Systematic trading | Page 4 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Trade another day through Systematic Trading 203 replies

Nonparametric Association Measures in Systematic Trading 57 replies

Anyone use Systematic from Alpari? 1 reply

A Systematic Approach to Markets 68 replies

Systematic Trading vs. Discretionary Trading 14 replies

Attachments: Systematic trading
Exit Attachments

Systematic trading

  • #61
  • May 13, 2009 5:35pm May 13, 2009 5:35pm
Hi Mikkom
,
Quoting mikkom
I would like to discuss a bit about trade risk sizing.
As most people probably know, one method of calculating "optimal" risk is kelly but kelly does not take volatility into consideration at all.
Very good point! I'm sure I don't have to say that but I'll always emphasize it - volatility measurement and its later application is critical for systematic trading. I'm even tempted to say that systems that don't take vol into account are useless long-term. The amount of effort put into research of it at my workplace seems to confirm my standpoint

Quoting mikkom
Does anyone know is there any kelly like formula that would take volatility of the profit curve into consideration (= sharpe etc.)?
What about treating vol directly as a measure of risk (or the inverse of it) ? Possibly as a scaling factor for what Kelly formula produces? Possibly relative vol ??? Nice area for research. Anyway - Euan Sinclair - "Volatility Trading" is all you need (and much more). There's an entire chapter on how to make risk and vol live a long and happy life together. There's a place for Kelly in it as well.

As a matter of interest - could you share you thoughts on the volatility estimation / prediction methods you use? This is so crucial that I'm always happy to know other points of view.

Regards,
Ipso
  • #62
  • May 13, 2009 6:49pm May 13, 2009 6:49pm
pls this souds very good. i m waiting to see how it gose. keep me posted.
  • #63
  • Edited 6:11am May 14, 2009 2:05am | Edited 6:11am
Quoting Ipso
As a matter of interest - could you share you thoughts on the volatility estimation / prediction methods you use? This is so crucial that I'm always happy to know other points of view.
Are we talking about profit curve volatility or market volatility (I was talking about profit curve volatility in my previous post ie. maximal sharpe/sortino/etc)?

I trade directionally so I personally don't try to predict volatility in a sense that you are probably talking about - I just try to find points in price where the risk for entry or exit is lower than on other points.

Only place where I try to predict short term volatility is my initial stop position and even there I don't try to predict volatility per se but maximum directional movement in that direction for the current day after the point of entry has been hit.

That said, I plan to take a good look at delta neutral strategies too in the future.

ps. thanks for the book pointer, I'll probably order it.

pps. why do all the trading books have to have a psychology section?
  • #64
  • May 14, 2009 3:37am May 14, 2009 3:37am
Quoting mikkom
pps. why do all the trading books have to have a psychology section?
Well, that is the standard recipe of trading books these days, especially books from Wiley Trading/financial press. Little bit of meaningless statistics, bits of history, some trading myths, some thoughts on psychology, some correlation stories or fallacies, back test results of few strategies, few charts, comments/reviews of well known traders or analysts, and a declaimer that may resemble "The systems presented in this book are by no means exhaustive, and they are limited by space and the author’s imagination and experience", but the truth is, the concoction they present can fool any curious novice trying to find an edge in the realm of trading... it is indeed a great business..

I do have some questions related to systemic trading.

How to measure Risks of trading in the financial markets?
How a quant quantifies risk when they are developing trading systems, I guess, there are many types of risks and risks depend on the type of strategy, I was wondering Mikkom or other quants could give pointers to measure risks for various types of trading strategies.

What is the best algorithmic trading platform?

What is the average lifespan of an algorithmic trading system?
　
No Fear, No Regrets A sense of purpose makes life meaningful
  • #65
  • Edited 7:53am May 14, 2009 7:01am | Edited 7:53am
Quoting Bizken
I do have some questions related to systemic trading.

How to measure Risks of trading in the financial markets?
If you know trade frequency and risk for each trade you know the maximum risk too.

Quote
How a quant quantifies risk when they are developing trading systems, I guess, there are many types of risks and risks depend on the type of strategy, I was wondering Mikkom or other quants could give pointers to measure risks for various types of trading strategies.
If you trust backtest results for the systems you trade (I do, I know how my systems work and what they are based on), you can run backtests and see what the historical drawdown has been. Drawdown basically defines risk. Backtest data has to be quite long to mean anything, I always want at least 10+ years of backtest data for multiple markets.

Many people use montecarlo simulations to simulate different curves but I don't like them because they are just simulations, nothing more. I know many people disagree with me on this.

There are some well known ways to define proft/risk ratio like sharpe, sortino etc.

Quote
What is the best algorithmic trading platform?

What is the average lifespan of an algorithmic trading system?
　
Best is relative term, I didn't found a good platform so I decided to go with my own.

Average lifespan depends on the system, as I don't know all the systems I can't know the average span. I wouldn't trade a system that wouldn't make good profits with relativety low dd on 10+ years of data.

Usually systems have periods when they behave better and periods when they behave worse. That's why it's important to haev multiple systems that work on different kinds of market types and basically hedge the risk of one system with another system.

ps. I would very much like to hear other peoples answers to the questions too.
  • #66
  • May 15, 2009 2:49pm May 15, 2009 2:49pm
Hi Mikkom,

I admire your courage for starting this massive project on your own. I wish you will do very well.

On the topic of APIs, I was looking today at forex com which is the retail subsidiary of Gain apparently. I noticed something very interesting, although I can't confirm it, but this is how I understood from their 2 websites. It seems that they offer tha ability to use the API even when trading small amounts via forex com, and I would imagine this will be the same API used when trading with Gain. Now, Gain has minimum account 100k USD, so it would seem as advantage to start using the API with the smaller accounts in the retail business, and if and when you ever have the capital for an account with Gain, all your software would be ready to go.

As I said I can't confirm any of this, but it was my first impression. Anyone had any experience with this API?

Also, Gain is one of the best capitalised brokers, which is an advantage, however, I think they are based in the USA, which, could be a disadvantage, given the latest developments about NFA and anti hedging rules. Plus I dont know anything else about the reputation of their retail broker.
  • #67
  • May 17, 2009 9:34am May 17, 2009 9:34am
Looks like I got things sorted out with ActivTrades and they reassessed that I'm not scalper

I'm still going to go with the FIX gateway, this whole episode reminded me of why I want to use ECNs rather than market makers. The good thing is that now I'm in no hurry anymore.

About gain, I have heard that Gain is an awful bucket shop so I would never trade with them.
  • #68
  • May 18, 2009 6:10am May 18, 2009 6:10am
ok mikkom thaks for opening my eyes on this one before I spend any time on them.

Quoting mikkom
About gain, I have heard that Gain is an awful bucket shop so I would never trade with them.
  • #69
  • May 22, 2009 5:01am May 22, 2009 5:01am
Some more simple stats from EURUSD below, this chart tells on what day the weekly high or low has been hit. This should tell you something about what is statistically the best way to enter/exit the trade.
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot-3.png Size: 7 KB
  • #70
  • May 22, 2009 1:59pm May 22, 2009 1:59pm
What are you using to generate these graphs?
The breaking of a wave cannot explain the whole sea.
  • #71
  • May 23, 2009 5:48am May 23, 2009 5:48am
Quoting Craig
What are you using to generate these graphs?
I just take output into file from mysql and create graphs with openoffice.
  • #72
  • Edited 2:06pm May 24, 2009 1:34pm | Edited 2:06pm
Nice, I wish I had the SQL skills to do this sort of thing.
One does wonder when viewing graphs like the one above if we are not just finding patterns in randomly generated data (like the frequency histograms one can produce for lottery numbers, for example). The Monday and Friday bars do seem quite pronounced, but how do we tell the difference between an actual effect and a randomly produced skew of the type mentioned above?

My own answer to this would be that some type of statistical confidence test would need to be applied to data so that we could assign a probability to the Monday or Friday data not been due to chance alone, but I'm interested in what other people have to say on the subject.

Edit: If we satisfy ourselves this is a non-random effect, I'm also very interested in how you turn this into a trading rule.
The breaking of a wave cannot explain the whole sea.
  • #73
  • Edited 10:32am May 25, 2009 1:31am | Edited 10:32am
Quoting Craig
Edit: If we satisfy ourselves this is a non-random effect, I'm also very interested in how you turn this into a trading rule.
I'm trading a system based on this and it has been effective on multiple markets (I trade and test majors) for at least 20 years. The funny thing is that when your usual trend following systems (turtle etc) have went from very profitable to profitable, this one has gained profitability or has been quite stable.

I'm not going to spoil the fun but think about all the typical trading systems, one particular type of them (very well known) is using this exact thing to make profits - also look at the hourly chart data I posted, they tell the same thing (bar open and close are very pronounced in high/low distribution). It's really quite obvious when you get it.

I have been thinking of doing a matrix that shows high of week (daily+hourly) and distribution of lows with that high. I think it would be a very interesting chart and would tell something even I'm not expecting.
  • #74
  • Edited 10:24am May 25, 2009 9:47am | Edited 10:24am
Quoting Craig
My own answer to this would be that some type of statistical confidence test would need to be applied to data so that we could assign a probability to the Monday or Friday data not been due to chance alone, but I'm interested in what other people have to say on the subject.
And about this, as this is EURUSD data, you could also make similar graphs for other majors. If all 4 majors show similar results it's IMHO fair to say that this is not a random occurence. I'll see what I can do..

edit: below
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot-symbol_high.ods - OpenOffice.org Calc.png Size: 8 KB
  • #75
  • May 25, 2009 10:33am May 25, 2009 10:33am
Quoting mikkom
And about this, as this is EURUSD data, you could also make similar graphs for other majors. If all 4 majors show similar results it's IMHO fair to say that this is not a random occurence. I'll see what I can do..

edit: below
Would you mind doing above for USDCAD, AUDUSD and EURJPY ?

Thks..
The future depends on what we do in the present.
  • #76
  • May 25, 2009 10:39am May 25, 2009 10:39am
Quoting compro99
Would you mind doing above for USDCAD, AUDUSD and EURJPY ?

Thks..
Sorry, I only trade and analyze majors (I haven't even downloaded any other data than majors).
  • #77
  • May 25, 2009 7:08pm May 25, 2009 7:08pm
mikkom,

Can you analyze below based on your majors data ? Thanks.

1) which week is the High/Low of the month

2) Which month is the High/Low of the year

Would love to see the results..
The future depends on what we do in the present.
  • #78
  • May 26, 2009 1:28am May 26, 2009 1:28am
Quoting compro99
mikkom,

Can you analyze below based on your majors data ? Thanks.

1) which week is the High/Low of the month

2) Which month is the High/Low of the year

Would love to see the results..
I'll see if I have time to do that.. However I only have 10 years of data meaning 10 highs/lows and that really doesn't tell much. I also aren't believer in seasonal movements in forex (in some other markets like commodities I might very well be).
  • #79
  • May 26, 2009 1:39am May 26, 2009 1:39am
Here is another interesting chart. This one is from "Bollinger on Bollinger Bands" (good book, worth buying if you don't have it). This is the reason why breakout methods work on market data but wouldn't work on randomly generated data

(This is stock market data - I assume forex data would be even more non normally distributed)
Attached Image
  • #80
  • May 26, 2009 8:14pm May 26, 2009 8:14pm
Quoting mikkom
Here is another interesting chart from EURUSD. This chart tells what hour the daily high was hit. As you can see, it's heavily biased to first hours of day. This should tell you something Also note how the last hour just jumps out from previous hour.

My data is gmt+1 so all the charts are the same.

edit: added daily EURUSD low hour below the daily high chart

Just to clarify, the chart is saying that the daily high most often occurs at hour 0 (gmt + 1), so hour 23 gmt (11 o'clock in the evening London time). Isn't this about Asian session lunchtime?
The breaking of a wave cannot explain the whole sea.
0 traders viewing now