Do you think statistics work in trading? | 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?

  • #1
  • First Post: Jul 29, 2013 3:35pm Jul 29, 2013 3:35pm
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 work in market, he ran a lot of long term simulations and says that the price has probability to go up or down around 50% all the time. Tiny imbalances leading to trends do not last and almost impossible to catch with retail data.

I need to point out that we were talking about Forex, he said that in equities picture is much better for math traders. About my systems he said that the success is temporary and soon will be eliminated.

What do you think? Do you calculate any stat for the market? Have you found it useful or not useful?
Wookey
  • #2
  • Jul 29, 2013 4:09pm Jul 29, 2013 4:09pm
Statistically we have a better chance of going with the trend...that's all there is to it.

"he said that in equities picture is much better for math traders."

Why?
  • #3
  • Jul 29, 2013 4:27pm Jul 29, 2013 4:27pm
Quoting Vitez
Statistically we have a better chance of going with the trend...that's all there is to it.
That is exactly what he is arguing with. He says that every moment of time the probability of going up is within margin of error of 50%. So, basically, you cannot develop a model which would allow you to trade profitably basing on statistical research.
For example, you research how many up bars market can possibly make before making a down bar. And you found that at maximum market makes 13 up bars. Do you think it is safe for you to go short on 9th bar or go long on 5th?

Quoting Vitez
"he said that in equities picture is much better for math traders." Why?
Well, I am not too much into equities, but he said that there you can develop a statistical model giving you a statistically significant advantage.
Wookey
  • #4
  • Edited 6:41pm Jul 29, 2013 6:25pm | Edited 6:41pm
Quoting wookey
he ran a lot of long term simulations and says that the price has probability to go up or down around 50% all the time. Tiny imbalances leading to trends do not last and almost impossible to catch with retail data.
If that is true, and it actually might be, the whole forex market is a scam for retail traders. if there are no inbalances, nobody can win over long term.

To the Question of the topic: if not statistics, what else could work in trading? the only choice other than relying on statistics would be guessing.
  • #5
  • Jul 29, 2013 7:38pm Jul 29, 2013 7:38pm
Quoting pwnpwn
{quote} If that is true, and it actually might be, the whole forex market is a scam for retail traders. if there are no inbalances, nobody can win over long term. To the Question of the topic: if not statistics, what else could work in trading? the only choice other than relying on statistics would be guessing.
I do not know if it true or not, that it is why I came here to ask. Imbalances always exist, otherwise the price would not move. The question is if these imbalances are measurable with the retail data and if there are statistically significant patterns in them.

Forex is a playground of big banks. Equities are still traded. Now scalping and HFT almost killed short term manual trading in equities, but medium and long term trading is still dominated by human traders both individual and hedge fund. The human traders trades patterns, that is why you can find that they work there.

Completely different situation is in Forex. It is dominated by big banks including central banks. Which more often than not do trades for other than simple trading purposes. They do not care that you have a perfect butterfly pattern and the market should go down. They need those 5 billion Euro now. The other thing is manipulation, it is not a secret that banks make a lot of money out of foreign exchange operation. If you are a big international company which needs 3 billion Euro to pay, say taxes, but you do not have those, but you have dollars on your US account. You call your bank manager and give him instruction to exchange necessary amount of dollars to get Euro. The bank manager says: "no problems will be done". And here the game starts. No matter what rate the bank actually exchanged your money at. You get charged at the end of day rate called fix. So you can notice some small spikes near the fixing time. These are banks trying to push the price in needed direction. Because the further away they push the price the bigger difference they pocket courtesy of your company. They can push the price just 10 pips higher and your company is almost 4 million dollars lighter and the bank manager gets a nice bonus, thank you very much.
Wookey
  • #6
  • Jul 30, 2013 4:52am Jul 30, 2013 4:52am
Quoting wookey
{quote} I do not know if it true or not, that it is why I came here to ask. Imbalances always exist, otherwise the price would not move. The question is if these imbalances are measurable with the retail data and if there are statistically significant patterns in them. Forex is a playground of big banks. Equities are still traded. Now scalping and HFT almost killed short term manual trading in equities, but medium and long term trading is still dominated by human traders both individual and hedge fund. The human traders trades patterns, that is...

Agree to that. so many different people and different reasons people buy or sell eu/usd that it must result in randomness. But regardless of irrational behavior there still could be constant repeating periods that could lead to an edge. The biggest challenge is probably to have an edge that beats the trading cost.

Iam planning to run some 5+ year backtesting for simple EAs like your friend did. Not for purpose for a working EA but to find out if market really is random or not over long term. If you want i can post result if i find out some relevant infos.
  • #7
  • Jul 30, 2013 5:07am Jul 30, 2013 5:07am
IMHO statistics is all that works. I could not trade without statistics. Be it backtest statistics, or live performance statistics, couldn't do without them.
"To hold, you must first open your hand. Let go." - Lao Tzu
  • #8
  • Jul 30, 2013 6:26am Jul 30, 2013 6:26am
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.
No greed. No fear. Just maths.
  • #9
  • Jul 30, 2013 6:59am Jul 30, 2013 6:59am
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...
Did he look at using monte carlo method with multiple timeframes...
tradewith60
  • #10
  • Jul 30, 2013 7:10am Jul 30, 2013 7:10am
I did not say that the marker is a random walk and I do not think so. The market alternate from balance to not balance. In balanced move you can treat market as random and trade mean reverse, you cannot do that then there is no balance and the market is trending. However, the thing is that these transitions are pretty much random for us. We cannot extract statistically significant clues that market is going to change.

For example, you found a pattern, for simplicity lets say bullish harami, which you found working pretty well for some testing period of time. My friend insists, that over long term the probability of this patter getting positive result is 50%. So if you found it working in the past 6 months, in the next 6 months it will make losses to compensate.
Wookey
  • #11
  • Jul 30, 2013 7:14am Jul 30, 2013 7:14am
No need for MC, a histogram is enough. The 50% up/down is true. This is true for any TF. I even looked at range bar chart, same thing; except you get a M-shaped bi-modal distribution. Even taking the previous bar into account doesn't improve the forecast.

When you find a good forecaster (kalman, ARIMA or whatever) it comes with a big confidence interval...
No greed. No fear. Just maths.
  • #12
  • Jul 30, 2013 7:15am Jul 30, 2013 7:15am
Attached Image (click to enlarge)
Click to Enlarge Name: histogram_EUR_H4.png Size: 28 KB
Attached Image (click to enlarge)
Click to Enlarge Name: histogram_EUR_RG20.png Size: 19 KB
No greed. No fear. Just maths.
  • #13
  • Jul 30, 2013 7:28am Jul 30, 2013 7:28am
Quoting wookey
I did not say that the marker is a random walk and I do not think so.
My misunderstanding.

Quoting wookey
The market alternate from balance to not balance. In balanced move you can treat market as random and trade mean reverse, you cannot do that then there is no balance and the market is trending. However, the thing is that these transitions are pretty much random for us. We cannot extract statistically significant clues that market is going to change.
You cannot even find statistically significant clues that the market is now in balance. Unit root tests fail with market data (fractionally integrated time series).

Quoting wookey
For example, you found a pattern, for simplicity lets say bullish harami, which you found working pretty well for some testing period of time. My friend insists, that over long term the probability of this patter getting positive result is 50%. So if you found it working in the past 6 months, in the next 6 months it will make losses to compensate.
50% is good as long as SL << TP. Let's let the fat tails do the job the open side of the trade.
No greed. No fear. Just maths.
  • #14
  • Jul 30, 2013 7:29am Jul 30, 2013 7:29am
Statisticians have gone back over 60 years examining distribution curves for all markets traded. The results - the markets are indeed random, at least to the extent of ever predicting the next candle. I personally like to say that the market is"random with structure", meaning that if enough traders are looking at the same information at the same time, then an increased probability can exist. This should not be a surprise to the experienced trader. I've said many times that knowing the various times frames that are being traded - from one minute to one month, different account sizes being leveraged, the utilization over 300 major indicators, trading a plethora of strategies and methodologies; add in manual, mechanical, robotic, and high-frequency trading; in addition, technical and fundamental analysis; throw in spot, futures, gold, options, and everything else that factors into currency value; then factor in retail and institutional; now add all the human emotions that determine every trading decision; all of this going on at the same time - and we think we can determine the next movement of a candle? I think not.

For the most part, we've all been had. When we finally understand the above, we will no longer look to financial analysts and advisers; pay money for worthless software programs, study fallacies such as moving average cross-overs, hidden divergence, and back-testing (and many more I could add). Our eyes will be opened to the truth - as we learn to trade using numerical probabilities of reward versus exposure. The best piece of advice I was ever given was "stop thinking like a trader and start thinking like a casino". The casinos added just one slot using a green double '00' to a roulette wheel and jumped the house's probability of winning by 5.26%. That's the way we must trade - we must find our own "double 0".
  • #15
  • Jul 30, 2013 7:34am Jul 30, 2013 7:34am
If you become 'competent ' in reading S/R then the markets become non random.The problem is this art of chart reading is highly subjective and appears random to the statistician. I believe for people with a firm grounding in maths/applied physics should focus on statistical arb and mean reversion strategies. They have a good advantage in this methods.
wo-yoy! wo-yoy! wo-yoy! wo-yoi! wo-yoy-yoy-yoy!
  • #16
  • Jul 30, 2013 7:42am Jul 30, 2013 7:42am
try adding fundamentals. That helps

Oh, I see, you cant backtest fundamentals because they are not written on chart and is no convenient software to do it.

winning in markets consistently will take more than stats, even if talking kalman filter.

And there is more!

You then have to add sentiment. Unfortunately, not on chart either. You can use option volatilities for that (right move!) but instead of 52% you will get 55% win rate.

And then you will also need stomach to cash in on these odds, which most underestimate. Going thru large drawdowns is no fun, I can tell you.
  • #17
  • Jul 30, 2013 8:08am Jul 30, 2013 8:08am
Fundamentals not written on the charts? Euro crisis is quite visible. Bernanke printing money is quite visible too. BNS' decidion that EUR/CHF cannot go below 1.2 you can't miss it on the chart. Aussie and its interest rate curves are quite in sync to me.
Attached Image (click to enlarge)
Click to Enlarge Name: australia-interest-rate.png Size: 18 KB

If a currency is fundamentally doom to fall it will fall and will form a down trend on the chart, no?
No greed. No fear. Just maths.
  • #18
  • Jul 30, 2013 8:17am Jul 30, 2013 8:17am
Well, fundamentals may help, no doubts. But this thread is about market statistics. I hope traders beside me use it and have at least moderate success. I want to know what people think about it. What stats people use or what they think may work.
Wookey
  • #19
  • Jul 30, 2013 8:35am Jul 30, 2013 8:35am
There is a harsh reality the average "trader" will never fully accept; that this system of facilitating trade is designed to move capital and it needs to be fed losers. The retail crowd is inherently trained to trade (via the internet and purchasable "guru" programs) the same way and that way consistently allows for dealers to maintain target inventories and adjust quotes appropriately. Realistically, we're all just sports bettors dealing with multiple bookies who are constantly adjusting the odds dependent upon where the crowd is positioned. Think about it;

Go to:
-FXCM
-OANDA
-SAXO BANK
-DUKASCOPY
-MT4i
-FXPRO

Look at current and historical retail positioning data as it relates to price adjustments. I don't think the two are contradictory (most of the time) because the retail trading crowd is "stupid". I think it does this because this is how trade is facilitated for the entities that "matter". We're not supposed to win long term and when we do, it's a statistical anomaly. I can guarantee that I'm not the only person on this forum who's entered a buy position in this market only to have my exact entry price become resistance a few moments later - this happens for a reason and that is the one thing we can't see on a chart.


Market statistics are fascinating and I'm no mathematician. I'm a simple guy who still hits the button with his own finger, lol. But I've studied a lot of the methods your average retail trader who's been in the game long enough has and honestly, none of it will really help you overcome the forces you're competing against. You may do well for a while - just long enough for you to open a trade explorer and brag on a forum - but eventually, the dealer always gets that money back; balance and imbalance.

Now, I'm not saying folks like us can't win long-term. We absolutely can. But it will always be the minority of us.
  • #20
  • Jul 30, 2013 10:22am Jul 30, 2013 10:22am
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 money?
There are so many myths - the biggest being "trade with the trend". Again I ask - which one? The truth is there is
no such thing. What you see are continuations of buying and selling. You may say well I call that a trend. Don't be
blindsided by the fact, as we all know, that a continuation can end on the next candle. Imagine emptying a container of
red and white marbles down a transparent pipe. 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. This is what you're seeing in what some refer to as a trend.
I'm not saying to not trade with the movement of the money but we must realize what it is and what it isn't.

The first thing we must do in trading is to forget time frames - the price doesn't know which one you're on. Time frames
should only be used for your personal time convenience and nothing else. Learn what price the market will accept and what
it won't. Understand the probabilities of gain and your exposure with every trade you make. Although trading is challenging,
it is still the best business in the world. The problem is that everyone starts trading the same way - MACD's and Bollingers.
For most, it will be years before they finally look behind the curtain and see who's really pulling the ropes.
0 traders viewing now