the mean over time.
  • #27
  • Jul 30, 2013 12:13pm Jul 30, 2013 12:13pm
Quoting pwnpwn
{quote} Very good post. The whole thread is great and would make 99% of other threads in this forum useless if read. I used to play poker for a living for six years. The game is very similar to trading. 95%+ people are losing long term, but people don't stop playing...why? Because they don't realize that they dont have an edge. They win some hands and think it's skill. Then they lose some hands and now they think it's bad luck or lacking of concentration or whatever
I remember a while back I went to the Hard Rock for the first time with my wife. I figured heck, I'm trading, may as well try this crap. We brought our change jar so we figured "what the hell?" Her mother - a gambling addict (seriously) - was with us. We sat at a row of those electronic poker machines. It took me a minute to get used to it because I'm not a card player but I quickly understood something about the machine - I'd always win a little bit of money before quickly losing it. As her mother sat next to me dumping hundos into this machine and telling me I have to play fast because it's lucky - I was cashing out every time I won a little and coming back and putting in $10 less than I did the last time.

I did this for four hours. No sh*t, her mother lost like $8,000 and I wound up making $450. It's not much, but heck, I wasn't a gambler guy. I just considered the machine to be a constant downtrend for your equity and when it was up, take it, cash out, come back, put in a little less money than before. I just used $10 as my marker because I was starting with $100 bill. It was fun
  • #28
  • Jul 30, 2013 9:34pm Jul 30, 2013 9:34pm
The reason why I and many other traders use statistics is advantages given to us by usage of computers. I wrote tonnes of code trying to make computer to recognize patterns. But computers are much less able to do it than humans. Even most sophisticated pattern recognition programs I cam across or created myself are easily surpassed in quality by an average trader.

Where computers have edge over us is calculation. On of my statistical program monitors 700 symbols at once for trading opportunities in real time. It keeps data for up to 3 years for each symbol and does extensive calculation. Needless to say that no human trader can come even close to do the math even for one symbol.

As I said, my strategies are a moderate success, so I do think that there is some merit in using statistics, but I have concern about long term.

Good systems never use 1 pattern. If you know that 1 pattern has 50% of success and another one is 50% it is very good information. You can look for combination of those patterns to increase your probability. For instance if you take any bullish engulfment you get 50/50, but if you get the engulfment near the pivot point the probability will be different. It is almost impossible to calculate but a simple strategy will till you quickly. If joint probability increases you have a winner, if it decreases then you have a filter if it does not change then the patterns are not to be used together.
Wookey
  • #29
  • Jul 30, 2013 9:50pm Jul 30, 2013 9:50pm
litteraly you can't know what will be the trend within 1 day or 1 week or 1 month but you have some " hack methodes " to get profits from forex without knowing if euro/usd will grow up or down within the next hours for exemple martyngale but i DON'T recommend you the martyngale except if you can invest at less 1000€ and profits will not be huge
  • #30
  • Jul 31, 2013 3:31am Jul 31, 2013 3:31am
Quoting Fundaments
What you are trying to do is study data series in order to obtain patterns that give you better probabilities than 50%+spread in order to make profits.
I just can't understand this. Why those 50%??? 30% is enough if the reward is 4-5 times the risk. Where is this "need" of being "right" coming from? It is a probability game. If a trader buys a bullish flag breakout and the price goes against his SL, was he "right" or "wrong"? To me he was right because odds were 50-50 while the risk was smaller than the potential reward. This is independent of the outcome of one single trade.

Quoting aketo25
litteraly you can't know what will be the trend within 1 day or 1 week or 1 month but you have some " hack methodes " to get profits from forex without knowing if euro/usd will grow up or down within the next hours for exemple martyngale but i DON'T recommend you the martyngale except if you can invest at less 1000€ and profits will not be huge
Martingale IS NOT a way to win. It is just differing the loss. Living on credit is NOT being rich.

Quoting wookey
Good systems never use 1 pattern. If you know that 1 pattern has 50% of success and another one is 50% it is very good information. You can look for combination of those patterns to increase your probability. For instance if you take any bullish engulfment you get 50/50, but if you get the engulfment near the pivot point the probability will be different. It is almost impossible to calculate but a simple strategy will till you quickly. If joint probability increases you have a winner, if it decreases then you have a filter if it does not change...
I completely agree with the idea of using stats to estimate the PDFs of the events and Bayes' theorem to estimate a likelihood of a confluence of technical elements.
No greed. No fear. Just maths.
  • #31
  • Jul 31, 2013 6:45am Jul 31, 2013 6:45am
Quoting PipMeUp
{quote} I just can't understand this. Why those 50%??? 30% is enough if the reward is 4-5 times the risk. Where is this "need" of being "right" coming from? It is a probability game. If a trader buys a bullish flag breakout and the price goes against his SL, was he "right" or "wrong"? To me he was right because odds were 50-50 while the risk was smaller than the potential reward. This is independent of the outcome of one single trade. .
The 50%/50% is a theorical simplification.

Regarding "To me he was right because odds were 50-50 while the risk was smaller than the potential reward". Well, if your profit target is bigger than your stop loss, usually your chances decrease of 50-50.

Regards
  • #32
  • Jul 31, 2013 8:10am Jul 31, 2013 8:10am
I think that mathematics will never working on market. it's not exact things, it's very fluctuate and unpredictable...
We only can use trap on there, with EA, fundamental,and with technical analysis..
  • #33
  • Jul 31, 2013 9:40am Jul 31, 2013 9:40am
Quoting PipMeUp
Your friend is right about the 50-50% for market to go up X or down X. FX is symmetric. Isn't this already statistically significant? If the market was a random walk, could E/U and G/U be so well cointegreated while U/J is almost independent from them while they have USD in common? At the end of the day what's the difference trading a fundamentals driven up trend or a spurious up trend as long as you're long? Using stats to forecast the direction is vain IMO but we can use stats on a lot of other things, like the TA patterns.
This is the foremost if not only credible way to use statistics in trading. i.e. "forecasting (edge based a probability)" price behavior around statistically meaningful points around technical analysis. And you will find that MA's have significant bearing in that analysis. Also to note, the MA is a statistical number at a point in time on the chart.

That the market is likely to go up or down X at any point is true. But what also is true is that the market tends to take or go in particular directions (behave a certain way) for certain periods of time from time to time. And it is in that analysis lies the edge that traders can exploit successfully. I really don't pay attention to individual price bars. I think that what is of utmost significant in trading is directional analysis over select samples of price bars - easily derived by MA and THE RANGE (aka HEIGHT).
  • #34
  • Jul 31, 2013 10:17am Jul 31, 2013 10:17am
Quoting wookey
What do you think? Do you calculate any stat for the market? Have you found it useful or not useful?
Maybe a more useful quest would be to answer "Why" certain instances defied the bell curve, rather than just observing that they did, and then assuming they will continue to.

This is the problem with TA and FA... repetitive patterns do exist but they represent some kind of repetitive behavior, which is a product of a certain structure, which can be undone with one certain event..

its a jungle out there, it will never agree to stay in your box. You need to put a face and a name to the numbers so to speak.
Time hides Nothing
  • #35
  • Jul 31, 2013 3:51pm Jul 31, 2013 3:51pm
Quoting CindyXXXX
{quote} Maybe a more useful quest would be to answer "Why" certain instances defied the bell curve, rather than just observing that they did, and then assuming they will continue to. This is the problem with TA and FA... repetitive patterns do exist but they represent some kind of repetitive behavior, which is a product of a certain structure, which can be undone with one certain event.. its a jungle out there, it will never agree to stay in your box. You need to put a face and a name to the numbers so to speak.



my best trades com from this logic :

FA or sentiment derived signal -> and only then check TA/stats to set parameters for trade.
Same applied to scalping where I anticipate good environment for my type of scalping and only then start scalping.
If none of above, I still trade and usually lose but try to keep bet size as little as possible.
  • #36
  • Jul 31, 2013 4:11pm Jul 31, 2013 4:11pm
Statistics is somewhat useful in the inside game, but as with any technical or fundamental or psychology or charts, a Z at extremes for example, can and usually does redevelop. A stochastic at 90 can flip over and start a whole new trend. An analogy, if you're a soldier on the frontline, it's important to have a solid grasp of the inside game, your tools, but it's critical to have a very clear understanding of the most important outside game, your total surroundings.
  • #37
  • Jul 31, 2013 4:43pm Jul 31, 2013 4:43pm
I believe you look at the wrong values. Simple bar count and pip values won't work since the start and end of a time frame is completely random. (I did the same with renko bars).

You might want to take a closer look at retracement values. Look at the chart I traded from the past 2 days. 2 Red lines always belong to each other with the second line drawn in advance of the price movement and those lines are just simple 1 to 1 extensions. (Look at the Dow theory)
Thats what I call precision. You can't tell me that this is random!

I am all into math and statistics doing a lot of research (countless of excel spreadsheets) on martingale and antimartingale strategies with standard deviation and confidence intervals. Grid trading correlation. I have been there I have done that.


I did retracement research on big indexes like the DAX. If you take a look major swings (they can automatically be detected using a model with zigzags (of course the best zig zag values has to be computed for each situation in advance!! Hint:I would take a look at volume and the ATR ) and fibonacci values (at the beginning I was strongly opposed those values as they seem random and if you just draw enough lines price will always come close to one of those), you can find a significant distribution which differs from a Gaus curve.
Attached Image (click to enlarge)
Click to Enlarge Name: EURUSD.png Size: 41 KB
  • #38
  • Jul 31, 2013 5:38pm Jul 31, 2013 5:38pm
Trading forex is gambling.
  • #39
  • Edited Aug 2, 2013 1:16am Aug 1, 2013 10:22am | Edited Aug 2, 2013 1:16am
none
  • #40
  • Aug 1, 2013 10:40am Aug 1, 2013 10:40am
Quoting wookey
Hello, I am a mathematician, so I like when every action is explained mathematically. I use various statistics in my trading systems starting from support/resistance (to see which level statistically more accurate) to validating patterns (for instance if a certain kind of engulfing patter has more probability to work). My systems have moderate success trading automatically. So I am pretty happy with them and math behind them. However few weeks ago my friend, also a math graduate, pointed out to me that his research shows that statistics does not...
There are imbalances in the market but not statistical imbalances that can be spotted by technical analysis. Imbalances are created by economic growth, central bank rates, liquidity etc... plain statistics doesn't spot imbalances, you need to have a concept before you can start statistical modelling.
0 traders viewing now