NFP System - Backtest data provided

In addition to how I said that the trade should increase with every buy or sell order (so add a line after ticket made, TradeCount = TradeCount+1), TradeCount can be set to 1 manually in inputs, so that if the user starts sometime in the middle of the month instead of right after NFP cooling off period, then they can tell the EA that it's already had a trade on the pair, so that the extra conditions of price above highest high/below lowest low, for previous and current week, are also taken into consideration, before the trade is taken. Very nifty way of psyching out the EA to not just using the "1st trade after NFP cooling off period" rules (since entering in the middle of the month may make someone nervous, using the extra conditions are a great way to limit losing trades).

Don


[quote=ForexNewDude;2444941]
What is TradeCount = 0 for?

I do not see any code that adds to this counter. It seems to stay zero forever unless I am wrong. Is there code to increase this once an order has been made?

Place these indicators on your 6 charts:
EURUSD, GBPUSD, AUDUSD, USDCHF, USDCAD, USDJPY

I have been told that NFP was not released Friday, so the user will have to put in the time manually.

How I figured when NFP is (normally) released my time:
NFP Start day is:
one day in the range of first 7 days of the month
day of week is Friday
*IMPORTANT*
NFP Start Time is 15:30 Central Time
Note: I calculated this by knowing that FXDD's Server time goes to 00:00 at 4PM Central time my time. From this point, there are 8 hours until midnight Central time, then the next morning NFP is released at 7:30 AM Friday Central time. So add 8 + 7:30, and you have 15:30

How I figured when NFP cooling off period is over (normally):
int OneSecond = 1;
int OneMinute = 60*OneSecond;
int OneHour = 60*OneMinute;
int OneDay = 24*OneHour;
int NFP_EndTime = NFP_StartTime +
From NFP release until 4PM Central time +
Saturday and Sunday +
From 4PM CST Sunday to 11PM CST Sunday/00:00 Midnight EST Monday

So, NFP_EndTime = NFP_StartTime + ((8*OneHour)+(30*OneMinute)) + (2*OneDay) + (7*OneHour)

Once you know how your time is compared to server time, you should also be able to calculate your correct time. Note: For those that don't know, it's interesting to note that Monday Forex time starts on Sunday afternoon 4PM Central time, so when looking at days and you are in Central time zone:
Before 4PM Central, then the current Forex day is the same as the today's date/day you see on your calendar. After 4PM Central, then the current Forex day is the same as tommorrow's date/day you see on your calendar.

This is all I can comment on using and changing the time inside the indicator. For someone that knows how to fix the indicator so that the indicator uses server time and local time to figure when NFP is going to be released, so that the time can be done automatically, then that would be great.

The way MT4 counts the start time's bar and end time's bar for NFP, previous and current week showed me that these needed fixed so that the correct bars would be chosen for us to manually trade now, and hopefully/eventually autotrade.

Note: There is no EA built to work with these indicators. If someone can optimize the code further than I have, that would be great. I have put comments as much as I can into this latest NFP indicator, so please look at the code if you have questions.

I have thought about using an approach to move the start and end bars for NFP by a certain number of days (user input) to the left or right, so that the user wouldn't have to override the dates by putting in actual dates, but override the dates by shifting days. If someone knows how to get the date and time for NFP release into MT4, so that the user wouldn't have to do this, that would be great.

note:the indicators in this post have been removed because of problems found

Happy New Year

Don

Hi disbellj,

i thinks this thread can help you more.

http://www.forexfactory.com/showthread.php?t=123284

Don,

I dunno what d drama was, how ever this morning those indicators werent looking that bad.

I stayed up for like 18 hours Saturday from morning until after midnight the next night, to try and get the NFP indicator fixed. Since the market wasn't opened yet, I couldn't test the new NFP indicator. Once the market did open, I saw that I had done something wrong be because of the way NFP end was not updating correctly. I wanted the red box to begin drawing at the start of NFP cooling-off period, but not to draw the red line marking the NFP cooling-off period end until the actual end of the NFP cooling off period (bar whose closing time is at the end of NFP cooling-off period), then draw the green line when it's OK to start looking for a trade.

The end red rule and the green line are being drawn too soon, and somehow the orange line is affected too. Should be an easy fix, which I already have ideas about how i will fix the problem(s).

Don

Thank you Wartercooloer. I'll take a look when I get a chance. The files only being .ex4 doesn't allow me to learn anything about how to make my code more efficient. When I have a chance to go through 165 pages, I'll read it. If there is a specific post or something that you can link to, so I don't have to wade through them all, that would be great.

Thank you,

Don

Looking at the chart with the latest indicator, (which I have since removed since there was believed to be a glitch, which there may have not been a glitch in the latest indicator, since I had an old indicator on the chart, duh), it looks like if NFP was released Friday, and the NFP cooling-of period end was at the regular time (Sunday 11PM Central):

Would have sold two positions of EURUSD at FXDD server time 2009.01.05 10:30 for 1.3712, bought back one of the positions at 50 pips profit 1.3662, and the rest would be in profit at current price 1.3589, for a gain of 123 pips on the 2nd position (or remaining half of original position, however you'd rather look at it).

Would not made any trades for GBPUSD since all conditions weren't met.

Would not made any trades for AUDUSD since all conditions weren't met.

Would have bought two positions of USDCHF at FXDD server time 2009.01.05 10:30 for 1.0928, sold one of the positions at 50 pips profit 1.0978, and the rest would be in profit at current price 1.1059, for a gain of 127 pips on the 2nd position (or remaining half of original position, however you'd rather look at it).

Would have bought two positions of USDJPY at FXDD server time 2009.01.05 10:30 for 92.69, sold one of the positions at 50 pips profit 93.19, and the rest would be in profit at current price 93.19, for a gain of 50 pips on the 2nd position (or remaining half of original position, however you'd rather look at it).

At this time, Monday 01-05-09 11AM Central Time, 19:00 FXDD server time, if NFP was released on Friday, and we took trades as scheduled, would be 450 pips in profit, and still have half positions in those trades going, to make more profit. Not bad for the trades being half of a day old (12 hours to be exact), huh?

I would consider putting in money management (if an EA were written, my money management to pick number of lots, and then to lock 50 pips profit on the second position at 100 pips, lock 100 pips profit when profit at 150 pips, lock 150 pips profit when profit at 200 pips, etc., so that we would never lose more than 50 pips of profit from the high peak of profit)

Just wanted to let people see that quick profit can be possible, and to also ask, how can I find out when NFP is released? and does anybody know a way that NFP release can be programmed? Then the cooling-off period would shift with the NFP release time, the way it is coded.

Don

I am programming the NFP date thing. I do not see a need to enter the dates. It should either the first week or the second week.

I simply added a external variable that sets true or false for 1st week. If it falls on the first week, do this. If not, do that.

I think it would be very hard to get a user to enter the dates in that format.. Too much room for errors. True and false is easier.

Email me. I can send you the EA I created. It is just without the entry and exit code. Mine runs on daily charts.

Let's put a scenario together using my money management to pick number of lots to trade for each pair, coupled with Kevin's technique of taking off half the position at 50 pips take profit for the first half of position, while putting the remaining half of position at stop loss of break-even:

I have determined that I can't risk more than 8% per pair trading six pairs with 100 pips stop loss each, or I will get a margin call. But Kevin said (my understanding at least) to trade 2 trades for each pair, with 100 pips stop loss on each. If I put 200 pips stop loss into my OpenOffice calculator for each pair, I can risk 10% per pair. At $1,000 initial deposit, 10% risk per pair, with a stop loss of 200 pips for each pair, and tick value of $10 per pair, my money management chooses the following number of lots for each pair:
=(Account Balance x (RiskPercentPerPair/100)/(StopLoss x TickValue)
=(1000 x (10/100))/(200 x 10) (Note: tick value will change and not always be $10, so number of lots will change, as seen below in USD-based pairs)
=(1000 x 0.1)/2000
=100/2000
=0.05 standard lots, or 5 micro lots each pair, to begin trade.

EURUSD 50 pips take profit x $0.30 per pair (three micro lots = half of position) = $15
EURUSD 123 pips profit x $0.20 per pair (two micro lots = other half of position) = $24.60
USDCHF 50 pips take profit x $0.27 per pair (three micro lots = half of position) = $13.50
USDCHF 127 pips profit x $0.18 per pair (two micro lots = other half of position)= $22.86
USDJPY 50 pips take profit x $0.21 per pair (two micro lots = half of position) = $10.50
USDJPY 50 pips profit x $0.21 per pair (two micro lots = other half of position)= $10.50

Add all these together = $96.96 gained on a $1000 account. Percentage gained: 9.69% - in 12 hours. Now because my money management is set up the way it is, you'd get 9.69% gain on your account (given these trades) no matter what size the account. So if the account was $100,000, you'd still risk the same percent and still gain the same percent from these trades (of course, if you had $100,000 account size, that 9.69% gain in 12 hours would amount to around $9,690). And remember, we would've only traded three of the six pairs so far.

Don

ForexNewDude,

So you're saying that NFP is always released on a Friday at the same time, and if it's not the first week, it's the second? Cool. Yeah, a true/false input to just forward to the second week would be great and easy on the user. The time to use still is going to be trouble for some I know, because it is hard-wired into the code, and not a user input, and some people don't want to dig in code. However, I don't know how to fix this (yet) without someone having to go into the code to choose the correct hour, since one has to think about their broker's server time, their local time, and maybe even Eastern time that NFP is released, and calculate by themselves to figure out the hour and minute in their broker's MT4 that the lines SHOULD be, and fix the code - geared to their particular broker and local time.

Don

I have revised the new indicators, and I've implemented ForexNewDude's recommendation to have user input of 'Shift_NFPStart_OneWeek'. When set to true, it NFP_Start (start of NFP cooling off-period) and NFP_End (end of NFP cooling-off period) will be pushed forward one week exactly. This month, the user will want to set this to True, since NFP is being released the 2nd Friday this month.

Also, there is an input 'Shift_LastNFPStart_OneWeek', that when set to true, will shift last month's NFP_Start and NFP_End forward one week. Next month, the user will want to set this to True, since next month, when looking back at the graphics for this month, they will not be correct unless this is set to True.

I've also added my money management. If 'ShowComments' is True, then all sorts of relevant information is shown on-chart, including my money management discussed in previous posts. Also, Server time and Local time are displayed.

I've added TradeCount to this indicator, and I've added Alerts, but since I'm getting way too many Alerts (and therefore incorrect TradeCount), I've set Alerts to False - to not use alerts. But the conditions that are put into this portion of the indicator are the signals for buy, buy again, sell, sell again. ForexNewDude, this area may help you when working on your EA buy and sell conditions.

Indicator still could be improved, but since I don't know how to at this time, I'll leave it to someone else if they can fix the code to be shorter and neater (using functions, arrays, etc.)

Don

Some colored areas were not staying updated. That's now fixed.

Nope. NOT FIXED. latest "fixed" indicator removed.

I TRIED. if someone else wants to fix my stuff and post it, that would be great. I'm done for now.

Don

I think you done well Don. Personally Im really interested in the way this system is working. I have one minor request if I may...

Are the rules still the same? Or better still in a few bullet pionts can you jot them down for the sake of everyone. I read the trades you had and I was watching them too but AUDUSD was a trade according to what I saw. I use ODL.

It would be really helpful to summarise the rules in points, for me thats the case any way.
thanks for all the hard work.

dfxclub,

Sure, I'll post them again.

First, you must know what your broker's server time is when it's the first Friday of the month, at 8:30 AM Eastern time. I have Central time, so NFP is released at my time of 7:30 AM CST. I know my broker starts Friday at my time Thursday 4:00 PM CST. So on my charts, I'm looking for Friday's date, but the hour I need to look for "NFP Start" (where the vertical line should be drawn) is:
4PM-12AM (8 hours) + 12AM-7:30AM (7 hours 30 minutes)
... and 8 + 7:30 = 15:30...
...so I'm looking on my MT4 platform for Friday's date, at hour 15:30.

So now I know when the NFP cooling-off period starts. NFP cooling-off period ends at Midnight Monday EST/11PM Sunday CST. The times in my NFP indicator are based on my CST and my broker's (FXDD) Server time combination, so will use wrong Start time if not on CST (Central Standard time) or on a broker who has a different time than FXDD's MT4 does. It would be good if this wasn't the case, and the indicator looked at server time and local time, and made decisions automatically based on these, but I coded what I could.

The Sentinel indicator (I would recommend the latest one, which doesn't use NZD), tracks the pips gained from buying USDCHF,USDCAD, and USDJPY (the usd-based pairs), and selling EURUSD,GBPUSD, and AUDUSD (the non-usd-based pairs). This indicator shows "upside down" when placed on EURUSD,GBPUSD, and AUDUSD charts, and is intended to do so, so that an EA (if built) or if manually traded, it makes the user not have to think about direction of Sentinel on non-usd-based chart. When Sentinel indicator is placed on chart, it shows in a separate indicator window below the main chart window.

Rules for 1st buy (use M30 chart to go by the rules):
1. Time to start trading (or looking for trades) is after NFP cooling-off period ends (main chart).
2. Price must close above highest high of NFP cooling-off period (main chart).
3. Price must close above 336 SMA (main chart).
4. Sentinel must be above Sentinel moving average of 336 SMA (explained by Kevin to be 7-day moving average). (Note: This is in the separate indicator window. See Sentinel_indi_Don indicator, or Sentinel 3 indicator if you want to use/trade NZD (and if trading NZD, you'll want to use it when calculating the Sentinel also)

If all conditions are met above in 1-4, place an order for that pair. Risk 10% of the account when choosing number of lots to trade. When the price hits 50 pips above where you bought, remove half the number of lots you're trading on the pair. Move the stop loss on the remaining half position to where you bought at (which is 50 pips below where you are now). The 2nd half of the position is closed when price closes below the 336 SMA on main chart. So if the price goes at least 50 pips, you've made money and won't lose any money on the trade. If it doesn't get to 50 pips above where you bought, and you use Kevin's recommendation of 100 pips stop loss for both halves of the position, you end up losing 200 pips on this pair. So before you ever put on a trade on a pair, account for 200 pips stop loss on that pair, and it will give you correct amount of lots (my money management equation) to trade.

So you see (this is the short explanation):
1st Buy, Condition 1: Price closes ABOVE NFP cooling-off period's highest high
1st Buy, Condition 2: Price closes ABOVE 336 SMA
1st Buy, Condition 3: Sentinel is ABOVE 336 SMA in indicator window.

Exit if price closes BELOW 336 SMA on main chart. After the 1st Buy, any Buys after that must also include:
Buy Condition 4: Price closes ABOVE current week's highest high.
Buy Condition 5: Price closes ABOVE previous week's highest high.

The rules for sell on a pair are the exact opposite - switch ABOVE with BELOW, and BELOW with ABOVE in the above statements. If you tackle the time stuff first mentioned, understand the Sentinel I've explained above, and understand the way Kevin take's profit, then the system's rules are all in these last paragraphs.

The following is assuming the NFP was released the first week of the month, which it wasn't, so this is just for purpose of discussion for understanding:
On my AUDUSD chart, I show that WHEN the price closed above the NFP cooling-off period's highest high (2009.01.05 23:00 on FXDD MT4), the Sentinel was below the Sentinel_MA on that chart's indicator window, so no trade would be taken. Sentinel kept us out of a losing trade. Again, NFP is being released the second Friday this month, so no trades would be taken yet for January 2009, and this little discussion is only for learning (scenario if NFP would have been released the first week).

Don

Thanks alot Don

Much appreciated

You're welcome.

I'm looking at Kevin's NFP.txt from the first post, and the posts in between, and I still don't understand how he got number of lots with the math he provided. Does anybody else have a clue?

I just know if someone were to trade 2 "lots" (trades) per pair (look at them as two trades, with each one having a 100 pips stop loss), then at most, I could only risk 4% per trade, trading six pairs, with 2 trades each pair, all at once, and any more than that and I could get a margin call (and even at that, that's 4 x 12 = the percentage of the account at "risk" - although I wouldn't really expect to lose all the trades if going by the rules, so this risk is just for money management and worse-case scenario, which money management should be looking at when choosing amount of lots to trade IMO).

Don

Hi Kev, could you please sent me an examplea few chart of your system, about NFP rule i want to try for backtest to firm my trade,

Hi all,

anybody still interested in this method ?
I went through all the posts and adapted / modified the NFP/sentinel indicators and EA to Tradestation/Multicharts (without MM yet) because the backtesting capabilities of MT seem to be awful.
I backtested all currency pairs for the last 7 years:

The adapted strategy works very well and consistently for GBP/USD only.
All other currency pairs did not work at all.
Maybe this is due to its focus on GBP/USD while it was developed.
Can anybody confirm this ?

If anybody is interested in this, let me know.

Hi Pippiee,

Very interesting! Could you please post some of your charts from backtesting, and give some information on parameter values used?

It is very difficult to give a general answer without any specific information on patterns.

It is possible that you benefit for the GBPUSD because you benefit from information from the huge volume of EURUSD and USDJPY for predicting trends in USD, while some fringe crosses of the USD may be affected more by other factors than trends in USD. This would leave the GBPUSD as a reasonable choice with a fair amount of volume but not absolute top. It could be that trends in the GBP are related but somewhat lag behind both USD and EUR.

But you could also say that the system has serious limitations, and sooner or later these limitations wil become apparent for all pairs. It may just be a matter of chance why the GBPUSD has lasted positive so long. In other words, you may have been out for a statistical fishing expedition (when you look hard enough for something you will eventually find it but may be a fluke with no predictive value).

Without seeing more information on how the equiry curves of different pairs look like, and how consistent some of the equity growth is, this is all mute speculation.

Would be great to see some charts of yor work!

Hi Bubo,

i have attached EUR/USD, GBP/USD and USD/CHF.
Parameters:

- fixed trade size = 0.1 lot
- pessimistic comission (i dont have a spread in multicharts) of 4 pips per transaction (so this equals to a 8 pips spread/slippage per roundtrip)
- signal/ea parameters are the original ones (50 Profit Target, 100 SL)

The starting equity of 100.000 is of course way to high for a trade size of 0.1 lot.
Lets assume a starting equity of 10.000, the maximum DD is 1.500, which is 15 %.
Net profit is 8.000 which is not much in respect to the 7 year trading period, but remember, I always traded 0.1 lot.

I am curious to see how it will look like after implementing MM code.