The Stochastic Indicator

Discussion in 'Technical Analysis' started by jack hershey, Feb 17, 2003.

Thread Status:
Not open for further replies.
  1. See notes to right on chart.

    after trade 2 there was mostly congestion. when you have made several multiples of intial capital you can trade the congestion off the MACD. i will detail it out on expanded charts.

    I am a CERT start up this weekend so I will catch as catch can here.
     
    #511     Mar 28, 2003
  2.  
    #512     Mar 28, 2003
  3. dawg

    dawg

    your comments accurately convey what i was 'feeling' during the trades, but somewhat hesitant to act upon....now i have a set of pictures i can think of when faced with a similar situation. while today was not a profitible $ wise, it was definitlely an addition to my knowledge that can be drawn upon in the future. i feel like these are all steps in the right direction.

    thanks for all your help. enjoy the weekend.

    dawg
     
    #513     Mar 28, 2003
  4. vorzo

    vorzo

    Here are the backtesting results for Jack's methodology.

    Size traded: 1 contract
    Commissions: $5/round trip (IB's rounded up)
    Time frame: 5 min
    Time of day: 9:30-12:00 EST except for Mon 9:30-16:00 (the hours that I trade).

    Entry: the signals used for variations of the method are indicated in the first column, and the entry price was set to the midpoint of the entry bar (average of open and close) to approximate live entry as soon as the line crosses rather than on bar close. I've included several results for entry on bar close for comparison.

    Exit: on close of signal bar, or at indicated stop.

    Settings: Stochastic 14,1,3, MACD 5,13,6
    Software used: WealthLab


    First I tested entry on fast line in and MACDhist +-0.4 /exit on fast line out with different point stops (and also a stop on low of bar), on the most recent expired contract (ES03H). All entries after 9:30. Entry after 9:45 - synch period - included as it has better results then entry after 9:30. Two different triggers were tested 20/80 and 25/75.

    Contract Stop Stoch P&L Win/Loss(max consec) Max Ddown
    ES03H
    in:fast line 1 pt 20/80 $8,950 135/138 (8/10) -$444
    and MACDhist 25/75 $8,953 123/159 (6/10) -$700
    out:fast line
    1.5 pt 20/80 $9,595 145/113 (8/9) -$497
    25/75 $9,629 130/129 (7/9) -$588

    2 pt 20/80 $9,177 146/110 (8/8) -$598
    25/75 $9,358 131/121 (7/8) -$592

    [in:after 9:45] 2 pt 20/80 $9,087 135/85 (8/8) -$452
    25/75 $9,456 121/94 (8/8) -$786

    3 pt 20/80 $9,859 145/100 (8/8) -$520
    25/75 $9,995 130/110 (8/8) -$560

    bar low 20/80 $8,741 134/134 (5/8) -$481 (-$130 largest loser)
    25/75 $9,351 122/139 (6/8) -$481 (-$130 largest loser)

    From these results it appears that the best results (highest return, highest win/loss ratio, lowest drawdown) were obtained with a 3 pt stop, and a 25/75 trigger. A 3 pt stop 20/80 trigger, however, offers a better win/loss ratio and a smaller drawdown, although total return is slightly less. Results for 2 pt stop were the next best thing, and when tested over 5 contracts the 2 pt stop was better than the 3 pt (results not shown).

    Next, I tested entry on both lines in, or on fast line in/no MACD hist.


    Contract Stop Stoch P&L Win/Loss(max consec) Max Ddown

    in:both lines 2 pt 20/80 $4,856 111/112 (10/9) -$824
    out:fast line 25/75 $4,765 116/133 (7/11) -$1,360

    in:fast line 1 pt 20/80 $9,946 177/175 (7/10) -$490
    exit fast line 2 pt 20/80 $9,807 174/173 (7/10) -$490

    Waiting for both lines to get in misses out on signals with almost half the return of entry on fast line, without much improvement in win/loss ratio. Drawdown is significantly higher too.

    Next, I tested entry on close of signal bar (tested volume greater than 5,000 as entry condition to see if it makes a difference).

    Contract Stop Stoch P&L Win/Loss(max consec) Max Ddown

    in:fast line
    and MACDhist 2 pt 20/80 $1,041 93/168 (5/10) -$2,224
    on close 25/75 +$1,857 92/163 (5/10) -$2,421
    out:fast line

    in:fast line 2 pt 80/20 +$480 62/93 (7/7) -$1,722
    and MACDhist 75/25 +$567 55/99 (4/7) -$2,265
    and vol>5000
    on close
    out:fast line

    Return is down dramatically because of the late entry. Entry only on vol>5,000 misses out on signals with lower return.

    Next I tested entries a la dawg, on the break of the signal bar.

    Contract Stop Stoch P&L Win/Loss(max consec) Max Ddown

    in:fast line 2 pt 20/80 -$437 61/109 (4/8) -$1,667
    and MACDhist 25/75 -$172 63/113 (3/8) -$2,538
    on hi/lo break 3 pt 20/80 +$262 62/99 (7/7) -$1,601
    out:fast line

    in:fast 80/20 3 pt 80/20 +$967 45/69 (7/9) -$1,843
    and slow 75/25
    and MACDhist
    on hi/lo break
    out:fast 75/25

    Returns are very low most likely due to even later entry (above close of bar).

    Next, I tested a variation of the exit signal - on MACD cross instead of fast line out.

    Contract Stop Stoch P&L Win/Loss(max consec) Max Ddown

    in:fast line 2 pt 80/20 +$7,190 114/114 (6/6) -$690
    and MACDhist 75/25 +$7,760 122/127 (7/6) -$640
    out:MACD cross

    in:both lines 2 pt 80/20 +$3,762 83/90 (5/6) -$996
    and MACDhist 75/25 +$3,731 97/118 (5/9) -$1,266
    out:MACD cross

    Returns and win/loss are slightly worse than exit based on fast line out, so will stick to the fast line signal.

    Since the best returns, highest win/loss ratio and lowest drawdown were obtained with:
    - entry after 9:45 on fast line in and MACDhist >< +-0.4, as soon as line crosses
    - exit on fast line out, on close of bar
    - 2 pt stop
    this combination was tested 5 contracts back (for some reason, software failed to retrieve earlier contracts data from Quote.com servers - will try it at a different time).

    Contract Stop Stoch P&L Win/Loss(max consec) Max Ddown


    ES03M to date 2 pt 20/80 $1,136 23/21 (6/3) -$397
    25/75 $1,290 23/24 (4/4) -$412

    ES02Z 2 pt 20/80 $10,283 146/106 (6/5) -$545
    25/75 $10,250 123/122 (6/5) -$575

    ES02U 2 pt 20/80 $15,612 148/97 (11/4) -$650
    25/75 $17,251 147/104 (8/4) -$725

    ES02M 2 pt 20/80 $9,175 128/112 (5/6) -$461
    25/75 $9,047 116/122 (6/6) -$478


    ES02H 2 pt 20/80 $5,038 96/80 (11/4) -$466
    25/75 $6,015 90/92 (7/6) -$730

    ES01Z 2 pt 20/80 $3,805 89/105 (7/5) -$648
    25/75 $2,663 87/124 (5/7) -$701

    As you can see results ar consistent with a peak return on Sept 2002 contract, and lower returns on older contracts.
    Roughly, 250 trades are taken for a given contract, which comes to 3 trades/morning and 7 trades/full trading day.

    The results indicate that the method is profitable historically, which gives one confidence in the method when trading it.
    If you remember that these results are for mornings only (and one full day) than they are even more impressive!

    Contract Stop Stoch P&L Win/Loss(max consec) Max Ddown

    ES03H 9:30-4:00 2 pt 80/20 $16,783 272/203 (10/8) -$800
    75/25 $18,266 240/221 (6/7) -$800


    Of course, live results will differ from backtested due to differences in entry prices, false signals that do not show up on the historical charts, rationalization of signals etc.
    Better exits may be obtained in live trading since in backtesting the exit was on bar close.
    Also, using sequences, 1 min chart prorating, volume, channels, wash trades and all the other refinements that Jack is kindly teaching us may greatly improve returns and win/loss ratio.

    Hope this is helpful. If anyone is interested, I'll post the WealthLab script that I used.
     
    #514     Mar 31, 2003
  5. vorzo

    vorzo

    Formatting is gone, so if it's hard to read here is an excel table.
     
    #515     Mar 31, 2003
  6. Vorzo,
    You did a great job ! If i am not mistaken , Jack also mentioned divergence in a past as an confirmation tool. If you have a time,try to test entries only after Stoch. and price divergence. It will cut down number of trades, but it should improve profitability.
    Thanks,
    Walter
     
    #516     Mar 31, 2003
  7. dawg

    dawg

    wow. great job, gives you a lot to think about. thanks for all your work.

    dawg
     
    #517     Mar 31, 2003
  8. ges

    ges

    Thanks very much for the testing.

    I'm not sure what you mean by the above...entry price set to midpoint of entry bar. Don't you have to wait until the signal bar has closed to know if you've got a valid entry? How can you anticipate this in real time? It would seem to me you'd have to enter on the open of the bar following the signal bar (crossing of fast line over 75 or 80). Am I missing something?

    ges
     
    #518     Mar 31, 2003
  9. dawg

    dawg

    what does the software do with the following:

    during the bar the fast stoch goes above 80, but closes below 80....would it exit you at the close or once in the trade your 2pt stop kicks in? thanks in advance.

    dawg
     
    #519     Mar 31, 2003
  10. vorzo

    vorzo

    ges,

    As Jack mentioned several times, the entry should be as soon as the fast line crosses into signal area (be it 75 or 80). Yes you get more false signals (fast line crosses during the bar and snaps out of the area before close of the bar) but that's where sequences, volume, etc. come into play.

    vorzo
     
    #520     Mar 31, 2003
Thread Status:
Not open for further replies.