MFI CROSS strategy to test and improve on Wall Street 6 minutes

• 03/12/2018 at 9:29 AM #64947
  DEIO

  HI,

  Here an idea to use MFI that I consider very interesting.

  The strategy (ONLY LONG) waits for a cross of a MFI fast and slow.

  Moreover the average must always go up in the last period.

  The system perform well on wall street (6 minutes)  and has a little DD.

  It would be nice if anyone can tests it on a frame of time greater than one year (as I can only do).

  Let me know….

  bye

  MFI CROSS

  //----------------------------------------------------------------------------------------------- //WALL STREET // 6 MIN. DEFPARAM cumulateorders = FALSE CONDAVG = 1 AVG = Average[S](close) FOR C = 1 TO F IF AVG[C-1] > AVG[C] THEN IF CONDAVG = 1 THEN CONDAVG = 1 ENDIF ELSE CONDAVG = 0 ENDIF NEXT TIMEWORK = 000000 TIMESTOP = 240000 F = 20 S = 200 MFIFAST = MoneyFlowIndex[F] MFISLOW = MoneyFlowIndex[S] BASEMFI = MFISLOW[0] > MFISLOW[F] AND MFISLOW[0] > MFISLOW[ROUND(F/2)] AND MFISLOW[0] > MFISLOW[ROUND(F/8)] CONDMFIL = MFIFAST[0] CROSSES OVER MFISLOW[0] //TRAILINGSTART = 50 * PIPSIZE TRAILINGSTART = (close/((S*2)+(F*2))) * PIPSIZE AMNT = TRAILINGSTART * POINTVALUE * COUNTOFPOSITION //GRAPH AMNT SIZEBUY = 1 IF CURRENTTIME >= TIMEWORK AND TIME <= TIMESTOP THEN if not onmarket THEN IF CONDMFIL AND BASEMFI AND CONDAVG THEN BUY sizebuy SHARES AT highest[F](HIGH) STOP endif ENDIF endif IF ONMARKET THEN PERFP = CLOSE - TRADEPRICE PERFN = (PERFP / TRAILINGSTART) TRAIL = (TRAILINGSTART/PERFN) *POINTVALUE * COUNTOFPOSITION IF PERFP > TRAILINGSTART THEN SET STOP $TRAILING TRAIL ELSE SET STOP $TRAILING AMNT ENDIF IF OPENTIME > 213000 AND DayOfWeek = 5 THEN SELL AT MARKET ENDIF ENDIF

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

  //----------------------------------------------------------------------------------------------- //WALL STREET // 6 MIN.   DEFPARAM cumulateorders = FALSE   CONDAVG = 1   AVG = Average[S](close)   FOR C = 1 TO F IF AVG[C-1] > AVG[C] THEN IF CONDAVG = 1 THEN CONDAVG = 1 ENDIF ELSE CONDAVG = 0 ENDIF NEXT     TIMEWORK = 000000 TIMESTOP = 240000 F = 20 S = 200   MFIFAST = MoneyFlowIndex[F] MFISLOW = MoneyFlowIndex[S]   BASEMFI = MFISLOW[0] > MFISLOW[F] AND MFISLOW[0] > MFISLOW[ROUND(F/2)] AND MFISLOW[0] > MFISLOW[ROUND(F/8)] CONDMFIL =  MFIFAST[0] CROSSES OVER  MFISLOW[0]     //TRAILINGSTART = 50 * PIPSIZE TRAILINGSTART = (close/((S*2)+(F*2))) * PIPSIZE   AMNT = TRAILINGSTART * POINTVALUE * COUNTOFPOSITION //GRAPH AMNT   SIZEBUY = 1   IF CURRENTTIME >= TIMEWORK AND TIME <= TIMESTOP THEN if not onmarket THEN IF CONDMFIL AND BASEMFI   AND CONDAVG  THEN BUY sizebuy SHARES AT highest[F](HIGH) STOP endif ENDIF endif   IF ONMARKET THEN PERFP = CLOSE - TRADEPRICE PERFN = (PERFP / TRAILINGSTART) TRAIL = (TRAILINGSTART/PERFN) *POINTVALUE * COUNTOFPOSITION   IF PERFP > TRAILINGSTART THEN SET STOP $TRAILING TRAIL ELSE SET STOP $TRAILING AMNT ENDIF   IF OPENTIME > 213000  AND DayOfWeek = 5 THEN SELL AT MARKET ENDIF   ENDIF

   

  2 users thanked author for this post.

  Gianluca, GraHal

  

  DEIO

  Participant

  Topics: 3

  Replies: 21

  Been thanked: 5 times
  03/13/2018 at 10:03 AM #65169
  Nicolas

  Even with in 200k bars, I don’ have a longer history than you. Tested with 2 points spread, the strategy shares the same equity curve than your test in tick mode. Did you try to change the trailing stop instruction with the code versions I made?

  https://www.prorealcode.com/blog/learning/trailing-stop-max-favorable-excursion-mfe/

  https://www.prorealcode.com/blog/trading/complete-trailing-stop-code-function/

  

  Nicolas

  Keymaster

  Topics: 81

  Replies: 20828

  Been thanked: 4741 times
  03/13/2018 at 11:44 AM #65187
  DEIO

  hi Nicolas,

  It is true….. I started from your code about trailng stop… but then I tried to use the feature of PRT (using $trailing in this case)

  As you can see I tried to manage o sort of max favorable excursion identifying the explosion bar (not so frequent..)

  The reason why I didn’t use your code is why I feel more comfortable when I can see the line of stop loss, which I can’t see

  using the code which calculate an invisible stop loss.

  Maybe you consider this approach a little bit strange but It’s a problem of my mind !!!  🙂

  Regarding the strategy in a fist step I was very happy to see such results but it’s would be nice can see a larger frame of time as I said before.

  The best thing I noticed about the strategy is that during the strong market fall (as recently happened) the strategy remain out of market correctly

  and I think it’s a good quality notso banal …

  Regarding DD it’s quite low compared to the gains.

  Let me know if you intend to publish it or not.

  Thx for any suggestion.

  bye

   

   

  

  DEIO

  Participant

  Topics: 3

  Replies: 21

  Been thanked: 5 times
  03/13/2018 at 12:42 PM #65196
  GraHal

  @DEIO did you try the Nicolas TS Code on the Link below? You do get a Stop Line moving up / down with this code.

  https://www.prorealcode.com/blog/trading/complete-trailing-stop-code-function/

  I’m a tad confused, I got 10 months history if I recall correctly with 100,000 bars so why didn’t Nicolas get 20 months or so with 200,000 bars??

  Cheers
  GraHal

  

  GraHal

  Participant

  Topics: 204

  Replies: 8399

  Been thanked: 2041 times
  03/13/2018 at 2:09 PM #65223
  DEIO

  HI GraHal,

  I modified  the code as suggested and results (on 100K bars) are good but with less protection during the bear market phases.

  I have to test it on demo in order to check if actually I can see the stop loss line moving (but I quite sure that is not so – sorry if I insist on it – maybe I am wrong…)

  The reason why Nicolas couldn’t see on 200k is unknown to me, I can’t understand it as  I can see only 100k (if I have correctly understood your affermation).

  This the code modified:

  //----------------------------------------------------------------------------------------------- //----------------------------------------------------------------------------------------------- //WALL STREET // 6 MIN. DEFPARAM cumulateorders = FALSE CONDAVG = 1 AVG = Average[S](close) FOR C = 1 TO F IF AVG[C-1] > AVG[C] THEN IF CONDAVG = 1 THEN CONDAVG = 1 ENDIF ELSE CONDAVG = 0 ENDIF NEXT TIMEWORK = 000000 TIMESTOP = 240000 F = 20 S = 200 MFIFAST = MoneyFlowIndex[F] MFISLOW = MoneyFlowIndex[S] BASEMFI = MFISLOW[0] > MFISLOW[F] AND MFISLOW[0] > MFISLOW[ROUND(F/2)] AND MFISLOW[0] > MFISLOW[ROUND(F/8)] CONDMFIL = MFIFAST[0] CROSSES OVER MFISLOW[0] //TRAILINGSTART = 50 * PIPSIZE TRAILINGSTART = ROUND((close/((S*2)+(F*2)))) * PIPSIZE TRAILINGSTEP = ROUND(TRAILINGSTART / 4) //AMNT = TRAILINGSTART * POINTVALUE * COUNTOFPOSITION //GRAPH AMNT SIZEBUY = 1 IF CURRENTTIME >= TIMEWORK AND TIME <= TIMESTOP THEN if not onmarket THEN IF CONDMFIL AND BASEMFI AND CONDAVG THEN BUY sizebuy SHARES AT highest[F](HIGH) STOP endif ENDIF endif //reset the stoploss value IF NOT ONMARKET THEN newSL=0 ENDIF //manage long positions IF LONGONMARKET THEN //first move (breakeven) IF newSL=0 AND close-tradeprice(1)>=trailingstart*pipsize THEN newSL = tradeprice(1)+trailingstep*pipsize ENDIF //next moves IF newSL>0 AND close-newSL>=trailingstep*pipsize THEN newSL = newSL+trailingstep*pipsize ENDIF ENDIF IF newSL>0 THEN SELL AT newSL STOP ENDIF

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57

  //----------------------------------------------------------------------------------------------- //----------------------------------------------------------------------------------------------- //WALL STREET // 6 MIN. DEFPARAM cumulateorders = FALSE CONDAVG = 1 AVG = Average[S](close) FOR C = 1 TO F IF AVG[C-1] > AVG[C] THEN IF CONDAVG = 1 THEN CONDAVG = 1 ENDIF ELSE CONDAVG = 0 ENDIF NEXT TIMEWORK = 000000 TIMESTOP = 240000 F = 20 S = 200 MFIFAST = MoneyFlowIndex[F] MFISLOW = MoneyFlowIndex[S] BASEMFI = MFISLOW[0] > MFISLOW[F] AND MFISLOW[0] > MFISLOW[ROUND(F/2)] AND MFISLOW[0] > MFISLOW[ROUND(F/8)] CONDMFIL = MFIFAST[0] CROSSES OVER MFISLOW[0] //TRAILINGSTART = 50 * PIPSIZE TRAILINGSTART = ROUND((close/((S*2)+(F*2)))) * PIPSIZE TRAILINGSTEP = ROUND(TRAILINGSTART / 4) //AMNT = TRAILINGSTART * POINTVALUE * COUNTOFPOSITION //GRAPH AMNT SIZEBUY = 1 IF CURRENTTIME >= TIMEWORK AND TIME <= TIMESTOP THEN if not onmarket THEN IF CONDMFIL AND BASEMFI AND CONDAVG THEN BUY sizebuy SHARES AT highest[F](HIGH) STOP endif ENDIF endif   //reset the stoploss value IF NOT ONMARKET THEN newSL=0 ENDIF   //manage long positions IF LONGONMARKET THEN //first move (breakeven) IF newSL=0 AND close-tradeprice(1)>=trailingstart*pipsize THEN newSL = tradeprice(1)+trailingstep*pipsize ENDIF //next moves IF newSL>0 AND close-newSL>=trailingstep*pipsize THEN newSL = newSL+trailingstep*pipsize ENDIF ENDIF IF newSL>0 THEN SELL AT newSL STOP ENDIF

  Moderator’s edit: code edited to apply the correct PRT code format, please use “insert PRT code” button for future occurences (last one at end of first line of tools in message editor), thanks

  

  DEIO

  Participant

  Topics: 3

  Replies: 21

  Been thanked: 5 times
  03/13/2018 at 2:18 PM #65229
  GraHal

  I’ll give it a go later and let you know any findings.

  Due to //TRAILINGSTART = 50 * PIPSIZE you won’t see the stop line even or any movement of the stop Line until price has moved 50 points into profit. I guess you knew that?

  Also the stop line only gets a chance to move every TF period (6 mins in this case).

  Did you optimise //TRAILINGSTART = 50 * PIPSIZE?

  Maybe Nicolas will be along soon to tell us why he couldn’t get any more History  than our 100k bars?

  

  GraHal

  Participant

  Topics: 204

  Replies: 8399

  Been thanked: 2041 times
  03/13/2018 at 2:49 PM #65236
  DEIO

  HI GraHal,

  sorry to appear so unskilled…:)  anyway you are right about the stop line which will appear only over a trailingstart points of gain and only

  afer the code is proccessed every x minutes and the palcing of the sell order stop …     🙂

  Yes I tried to optimize trailingstart value according to the close of the instrument which can change a lot during the years, so 50 points

  on wall street NOW means a certain risk, while 2 years ago a different one.

  The way I did can seem a little strange but it’s enough find a simply way like – for example-  TRAILINGSTART = ROUND(close/400) * PIPSIZE

  .. waiting for Nicolas tests on 200k…..  (maybe could be troubles about the availability of the MFI indicator when tests are made on data older than 1 years?)

  thx to all

   

  

  DEIO

  Participant

  Topics: 3

  Replies: 21

  Been thanked: 5 times
  03/13/2018 at 3:03 PM #65239
  GraHal

  No problemo DEIO … I am unskilled also compared to many on here! 🙂

  Ah yes MFI needs volume and volume has prob only been available via IG on DJI since 12 months as you say … ah well we solved our own conundrum!  Well done to us! 🙂

   

  

  GraHal

  Participant

  Topics: 204

  Replies: 8399

  Been thanked: 2041 times
  03/14/2018 at 5:01 PM #65297
  DEIO

  hi all,

   

  I am making other test on the strategy and I would take advantage from your experience and the question is:

  if the statement TRADEPRICE(1) contains the price of the last trade executed…..

  what contains TRADEPRICE(0)  ???

  I ask that because I make an error editing my code using TRADEPRICE(0) instead of TRADEPRICE(1)

  and I see strange results (no bad results I mean..)   …. so I wonder what contains TRADEPRICE(0)….

  could you help me ??

  thx

  

  DEIO

  Participant

  Topics: 3

  Replies: 21

  Been thanked: 5 times
  03/14/2018 at 5:27 PM #65298
  GraHal

  TradePrice(0) is the price of the most recent trade … which may even still be trading (not yet ended).

  TradePrice(1) is the price of the trade before the most recent trade (so 2 trades ago).

  

  GraHal

  Participant

  Topics: 204

  Replies: 8399

  Been thanked: 2041 times
  03/14/2018 at 8:51 PM #65300
  JR1976

  DO you test in Demo account ???

  seems interest code , but I think could be more  secure with the stop loss set up in the code  ….

  FOllow the backtest with 0.2 % and with 1.7 %   Stop loss

   

  

  JR1976

  Participant

  Topics: 16

  Replies: 88

  Been thanked: 8 times
  03/15/2018 at 2:06 AM #65320
  manel

  Actually the lack of complete 200k bar history is a known bug in PRT. I’ve raised it a couple of years back with them but it hasn’t been addressed. And also, there is volume history on the Dow going back further than 12mths, in fact it starts on Jul 2016. The reason there is no more data on 200k bars than 100k bars above is because of a weird bug in PRT that only calculates the full 200k bars based on “standard” time units ie those that you get on the drop down menu. At least that’s my theory anyway. See below.

  As you see from the first image there is no more data on the 100k chart compared to 200k chart for the 6m TF

  On the second image there is twice as much data in 200k units for the 5m period as there is for 100k units – as expected (its the same with 10m tf as well)

  Also, I don’t think there is any correlation between the units shown in the drop down menu and the actual time scale on the charts.

  Eg. for the 6m tf – 6m x 200,000 = 1.2m / 60/ 24 = 833 days /365 = 2.28 years ie 2yrs and 3mths approx. And as you can see from the history we get only around half that. In fact in reality it should be around 15% more history than even that figure due to the fact that Saturday has no units. I think you just have to accept this limitation in PRT though it’s a shame that if the data is there and is not being made available.

   

  1 user thanked author for this post.

  GraHal

  

  manel

  Participant

  Topics: 4

  Replies: 259

  Been thanked: 121 times
  03/15/2018 at 7:21 AM #65324
  DEIO

  HI JR1976,

  I’m  cheking several version of the strategy(on demo) in order to find the best one.

  Anyway I keep in consideration your suggestion and I’m preparing the proper code to manage the SL with the code provided by Nicolas.

  Rusults with the version I posted (in attached) are quite good even on DD as your result in % on SL.

  I’ll let you know further develops..

  thx for you time.

  bye

   

  

  DEIO

  Participant

  Topics: 3

  Replies: 21

  Been thanked: 5 times
  03/15/2018 at 8:46 AM #65332
  DEIO

  Hi manel

  Thx a lot for your analisys

  I imagined What yo are saying  about PRT  lackness of data and I really  don’t know  if they can fix such  issue, but  I  agree with  your disappoitment….

  Obviously  this way traders could fly away on other  platforms…….

  The again

  Bye

  

  DEIO

  Participant

  Topics: 3

  Replies: 21

  Been thanked: 5 times
  03/15/2018 at 10:56 AM #65352
  GraHal

  manel wrote:

  Actually the lack of complete 200k bar history is a known bug in PRT.

  Your full explanation was so good Manel, I raised it on the Platform Support Forum under your Quote … hope you don’t mid?!

  This is one of my new ways to be controlled & constructive! 🙂

  I can’t ignore Issues (I know you can’t either) else I feel like giving up altogether!

  Thank You
  GraHal

  

  GraHal

  Participant

  Topics: 204

  Replies: 8399

  Been thanked: 2041 times
• Author
  Posts