A typical case of selective reading and reverse psychology… 😉

I mean, there are a lot of questions about the system and I had hoped to get an answer to that… 😉

Sorry MaoRai54, I thought that comment came from davidelaferia, is not about your system, was about the “Linear regression universel strategy” system.

@ JS you’re right, sorry. the TF ‘ve tested better is 15 M not 5M

The “linear regression universal system” which I started yesterday in demo (time frame 1 day) is still running and has made two transactions…

Probably something else going on yesterday, that’s why it rained error messages…

JS wrote:

I started yesterday in demo (time frame 1 day) is still running

On the strength of above, I just started my M5 version of David TS. First full 5 min bar it got Rejected for not enough historical data to calculate at least one indicator etc Seems Rejections are linked to TF’s at < 1 day fo some weird reason? Anybody got any ideas (Roberto?) … here’s the code (optimised for 10K bars)  that gets Rejected at first full 5 min bar,

//-------------------------------------------------------------------------
// Main code : David LinReg DJI M5 10K v1gh
//-------------------------------------------------------------------------
//https://www.prorealcode.com/topic/linear-regression-universal-strategy-how-to-improve-it/#post-199478//Linear Regression Universal Strategy
//DJI
// timeframe : M5
// Spread: 5
// created and coded by davidelaferla
//————————————————————————-
 
defparam cumulateorders=false
DEFPARAM PRELOADBARS = 10000
 
//***********************************************************************************************************
ONCE A15 = 25.0
ONCE A16 = 10.0
ONCE A17 = 0.8
N = 1
//------------------ VARIABILI DEL SISTEMA---------------------------------------
//CNHJPY Values:      -------------------------------------------- Ottimization info
Volatilityperiod=A15 //15// different fixed value for each currency pair: range=3-20, with step=1
Regression1period=A16 //18// Linear Regression Imput Signal:              range=6-100 with step=1
Regression2period=A17 //1.3//Linear Regression Output Signal:             range=1-2, with step=0.01
//***********************************************************************************************
//------------------ INDICATOR ---------------------------------------
 
hv=HistoricVolatility[VolatilityPeriod](close)
hv2=HistoricVolatility[VolatilityPeriod](typicalprice)
st=linearregression[ROUND(Regression1Period*hv)](close)
lr=linearregression[ROUND(Regression1Period*Regression2Period*hv2)](typicalprice)
 
// Condizioni per uscire da posizioni short e entrare su posizioni long
IF st<lr THEN
BUY n contract AT lr stop
ENDIF
// Condizioni per uscire da posizioni long e entrare su posizioni short
IF st>lr THEN
SELLshort n contract At st stop
ENDIF

Hi GraHal, hi all, maybe worth trying these 2 modifications together for lines 29-30 of your #199668? (a) change the period to max(1, ROUND(Regression1Period*Regression2Period*hv2)) in order to avoid a zero period in case of low hv or hv2, and still keep the “round” spirit (read: not use ceil when not necessary above 1) (b) add an “if” statement on barindex to make sure hv and hv2 actually exist when calculting these periods for lr and st: if barindex>=VolatilityPeriod then… All together:

if barindex>=VolatilityPeriod then
 lr=linearregression[max(1,ROUND(Regression1Period*Regression2Period*hv2))](typicalprice)
 st=linearregression[max(1,ROUND(Regression1Period*hv))](close)
endif

If it doesn’t work I can try to have another look, but it’s all I can think of for now…

ok, as it solved display issues with returned lr and st in probuilder, I was hoping to avoid the pain of adding another system in proorder… but I tried and got it rejected too. With it working in probuilder and not proorder it made me look at orders code lines, two thoughts:  (1) I never tried sending pending order « from the wrong side », and position of price wasn’t checked to decide if pending order at stop or at limit is more appropriate, so I modified code as follows in case wrong side pending order was messing things up : IF st<lr THEN if close<lr then BUY n contract AT lr stop elsif close>lr then BUY n contract AT lr limit ENDIF endif // IF st>lr THEN if close>st then SELLshort n contract At st stop elsif close<st then SELLshort n contract At st limit endif ENDIF and (2) st and lr are highly likely to be within minimum stop distance and without stop readjustment box checked system is likely to get stopped fast, so I tried again checking the box… But even with those further 2 changes, got it rejected again… sorry…