stoptrailing mechanics question in MACD example - Quantstrat

2 messages · Joshua Ulrich, Derek Wong

Fri, Mar 11, 2016 7:26 PM #

On Wed, Feb 17, 2016 at 2:22 AM, Derek Wong <treydog999 at gmail.com> wrote:

Hello All,

I am using Quantstrat 0.9.1709 with Microsoft R Open 3.2.3 on Rstudio.

I am having a problem understanding the trailing methodology of the
stoptrailing order type in Quantstrat. It seems to me that the
trailing stop is not replaced as frequently as it should. Only for
larger price changes (threshold value?) it moves but for smaller
increments it remains the same. Is this the intended result?

I am using the MACD example and commenting out the original exit rule
and un-commenting the trailing exit.

<snip>
#alternatives for risk stops:
# simple stoplimit order, with threshold multiplier
#add.rule(strat.st,name='ruleSignal', arguments =
list(sigcol="signal.gt.zero",sigval=TRUE, orderqty='all',
ordertype='stoplimit', orderside='long', threshold=-.05,tmult=TRUE,
orderset='exit2'),type='chain', parent='enter',
label='risk',storefun=FALSE)
# alternately, use a trailing order, also with a threshold multiplier
add.rule(strat.st,name='ruleSignal',
         arguments = list(sigcol="signal.gt.zero",
                          sigval=TRUE,
                          orderqty='all',
                          ordertype='stoptrailing',
                          orderside='long',
                          threshold=-1,
                          tmult=FALSE,
                          orderset='exit2'),
         type='chain',
         parent='enter',
         label='trailingexit')

# # exit
# add.rule(strat.st,name='ruleSignal',
#          arguments = list(sigcol="signal.lt.zero",
#                           sigval=TRUE,
#                           orderqty='all',
#                           ordertype='market',
#                           orderside='long',
#                           threshold=NULL,
#                           orderset='exit2'),
#          type='exit',
#          label='exit'
# )

<snip>


I am looking at the orderbook and comparing it to the portfolio and
price of AAPL. I can see that the trailing stop moves but not on every
new high created during the trade. The first trade is an example.

<output>

AAPL["2007-03-15/2007-03-26"]

           AAPL.Open AAPL.High AAPL.Low AAPL.Close AAPL.Volume AAPL.Adjusted
2007-03-15     89.96     90.36    89.31      89.57   139874700      11.84997
2007-03-16     89.54     89.99    89.32      89.59   142926000      11.85262
2007-03-19     90.24     91.55    89.59      91.13   178240300      12.05636
2007-03-20     91.35     91.84    91.06      91.48   122229100      12.10266
2007-03-21     91.99     94.00    91.65      93.87   171724000      12.41886
2007-03-22     93.73     94.36    93.00      93.96   140373100      12.43076
2007-03-23     93.35     94.07    93.30      93.52   112721000      12.37255
2007-03-26     93.99     95.90    93.30      95.85   216246800      12.68081

obook$macd$AAPL["2007-03-15/2007-03-26"]

           Order.Qty Order.Price Order.Type     Order.Side
Order.Threshold Order.Status Order.StatusTime      Prefer
2007-03-15 "100"     "89.570001" "market"       "long"     NA
    "closed"     "2007-03-16 00:00:00" ""
2007-03-16 "all"     "88.590001" "stoptrailing" "long"     "-1"
    "replaced"   "2007-03-19 00:00:00" ""
2007-03-19 "all"     "90.549999" "stoptrailing" "long"     "-1"
    "replaced"   "2007-03-21 00:00:00" ""
2007-03-21 "all"     "93.000002" "stoptrailing" "long"     "-1"
    "replaced"   "2007-03-26 00:00:00" ""
2007-03-26 "all"     "94.899999" "stoptrailing" "long"     "-1"
    "closed"     "2007-03-28 00:00:00" ""
           Order.Set Txn.Fees Rule           Time.In.Force
2007-03-15 NA        "0"      "enter"        ""
2007-03-16 "exit2"   "0"      "trailingexit" ""
2007-03-19 "exit2"   "0"      "trailingexit" ""
2007-03-21 "exit2"   "0"      "trailingexit" ""
2007-03-26 "exit2"   "0"      "trailingexit" ""

<output>

You can see that for the date 2007-03-15 to 2007-03-21 we would have
what I would believe to be the expected behavior. However at
2007-03-22 we have a new high of 94.36 so i thought the new
Order.Price would be at 93.36 given a threshold =-1 but the order is
not adjusted. The order is however adjusted to the correct value on
2007-03-26 after a new AAPL.High at 95.90 and Order Price is 94.899999

I am very curious if the threshold works not only as the distance
between price but also the threshold of a new high price change in
order for the new Order.Price to be changed. I was expecting every new
AAPL.high to generate a new Order.Price for a stoptrailing order type.

If someone could explain this to me or tell me the expected behavior
and mechanics of the trailing methodology I would very much appreciate
it.

For OHLC data, the sell stoplimit order price will be based on the
high price, but the order evaluation is based on the close price.
That means the stoplimit will only be moved if the close is >= order
price + threshold.  And when it is moved, the new order price will be
high - threshold.

For BBO data, a sell stoplimit order price *and* evaluation would be
based on the offer price.

I don't recall at the moment why they're different.  Perhaps someone
else will remember.

_______________________________________________
R-SIG-Finance at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions should go.

Joshua Ulrich  |  about.me/joshuaulrich
FOSS Trading  |  www.fosstrading.com
R/Finance 2016 | www.rinfinance.com

Derek Wong

Fri, Mar 11, 2016 9:44 PM #

Hi Joshua,

Thank you for that reply. I will take a look at that. But what you
explained seems to be the conditions I found by digging into the order
book and order update logic. The confusing part for that was using the
OHLC data and mixing the close price as a "filter" rather than just
highest high lowest low. Also for the fills as well, as a close below
the stop limit order is not really realistic. As you can be filled by
the low of the day and still have the close above (high of the day and
close below for a short), in reality you would have obtained a fill in
the market. This is assuming that the stoptrailing order is in the
market as it executes at the limit price. But in the back test you
would still be maintaining a position as the order would not be
filled. This leads to some disparity in the results as only winners
would not be filled and continue higher but any losers would be cut
off properly. When in reality both would have been closed. I believe
this backtesting behavior biases the returns upwards when using the
stoptrailing order.

Thanks again,

Derek

On Sat, Mar 12, 2016 at 11:26 AM, Joshua Ulrich <josh.m.ulrich at gmail.com> wrote:

On Wed, Feb 17, 2016 at 2:22 AM, Derek Wong <treydog999 at gmail.com> wrote:

Hello All,

I am using Quantstrat 0.9.1709 with Microsoft R Open 3.2.3 on Rstudio.

I am having a problem understanding the trailing methodology of the
stoptrailing order type in Quantstrat. It seems to me that the
trailing stop is not replaced as frequently as it should. Only for
larger price changes (threshold value?) it moves but for smaller
increments it remains the same. Is this the intended result?

I am using the MACD example and commenting out the original exit rule
and un-commenting the trailing exit.

<snip>
#alternatives for risk stops:
# simple stoplimit order, with threshold multiplier
#add.rule(strat.st,name='ruleSignal', arguments =
list(sigcol="signal.gt.zero",sigval=TRUE, orderqty='all',
ordertype='stoplimit', orderside='long', threshold=-.05,tmult=TRUE,
orderset='exit2'),type='chain', parent='enter',
label='risk',storefun=FALSE)
# alternately, use a trailing order, also with a threshold multiplier
add.rule(strat.st,name='ruleSignal',
         arguments = list(sigcol="signal.gt.zero",
                          sigval=TRUE,
                          orderqty='all',
                          ordertype='stoptrailing',
                          orderside='long',
                          threshold=-1,
                          tmult=FALSE,
                          orderset='exit2'),
         type='chain',
         parent='enter',
         label='trailingexit')

# # exit
# add.rule(strat.st,name='ruleSignal',
#          arguments = list(sigcol="signal.lt.zero",
#                           sigval=TRUE,
#                           orderqty='all',
#                           ordertype='market',
#                           orderside='long',
#                           threshold=NULL,
#                           orderset='exit2'),
#          type='exit',
#          label='exit'
# )

<snip>


I am looking at the orderbook and comparing it to the portfolio and
price of AAPL. I can see that the trailing stop moves but not on every
new high created during the trade. The first trade is an example.

<output>

AAPL["2007-03-15/2007-03-26"]

           AAPL.Open AAPL.High AAPL.Low AAPL.Close AAPL.Volume AAPL.Adjusted
2007-03-15     89.96     90.36    89.31      89.57   139874700      11.84997
2007-03-16     89.54     89.99    89.32      89.59   142926000      11.85262
2007-03-19     90.24     91.55    89.59      91.13   178240300      12.05636
2007-03-20     91.35     91.84    91.06      91.48   122229100      12.10266
2007-03-21     91.99     94.00    91.65      93.87   171724000      12.41886
2007-03-22     93.73     94.36    93.00      93.96   140373100      12.43076
2007-03-23     93.35     94.07    93.30      93.52   112721000      12.37255
2007-03-26     93.99     95.90    93.30      95.85   216246800      12.68081

obook$macd$AAPL["2007-03-15/2007-03-26"]

           Order.Qty Order.Price Order.Type     Order.Side
Order.Threshold Order.Status Order.StatusTime      Prefer
2007-03-15 "100"     "89.570001" "market"       "long"     NA
    "closed"     "2007-03-16 00:00:00" ""
2007-03-16 "all"     "88.590001" "stoptrailing" "long"     "-1"
    "replaced"   "2007-03-19 00:00:00" ""
2007-03-19 "all"     "90.549999" "stoptrailing" "long"     "-1"
    "replaced"   "2007-03-21 00:00:00" ""
2007-03-21 "all"     "93.000002" "stoptrailing" "long"     "-1"
    "replaced"   "2007-03-26 00:00:00" ""
2007-03-26 "all"     "94.899999" "stoptrailing" "long"     "-1"
    "closed"     "2007-03-28 00:00:00" ""
           Order.Set Txn.Fees Rule           Time.In.Force
2007-03-15 NA        "0"      "enter"        ""
2007-03-16 "exit2"   "0"      "trailingexit" ""
2007-03-19 "exit2"   "0"      "trailingexit" ""
2007-03-21 "exit2"   "0"      "trailingexit" ""
2007-03-26 "exit2"   "0"      "trailingexit" ""

<output>

You can see that for the date 2007-03-15 to 2007-03-21 we would have
what I would believe to be the expected behavior. However at
2007-03-22 we have a new high of 94.36 so i thought the new
Order.Price would be at 93.36 given a threshold =-1 but the order is
not adjusted. The order is however adjusted to the correct value on
2007-03-26 after a new AAPL.High at 95.90 and Order Price is 94.899999

I am very curious if the threshold works not only as the distance
between price but also the threshold of a new high price change in
order for the new Order.Price to be changed. I was expecting every new
AAPL.high to generate a new Order.Price for a stoptrailing order type.

If someone could explain this to me or tell me the expected behavior
and mechanics of the trailing methodology I would very much appreciate
it.

For OHLC data, the sell stoplimit order price will be based on the
high price, but the order evaluation is based on the close price.
That means the stoplimit will only be moved if the close is >= order
price + threshold.  And when it is moved, the new order price will be
high - threshold.

For BBO data, a sell stoplimit order price *and* evaluation would be
based on the offer price.

I don't recall at the moment why they're different.  Perhaps someone
else will remember.

Thank you

Derek

_______________________________________________
R-SIG-Finance at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions should go.