Francesco’s Hammernegated strategy

• 06/27/2019 at 2:08 PM #101538
  25sonalgo

  http://www.prorealcode.com/prorealtime-trading-strategies/oil-10min-hammernegated-pattern-strategy/

  I refer to the strategy in the above link. It seems to be working well in recent months. But can someone please do a 200k bars test for us so so the community can further improve the code?

  

  25sonalgo

  Participant

  Topics: 1

  Replies: 2

  Been thanked: 0 times
  06/27/2019 at 2:09 PM #101539
  25sonalgo

  Credit to Francesco, the true legend!

  

  25sonalgo

  Participant

  Topics: 1

  Replies: 2

  Been thanked: 0 times
  06/28/2019 at 2:33 PM #101616
  robertogozzi

  I converted it to MTF (1-minute default TF):

  HammerNegated OIL 10min mtf

  // HammerNegated OIL 10min mtf // // https://www.prorealcode.com/prorealtime-trading-strategies/oil-10min-hammernegated-pattern-strategy/ // https://www.prorealcode.com/topic/francescos-hammernegated-strategy/ // defparam cumulateorders = false timeframe(10 minute,updateonclose) ONCE p = 36 //36 4 ONCE t = 0 //0 0 av1 = average[2,t](close) //2 2 av2 = average[2,t](close) //2 15 av3 = average[50,t](close) //50 30 ONCE x = 0.53 //0.53 0.50 bull = close> av1 and close >av2 and close > av3 bear = close <av1 and close <av2 and close < av3 ONCE n = 150 //150 95 ONCE m = 30 //30 85 hammerup = min(open[1],close[1]) > high[1]-(high[1]-low[1]) / 3//and timeok hammerupnegated = max(open,close) < min(open[1],close[1]) and abs(open-close) / (high-low) > x cs = hammerup and hammerupnegated and bear hammerdown = max(open[1],close[1]) < low[1]+(high[1]-low[1]) / 3 //and timeok hammerdownnegated = min(open,close) > max(open[1],close[1]) and abs(open-close) / (high-low) > x cl = hammerdown and hammerdownnegated and bull timeframe(default) if cs AND Not OnMarket then sellshort 1 contract at market set target pprofit n*averagetruerange[p] set stop ploss m*averagetruerange[p] endif if cl AND Not OnMarket then buy 1 contract at market set target pprofit n*averagetruerange[p] set stop ploss m*averagetruerange[p] endif //TRAILING STOP ONCE TGL = 27 //27 30 ONCE TGS = 27 //27 18 if not onmarket then MAXPRICE = 0 MINPRICE = close PREZZOUSCITA = 0 ENDIF if longonmarket then MAXPRICE = MAX(MAXPRICE,close) if MAXPRICE-tradeprice(1)>=TGL*pointsize then PREZZOUSCITA = MAXPRICE-TGL*pointsize ENDIF ENDIF if shortonmarket then MINPRICE = MIN(MINPRICE,close) if tradeprice(1)-MINPRICE>=TGS*pointsize then PREZZOUSCITA = MINPRICE+TGS*pointsize ENDIF ENDIF if onmarket and PREZZOUSCITA>0 then EXITSHORT AT PREZZOUSCITA STOP SELL AT PREZZOUSCITA 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 58

  // HammerNegated OIL 10min mtf // // https://www.prorealcode.com/prorealtime-trading-strategies/oil-10min-hammernegated-pattern-strategy/ // https://www.prorealcode.com/topic/francescos-hammernegated-strategy/ // defparam cumulateorders = false timeframe(10 minute,updateonclose) ONCE p            = 36                         //36             4 ONCE t            = 0                          //0              0 av1               = average[2,t](close)        //2              2 av2               = average[2,t](close)        //2              15 av3               = average[50,t](close)       //50             30 ONCE x            = 0.53                       //0.53           0.50 bull              = close> av1 and close >av2 and close > av3 bear              = close <av1 and close <av2 and close < av3 ONCE n            = 150                        //150            95 ONCE m            = 30                         //30             85 hammerup          = min(open[1],close[1]) > high[1]-(high[1]-low[1]) / 3//and timeok hammerupnegated   = max(open,close) < min(open[1],close[1]) and abs(open-close) / (high-low) > x cs                = hammerup and hammerupnegated and bear hammerdown        = max(open[1],close[1]) < low[1]+(high[1]-low[1]) / 3 //and timeok hammerdownnegated = min(open,close) > max(open[1],close[1]) and abs(open-close) / (high-low) > x cl                = hammerdown and hammerdownnegated and bull timeframe(default) if cs AND Not OnMarket then    sellshort 1 contract at market    set target pprofit n*averagetruerange[p]    set stop   ploss   m*averagetruerange[p] endif if cl AND Not OnMarket then    buy 1 contract at market    set target pprofit n*averagetruerange[p]    set stop   ploss   m*averagetruerange[p] endif //TRAILING STOP ONCE TGL = 27     //27      30 ONCE TGS = 27     //27      18 if not onmarket then    MAXPRICE     = 0    MINPRICE     = close    PREZZOUSCITA = 0 ENDIF if longonmarket then    MAXPRICE = MAX(MAXPRICE,close)    if MAXPRICE-tradeprice(1)>=TGL*pointsize then       PREZZOUSCITA = MAXPRICE-TGL*pointsize    ENDIF ENDIF if shortonmarket then    MINPRICE = MIN(MINPRICE,close)    if tradeprice(1)-MINPRICE>=TGS*pointsize then       PREZZOUSCITA = MINPRICE+TGS*pointsize    ENDIF ENDIF if onmarket and PREZZOUSCITA>0 then    EXITSHORT AT PREZZOUSCITA STOP    SELL      AT PREZZOUSCITA STOP ENDIF

   

  

  robertogozzi

  Moderator

  Topics: 69

  Replies: 16469

  Been thanked: 3643 times
  06/29/2019 at 6:42 AM #101634
  solar

  Thank you robertogozzi for your modification on the code. Interesting result, I have several questions regarding your version of this strategy. For the sake of tidiness and legibility I make my questions in point form:

  1. whats is the difference between  av1 and av2 (line 10 and 11) ( Is it a typo or sth else?)
  2. would it make more sense for “hammerupnegated   = max(open,close) <= min(open[1],close[1]) “(line 19 and same for line 22) as i guess the open of the current bar should always equal the close of previous bar ?(except gaps after session break)
  3. Do the parameters like p,t,x,m,n c come from optimization? (any WF test done on these parameters?)
  4. Can you or any others help to test it for 200k bars?
  5. Have you tried other type of trailing stop or exit strategies?

  Meanwhile, I am also interested on the original code’s performance on 15min TF for 200k bar. Would you mind posting the backtest result here for my reference?

  Cheers

  

  solar

  Participant

  Topics: 13

  Replies: 57

  Been thanked: 2 times
  06/29/2019 at 8:35 AM #101638
  robertogozzi

  1. it’s not a typo, it’s the optimization tool that said it was the same (I kept all the three not to alter the original logic)
  2. I kept the same code as the original version, but you can try different settings
  3. Yes, I optimized all variables due to the fact that I was using a MTF approach, which is different from using a single TF
  4. Yes, I attach both 100k and 200k performance on the original code, but with a 15-minute TF as you asked (oddly enough there’s no difference in performance and dates on CL, while there are on DAX or Eur/Usd). I never use 200k tests!
  5. No, I just kept the original code, just changing values or replacing some constants with variables. I also reintroduced the last two lines, but placed them elsewhere to avoid SL & TP be altered every candle when onmarket

   

   

  

  robertogozzi

  Moderator

  Topics: 69

  Replies: 16469

  Been thanked: 3643 times
  07/02/2019 at 2:42 AM #101791
  solar

  I can see that very often the profit is secured after hours the position is opened. It somehow implies that “hammer negated” is not a very precise/ accurate signal, otherwise the market would always trend in direction favourable to the position.

  Would you mind sharing your thoughts on this strategy?

  

  solar

  Participant

  Topics: 13

  Replies: 57

  Been thanked: 2 times
  07/02/2019 at 3:15 AM #101792
  25sonalgo

  Thank you @robertogozzi! I found that the original code written by @Francesco also works pretty well on DJI

  Can robertogozzi please help the original code on DJI on 15min TF for 200k bar and share with us the results!

  Many thanks!

  

  25sonalgo

  Participant

  Topics: 1

  Replies: 2

  Been thanked: 0 times
  07/02/2019 at 9:53 AM #101800
  robertogozzi

  Performance for DJI-15min with 2.4 spread.

   

  

  robertogozzi

  Moderator

  Topics: 69

  Replies: 16469

  Been thanked: 3643 times
  07/02/2019 at 10:07 AM #101803
  solar

  IMO it works better if TGL or TGS are above 40 as the indices price is around 4times of that of oil.

  

  solar

  Participant

  Topics: 13

  Replies: 57

  Been thanked: 2 times
  07/02/2019 at 3:19 PM #101818
  jannus

  Hola Roberto. He copiado el código del Hammarnegated pero no me funciona.

  Hay algún fallo o algo que haya que hacer? yo hice un copy paste. Gracias.

   

  

  jannus

  Participant

  Topics: 3

  Replies: 11

  Been thanked: 0 times
  07/02/2019 at 3:44 PM #101819
  robertogozzi

  jannus wrote:

  Hola Roberto. He copiado el código del Hammarnegated pero no me funciona.

  Hay algún fallo o algo que haya que hacer? yo hice un copy paste. Gracias.

  Please speak English on the english forum.

  If it does absolutely nothing, then there must be an error in the backtest periods.

  Doesn’t it report any error?

  It works perfectly if copied as is. Should you have also copied line numbers please remove them.

  

  robertogozzi

  Moderator

  Topics: 69

  Replies: 16469

  Been thanked: 3643 times
  07/02/2019 at 4:45 PM #101830
  robertogozzi

  I attach the .ITF file to be imported.

  

  robertogozzi

  Moderator

  Topics: 69

  Replies: 16469

  Been thanked: 3643 times
  07/03/2019 at 3:57 PM #101887
  Paul

  interesting results, but I think there’s a point to take in consideration.

  Although very high win-chance and nice results, there’s no fixed stoploss.

  The highest MAE, is about the same as a 2.5-3% stoploss. (1 minute 200k bars us crude)

  So perhaps it’s good to have a fixed % set and then you get out. 3% is big seeing a market running against your trade which can happen apparently.

  The shorter the timeframe, the lower the stoploss should be?

  If switched that code to 10min, it goes up al the way to 11,75% loss on a trade and still a win-chance of more then 90%.

  pp=(positionperf*100) sl=2.00 //stoploss in % if pp < -sl then sell at market exitshort at market endif graph pp

  1 2 3 4 5 6 7 8 9

  pp=(positionperf*100) sl=2.00 //stoploss in %   if pp < -sl then sell at market exitshort at market endif   graph pp

   

  1 user thanked author for this post.

  solar

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
  07/05/2019 at 2:49 AM #101999
  solar

  Thanks Paul for your advice.

  It certainly makes me feel more comfortable by applying a fixed percentage stoploss, and of course, the trade off is a little bit lower win rate and profit.

  The next thing that comes to my mind is increasing the bet size after losses, since the maximum consecutive loser is just 2 or 3. I know each trade occur independently, however, statistically speaking, i think it maybe worth doing so. What do you think?

  

  solar

  Participant

  Topics: 13

  Replies: 57

  Been thanked: 2 times
  07/06/2019 at 6:21 PM #102054
  Paul

  Hi fatlung

  In my opinion it’s not a good strategy, only maybe if you have 1000 of trades to test.

  Example is an other strategy, nice run up and then after unexpected consecutive losses it gained momentum again.

  In the backtest it never had so many consecutive losses. To take those losses is bad enough without doubling down.

  1 user thanked author for this post.

  solar

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
• Author
  Posts