Central Banks & Big Players | Page 282 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

How do you know all the positions held by big players? 25 replies

ECN && STP && Scalping && Hedging - Broker Review - Oct 2009 24 replies

Central Banks and Big Players: Slight change in code reqd 7 replies

Japanese housewives are wrecking the forecasts of big players! 25 replies

Big players and Forex 4 replies

Attachments: Central Banks & Big Players
Exit Attachments

Central Banks & Big Players

  • #5,621
  • Dec 6, 2011 3:09pm Dec 6, 2011 3:09pm
Quoting jrodatus
Unfortunately it didn't do so well for 2009. But first remember I set the SL to 3 (instead of 2) to compensate for Duka's digit chopping. And second realize that that doesn't correct the effect of the chopping on the trailing stop: undoubtedly some trades that should've profited didn't.

Summary: it would probably do better if I had 5-digit data to test on. If anyone knows where to find some, let me know!

Results from (2009.01.01 00:00:0.000) to (2010.01.01 00:00:0.000)
----------------------------
totalTrades = 488
grossProfit = 700.0, grossLoss...
Wow, this is all definitely great news; the credit for the -5 SL idea goes to DDDD.

Is it possible to test something like a 5:50 risk/reward ratio? That is still a 10 to one ratio, and also run something like a 3:25 ratio as well?
  • #5,622
  • Dec 6, 2011 3:19pm Dec 6, 2011 3:19pm
Quoting jrodatus
Unfortunately it didn't do so well for 2009. But first remember I set the SL to 3 (instead of 2) to compensate for Duka's digit chopping. And second realize that that doesn't correct the effect of the chopping on the trailing stop: undoubtedly some trad
Summary: it would probably do better if I had 5-digit data t:0.000) to
Hi jrodatus,
Would it be possible for you to run the same but for TP = 20? I am going to calculate the system numbers tonight when I get home. Thanks!
r u ready?
  • #5,623
  • Edited 4:04pm Dec 6, 2011 3:39pm | Edited 4:04pm
Quoting rtrs111
Hi jrodatus,
Would it be possible for you to run the same but for TP = 20? I am going to calculate the system numbers tonight when I get home. Thanks!
Results for 2010: takeProfit=20, stopLoss=3
----------------------------
totalTrades = 520
grossProfit = 1020.0, grossLoss = 742.0, netProfit = 278.0
expectedPayoff = 0.535
drawdown: {0.00 | 2.07 | 3.00} (520)
shorts: won 26 out of 245. (11%)
longs: won 25 out of 275. (9%)
profitTrades: {20.00 | 20.00 | 20.00} (51)
lossTrades: {-0.00 | 1.58 | 3.00} (469)
consecutiveWins: {1.00 | 1.11 | 3.00} (46)
consecutiveLosses: {1.00 | 9.91 | 39.00} (46)

Results for 2009: takeProfit=20, stopLoss=3
----------------------------
totalTrades = 518
grossProfit = 880.0, grossLoss = 755.0, netProfit = 125.0
expectedPayoff = 0.241
drawdown: {-0.00 | 2.15 | 3.00} (518)
shorts: won 36 out of 267. (7%)
longs: won 25 out of 251. (10%)
profitTrades: {20.00 | 20.00 | 20.00} (44)
lossTrades: {0.00 | 1.59 | 3.00} (474)
consecutiveWins: {1.00 | 1.10 | 2.00} (40)
consecutiveLosses: {1.00 | 11.28 | 50.00} (40)

These took all of 40 seconds each to crunch
Imagine testing a year worth of tick data in MT4

If anyone knows where I can find 5-digit historical tick data, holler! Using 4 digits for this strategy is like driving the hairpin streets downtown... in a semi

EDIT: Hey rtrs, what do you mean by 'calculating the system numbers' when you get home? Just out of curiosity...

~Josh
Those who can kneel before God, can stand before anyone.
  • #5,624
  • Dec 6, 2011 3:42pm Dec 6, 2011 3:42pm
Im surprised the results for 20 pip TP were not better. I thought someone else did a backtest saying that 20 pip was hit many times.
  • #5,625
  • Dec 6, 2011 4:03pm Dec 6, 2011 4:03pm
Results for 2008: takeProfit=20, stopLoss=3
----------------------------
totalTrades = 519
grossProfit = 1100.0, grossLoss = 723.0, netProfit = 377.0
expectedPayoff = 0.726
drawdown: {-0.00 | 2.14 | 3.00} (519)
shorts: won 26 out of 259. (10%)
longs: won 29 out of 260. (11%)
profitTrades: {20.00 | 20.00 | 20.00} (55)
lossTrades: {0.00 | 1.56 | 3.00} (464)
consecutiveWins: {1.00 | 1.02 | 2.00} (53)
consecutiveLosses: {1.00 | 8.59 | 37.00} (54)
Those who can kneel before God, can stand before anyone.
  • #5,626
  • Dec 6, 2011 4:06pm Dec 6, 2011 4:06pm
Quoting Fedmayne
Im surprised the results for 20 pip TP were not better. I thought someone else did a backtest saying that 20 pip was hit many times.
Yes but I think that Jro stated that about half of the losses depicted in the results, were actually B/E trades. So the results are probably better than what is depicted here in the results.

Am I correct on my assessment, Jro?
  • #5,627
  • Dec 6, 2011 4:09pm Dec 6, 2011 4:09pm
Quoting Fedmayne
Im surprised the results for 20 pip TP were not better. I thought someone else did a backtest saying that 20 pip was hit many times.
Fedmayne, my tester is working with only 4-digit tick data. That's like driving hairpin streets downtown in a semi: you're going to hit a lot of unnecessary losses along the way. That considered, the 20 pip TP year-to-year had a very consistent success rate of about 10%, always bringing a long-term profit. But running the strategy on a 5-digit broker should produce even better results.

Still, I expect to see even better results for the 140 TP, since that is what Tony has found most profitable (after testing a wide range of TPs).

~Josh
Those who can kneel before God, can stand before anyone.
  • #5,628
  • Edited 4:52pm Dec 6, 2011 4:18pm | Edited 4:52pm
Quoting SmartGXL
Yes but I think that Jro stated that about half of the losses depicted in the results, were actually B/E trades. So the results are probably better than what is depicted here in the results.

Am I correct on my assessment, Jro?
Yes and no... as you say, the B/E trades are categorized as losses. But I think rtrs's concern was not how many losses there were, but rather how few times the 20 TP was hit. As the test stands, a 20 TP is hit about once per week (on average). Rtrs had the impression (based on another poster's manual backtest) that the 20 would be hit more often than that.

That may be the case. But I must keep mentioning that we need 5-digit data in order to find out, until it registers upstairs with some people
Those who can kneel before God, can stand before anyone.
  • #5,629
  • Dec 6, 2011 7:09pm Dec 6, 2011 7:09pm
Quoting jrodatus
Yes and no... as you say, the B/E trades are categorized as losses. But I think rtrs's concern was not how many losses there were, but rather how few times the 20 TP was hit. As the test stands, a 20 TP is hit about once per week (on average). Rtrs had the impression (based on another poster's manual backtest) that the 20 would be hit more often than that.

That may be the case. But I must keep mentioning that we need 5-digit data in order to find out, until it registers upstairs with some people
jrodatus,

Why you only got 4 digit tick data? I always get 5 digit from DukasCopy. Am I missing something here?

Regards,

screen
  • #5,630
  • Dec 6, 2011 8:45pm Dec 6, 2011 8:45pm
Quoting screenname
jrodatus, Why you only got 4 digit tick data? I always get 5 digit from DukasCopy. Am I missing something here?
Well I'll be! You see I actually obtained mine second-hand (did not realize that violates Duka's terms of use), because my computer would always crash trying to download more than a few days worth from their main history page. But since you mentioned it, I checked a CSV downloaded from the latter and- lo, and behold! 5 digits.

Thanks for the tip Mr. Screenname!

Now I've just got to download several years' worth of data... 1 day at a time

~Josh
Those who can kneel before God, can stand before anyone.
  • #5,631
  • Dec 6, 2011 10:03pm Dec 6, 2011 10:03pm
Quoting jrodatus
Results for 2010: takeProfit=20, stopLoss=3
----------------------------
totalTrades = 520
grossProfit = 1020.0, grossLoss = 742.0, netProfit = 278.0
expectedPayoff = 0.535
drawdown: {0.00 | 2.07 | 3.00} (520)
shorts: won 26 out of 245. (11%)
longs: won 25 out of 275. (9%)
profitTrades: {20.00 | 20.00 | 20.00} (51)
lossTrades: {-0.00 | 1.58 | 3.00} (469)
consecutiveWins: {1.00 | 1.11 | 3.00} (46)
consecutiveLosses: {1.00 | 9.91 | 39.00} (46)

Results for 2009: takeProfit=20, stopLoss=3
----------------------------
totalTrades = 518
grossProfit =...
That would be the system quality number. I am a big believer in Tharp's MM theories as u can see.
r u ready?
  • #5,632
  • Dec 6, 2011 10:28pm Dec 6, 2011 10:28pm
Quoting rtrs111
That would be the system quality number. I am a big believer in Tharp's MM theories as u can see.
Thanks, good reading there... I'm programming the SQN calculation into the tester.
~J
Those who can kneel before God, can stand before anyone.
  • #5,633
  • Dec 6, 2011 10:38pm Dec 6, 2011 10:38pm
Quoting jrodatus
Well I'll be! You see I actually obtained mine second-hand (did not realize that violates Duka's terms of use), because my computer would always crash trying to download more than a few days worth from their main history page. But since you mentioned it, I checked a CSV downloaded from the latter and- lo, and behold! 5 digits.

Thanks for the tip Mr. Screenname!

Now I've just got to download several years' worth of data... 1 day at a time

~Josh
Hi,

You can get the data more efficiently and more systematically by following this page:

http://eareview.net/tick-data

using either PHP or DukasCopier. With the php script you can update data much more easily esp. if you need data for many pairs.

You are very welcome. And thank you for posting your encouraging results.
  • #5,634
  • Dec 6, 2011 10:45pm Dec 6, 2011 10:45pm
I like the fact that people here are working together to achieve a common goal. The method tony has brought has some serious merit but certainly is not for everyone. Im curious as to whether or not 140 TP is actually more profitable than 20TP long term. If you can consistently make 20 pips with fewer losses it will compound much faster than 140 with many losses.
  • #5,635
  • Dec 6, 2011 11:11pm Dec 6, 2011 11:11pm
Quoting jrodatus
Yes and no... as you say, the B/E trades are categorized as losses. But I think rtrs's concern was not how many losses there were, but rather how few times the 20 TP was hit. As the test stands, a 20 TP is hit about once per week (on average). Rtrs had the impression (based on another poster's manual backtest) that the 20 would be hit more often than that.

That may be the case. But I must keep mentioning that we need 5-digit data in order to find out, until it registers upstairs with some people
I calculated the SQNs (where: 1-2 -> avg system, 2-3 -> good system, 3-5 -> excellent system, 5-7 -> superb system, >7 -> holy grail).
For TP = 20: 2009 had SQN = 0.89 and 2010 had SQN = 1.84
For TP = 140: 2009 had SQN = 0.0015 and 2010 had SQN = 1.98
If it was traded continuously for 2009 and 2010:
TP = 20 had SQN = 1.95
TP = 140 had SQN = 1.67

So we have a winner in TP 20? Would you say that with 5 digits we would probably have better SQNs? With AlienFrog's analysis we had a better SQN = 2.2, but he used minutes data. Still current results are impressive imo. Should we test other low spread pairs, times, TPs (10)?
r u ready?
  • #5,636
  • Dec 6, 2011 11:28pm Dec 6, 2011 11:28pm
Quoting rtrs111
For TP = 20: 2009 had SQN = 0.89 and 2010 had SQN = 1.84
For TP = 140: 2009 had SQN = 0.0015 and 2010 had SQN = 1.98
If it was traded continuously for 2009 and 2010:
TP = 20 had SQN = 1.95
TP = 140 had SQN = 1.67

So we have a winner in TP 20? Would you say that with 5 digits we would probably have better SQNs? With AlienFrog's analysis we had a better SQN = 2.2, but he used minutes data. Still current results are impressive imo. Should we test other low spread pairs, times, TPs (10)?
Hey, how did you get those SQN values? I thought Van Tharp's formula uses the standard deviation of the profits/losses of all the trades? Were you able to figure out the std.dev. from my report? Maybe I'm missing something. I got very different SQN numbers (higher). Will check my math/programming in the meantime. And yeah, I do expect higher SQNs for 5 digits, because we can have narrower stops, which means smaller risk, which means higher R-multiples for the profits.

Quoting Fedmayne
If you can consistently make 20 pips with fewer losses it will compound much faster than 140 with many losses.
Hey, I hadn't thought of that! You guys are starting to win me over to the 20 TP idea.

Quoting SmartGXL
Is it possible to test something like a 5:50 risk/reward ratio? That is still a 10 to one ratio, and also run something like a 3:25 ratio as well?
Sure. Just waiting till my 5-digit data finishes downloading.

Quoting screenname
...using either PHP or DukasCopier.
Nice. DukasCopier works on CrossOver Mac with Mono installed.

~Josh
Those who can kneel before God, can stand before anyone.
  • #5,637
  • Dec 6, 2011 11:36pm Dec 6, 2011 11:36pm
Quoting jrodatus
Hey, how did you get those SQN values? I thought Van Tharp's formula uses the standard deviation of the profits/losses of all the trades? Were you able to figure out the std.dev. from my report? Maybe I'm missing something. I got very different SQN numbers (higher). Will check my math/programming in the meantime. And yeah, I do expect higher SQNs for 5 digits, because we can have narrower stops, which means smaller risk, which means higher R-multiples for the profits.



Hey, I hadn't thought of that! You guys are starting to win me over to the...
I excluded all the BE trades - don't know 100% if it's good but u tell me.
r u ready?
  • #5,638
  • Edited Dec 7, 2011 12:15am Dec 6, 2011 11:58pm | Edited Dec 7, 2011 12:15am
Quoting rtrs111
I excluded all the BE trades - don't know 100% if it's good but u tell me.
Never mind, my programming was wrong!

EDIT: Now I can confirm your values. However, Van Tharp evidently said that sample sizes > 100 inflate the SQN so that it does not match the scale (as it would for # of trades < 100). Some say he recommended that you just put 10 in the numerator (i.e. sqrt of 100) for large samples to avoid this. That would lower the SQN value for our tests. But I say arbitrarily capping inputs sounds a little sketchy

~Josh
Those who can kneel before God, can stand before anyone.
  • #5,639
  • Dec 7, 2011 2:27pm Dec 7, 2011 2:27pm
Guys, you are getting WAY beyond my feeble ability to understand and keep up with all the programming and stat-speak. But it's VERY cool that everyone is working on it this way.

I'm having a terrible time with this. I downloaded about 4.5 years of Dukascopy tick data for several pairs. Each pair is one CSV file. Each file is 4-5GB in size!

The next major hurdle to overcome is that I have to convert these csv file to NinjaTrader import format. On that "Birt's EA Review" site there is a link to a script to convert Dukascopy data to NinjaTrader import format (if I am reading it right). But how do I run a .php script on the .csv file?

Sorry for dumb question, but I'm not a programmer.
Luck occurs when preparation meets with opportunity.
  • #5,640
  • Dec 7, 2011 2:31pm Dec 7, 2011 2:31pm
Quoting SmartGXL
Thanks for all your hard work Alien!
Thanks for the encouragment GX! I need it right now.
Luck occurs when preparation meets with opportunity.
5 traders viewing now