Hi @betson, my backtest (version 4.7) shows short positions on may 13,14 and two on 15th. You cannot see these ?
@betson, it only moves in the direction of the primary trend, ie if the 2hr chart is rising it can only go long and only goes short if the 2hr is falling. There’s been a pullback in the last couple of days but technically we’re still in an uptrend.
sorry , my mistake ,i can see those trades, my apologies
@nonetheless, thanks for the reply
regards
betson
Hi @nonetheless,
I have been working to enhance the strategy by limiting the loss. I have added a lot of filtering using BB%. Attached.
For each filter the parameter undergone WFA, but seems all 5 iterations having quite consistent value as per set in the .itf
I also did the VRT and attach result. Left is v4.7 result, right is the modified one (v4.7.1a)
I have added a lot of filtering using BB%. Attached.
Btw, I was utilizing the opening bar for one of the filters thus time function was used, time zone is UTC+8.
// --------- US DAY LIGHT SAVINGS MONTHS ---------------- //
mar = month = 3 // MONTH START
nov = month = 11 // MONTH END
IF (month > 3 AND month < 11) OR (mar AND day>14) OR (mar AND day-dayofweek>7) OR (nov AND day<=dayofweek AND day<7) THEN
USDLS=010000
ELSE
USDLS=0
ENDIF
once shortStep = 0
once longStep = 0
once openStrongLong = 0
once openStrongShort = 0
if not onmarket or (time <= 223000 - USDLS and time >= 050000 - USDLS) then
shortStep = 0
longStep = 0
openStrongLong = 0
openStrongShort = 0
endif
//detect strong direction for market open
once rangeOK = 45
once tradeMin = 500
IF (time >= 223500 - USDLS) AND (time <= 223500 + tradeMin - USDLS) THEN
openStrongLong = close > open AND close - open > rangeOK
openStrongShort = close < open AND open - close > rangeOK
ENDIF
In case of USA time zone, just need to remove the DLS (day light saving) and adjust the time something like this (not tested)
// --------- US DAY LIGHT SAVINGS MONTHS ---------------- //
//mar = month = 3 // MONTH START
//nov = month = 11 // MONTH END
//IF (month > 3 AND month < 11) OR (mar AND day>14) OR (mar AND day-dayofweek>7) OR (nov AND day<=dayofweek AND day<7) THEN
//USDLS=010000
//ELSE
//USDLS=0
//ENDIF
once shortStep = 0
once longStep = 0
once openStrongLong = 0
once openStrongShort = 0
if not onmarket or (time <= 093000 or time >= 160000) then
shortStep = 0
longStep = 0
openStrongLong = 0
openStrongShort = 0
endif
//detect strong direction for market open
once rangeOK = 45
once tradeMin = 500
IF (time >= 093500) AND (time <= 093500 + tradeMin) THEN
openStrongLong = close > open AND close - open > rangeOK
openStrongShort = close < open AND open - close > rangeOK
ENDIF
Haven’t check in case of Paris or UK time zone…need to keep the DLS code and adjust the time accordingly I think
Hi @DowJones, thanks so much for your work on this.
In this section
//detect strong direction for market open
once rangeOK = 45
once tradeMin = 500
IF (time >= 093500) AND (time <= 093500 + tradeMin) THEN
openStrongLong = close > open AND close - open > rangeOK
openStrongShort = close < open AND open - close > rangeOK
ENDIF
I would suggest adding another candle to the opening time, ie 094000 NY time, just to give it a bit longer to establish the direction.
Hi @nonetheless, thanks for your suggestion. I have a question for your strategy. There are many variables and in different time frame, how do you perform optimization?
I like to see how to apply this to another market, but it is hard to work with many variables without a good strategy to start. I’m thinking the following, do you have suggestion?
1. Disable trailing stop, SL and TP
2. Remove some of the indicators, keep just 2 condition pairs (e.g. c1, c3 for long, c2, c4 for short), optimize them until obtain a good range, and perform WFA
3. Keep the first 2 condition pairs parameter fixed, add 2 other condition pairs, and redo optimizing
4. Add back SL, TP and trailing stop and optimize them
5. Have all parameters (except SL, TP and trailing) open for optimization but just 1 step above and 1 step below, perform WFA
6. Perform VST
Here is the strategy optimised for SP100.
Haven’t run any robustness test on it. Feel free to run one and let me know if it is any reliable.
@Francesco
Sorry with my non-premium account thats the best I can do. Can you try it on a larger data , if you can? May be its a better test to see if the optimisation actually works.. 🙂
Please try to avoid double posts.
With 100k you can go back to jan 2019; anyway both 100k and 200k are a complete disaster.
I’ve been trading the Dow version on my live account but I’ve been surprised it hasn’t taken any shorts during this recent dip. The 2 hour is trending down, maybe the dip has been too sharp? Anybody had any short trades in the last day or so?
hello no trip home either
@Jabezz I have not detected any trades this week, there was a closing early monday. Latest short trades was in mid may
I’m impressed of this algo, even though I don’t understand the code fully yet. Especially to consider the market even after entry is from my small experience unique. The rented algos I’m running are blindfolded after entry. Thank You for sharing so we newbies are able to learn and test before wo contribute. Enjoy the upcoming weekend!
@ Dow Jones, this post #123625 gives some explanation of the overall structure and logic to the strategy. On that basis, my approach to optimisation is to treat each of the 3 trends separately; first the 2hr primary trend, then the 15m/30m secondary trend, then the 5m minor trend. SL, TP and trail come last. I expect that your approach could work perfectly well, but you might end up with a completely different algo. Try it and let us know.
