Btw the paul’s last version works very good on nikkei spread 5.
But huge drawdown.
Optimized performance with fairly conservative MM added … dream on!
PaulParticipant
Master
@nonetheless nice interesting find! I see is long & short on one bar. Is that a problem for daily bars?
maybe split the strategy long & short
and i.e. if in longposition exit also when short criteria is met.
@nonetheless your backtest is without tick by tick?
your backtest is without tick by tick?
Damn, I knew there’d be something wrong with it! Tick by tick is supposedly available from Aug 2010 but the most it will give me is from 2016.
Still semi-ok, but not nearly as good (MM is off) … so frustrating!
Bet you wish your Trades were Hot like Mine (from the song! 🙂 ) … see attached 🙂
All in the last 3 days!
Wow, that is hot! Gotta reload that one. Well done for persevering!
PaulParticipant
Master
don’t seem much interest in this, but in demo it’s awesome!
Grahal you’re still working on this?
Grahal you’re still working on this?
Not working on it, but I’ve had 2 versions on Forward Test since 25 Feb … attached are the results.
Lot size is 1 x $2 Contract, so halve the profit for near-direct comparison with 1 x £1 Contract SB.
Still good though even through all this turmoil in the markets?
I did a bit of work on the 1m version but couldn’t find much that improved it. The win% can be higher just by raising the stop loss, but then the draw down is so much worse. Seems to work best with lots of small losses. These are the only changes i made (position size is 1 x $2)
ctime=hour>=1 and hour<=20
if ctime then
buy positionsize CONTRACT at renkoMax + boxSize stop
sellshort positionsize CONTRACT at renkoMin - boxSize stop
endif
// Stops and targets
SET STOP %LOSS .2
SET TARGET %PROFIT 1.5
I did notice that with the 15m (longer back test) the best box size for the DOW was def 100; presently on the 1m TF it’s 110 but that may need to change if looking further back.
presently on the 1m TF it’s 110 but that may need to change if looking further back.
A good candidate for JuanJ Heuristics and then box size can change ‘on the fly’ … going forward?
PaulParticipant
Master
@grahal Stop orders like in this code are perhaps not reliable for entry, because it’s too close to the entry-criteria I think. Maybe change it to market orders. But indeed Heuristics would be a good candidate!
PaulParticipant
Master
A small stoploss would be better, no weekend or no overnight.
Here’s a pic from results so far. But I have to say, backtest-demo-live all difference. I come back to this later.
(left pic backtest, right pic demo)